LECTURE NOTES FOR 18.

155, FALL 2004

RICHARD B. MELROSE

Contents
Introduction 1
1. Continuous functions 2
2. Measures and σ­algebras 10
3. Measureability of functions 16
4. Integration 19
5. Hilbert space 30
6. Test functions 34
7. Tempered distributions 42
8. Convolution and density 47
9. Fourier inversion 58
10. Sobolev embedding 63
11. Differential operators. 67
12. Cone support and wavefront set 83
13. Homogeneous distributions 96
14. Wave equation 97
15. Operators and kernels 98
16. Spectral theorem 99
17. Problems 103
18. Solutions to (some of) the problems 130
References 136

Introduction
These notes are for the course the graduate analysis course (18.155)
at MIT in Fall 2004. They are based on earlier notes for similar courses
in 1997, 2001 and 2002. In giving the lectures I may cut some corners!
I wish to particularly thank Austin Frakt for many comments on,
and corrections to, an earlier version of these notes. Others who made
helpful comments or noted errors include Philip Dorrell, ....
1
2 RICHARD B. MELROSE

1. Continuous functions
A the beginning I want to remind you of things I think you already
know and then go on to show the direction the course will be taking.
Let me first try to set the context.
One basic notion I assume you are reasonably familiar with is that
of a metric space ([5] p.9). This consists of a set, X, and a distance
function
d : X × X = X 2 −→ [0, ∞) ,
satisfying the following three axioms:
i) d(x, y) = 0 ⇔ x = y, (and d(x, y) ≥ 0)
(1.1) ii) d(x, y) = d(y, x) ∀ x, y ∈ X
iii) d(x, y) ≤ d(x, z) + d(z, y) ∀ x, y, z ∈ X.
The basic theory of metric spaces deals with properties of subsets
(open, closed, compact, connected), sequences (convergent, Cauchy)
and maps (continuous) and the relationship between these notions.
Let me just remind you of one such result.
Proposition 1.1. A map f : X → Y between metric spaces is con­
tinuous if and only if one of the three following equivalent conditions
holds
(1) f −1 (O) ⊂ X is open ∀ O ⊂ Y open.
(2) f −1 (C) ⊂ X is closed ∀ C ⊂ Y closed.
(3) limn→∞ f (xn ) = f (x) in Y if xn → x in X.
The basic example of a metric space is Euclidean space. Real n­
dimensional Euclidean space, Rn , is the set of ordered n­tuples of real
numbers
x = (x1 , . . . , xn ) ∈ Rn , xj ∈ R , j = 1, . . . , n .
It is also the basic example of a vector (or linear) space with the oper­
ations
x + y = (x1 + y1 , x2 + y2 , . . . , xn + yn )
cx = (cx1 , . . . , cxn ) .
The metric is usually taken to be given by the Euclidean metric
n
�
2 2 1/2
|x| = (x1 + · · · + xn ) = ( x2j )1/2 ,
j=1

in the sense that

d(x, y) = |x − y | .
 LECTURE NOTES FOR 18.155, FALL 2004 3

Let us abstract this immediately to the notion of a normed vector

space, or normed space. This is a vector space V (over R or C) equipped
with a norm, which is to say a function
� � : V −→ [0, ∞)
satisfying
i) �v� = 0 ⇐⇒ v = 0,
(1.2) ii) �cv� = |c| �v� ∀ c ∈ K,
iii) �v + w� ≤ �v� + �w�.
This means that (V, d), d(v, w) = �v − w� is a vector space; I am also
using K to denote either R or C as is appropriate.
The case of a finite dimensional normed space is not very interesting
because, apart from the dimension, they are all “the same”. We shall
say (in general) that two norms � • �1 and � • �2 on V are equivalent
of there exists C > 0 such that
1
�v�1 ≤ �v�2 ≤ C�v�1 ∀ v ∈ V .
C
Proposition 1.2. Any two norms on a finite dimensional vector space
are equivalent.
So, we are mainly interested in the infinite dimensional case. I will
start the course, in a slightly unorthodox manner, by concentrating on
one such normed space (really one class). Let X be a metric space.
The case of a continuous function, f : X → R (or C) is a special case
of Proposition 1.1 above. We then define
C(X) = {f : X → R, f bounded and continuous} .
In fact the same notation is generally used for the space of complex­
valued functions. If we want to distinguish between these two possi­
bilities we can use the more pedantic notation C(X; R) and C(X; C).
Now, the ‘obvious’ norm on this linear space is the supremum (or ‘uni­
form’) norm
�f �∞ = sup |f (x)| .
x∈X

Here X is an arbitrary metric space. For the moment X is sup­

posed to be a “physical” space, something like Rn . Corresponding to
the finite­dimensionality of Rn we often assume (or demand) that X
is locally compact. This just means that every point has a compact
neighborhood, i.e., is in the interior of a compact set. Whether locally
4 RICHARD B. MELROSE

compact or not we can consider

� �
(1.3) C0 (X) = f ∈ C(X); ∀ � > 0 ∃ K � Xs.t. sup |f (x)| ≤ � .
x∈K
/

Here the notation K � X means ‘K is a compact subset of X’.

If V is a normed linear space we are particularly interested in the
continuous linear functionals on V . Here ‘functional’ just means func­
tion but V is allowed to be ‘large’ (not like Rn ) so ‘functional’ is used
for historical reasons.
Proposition 1.3. The following are equivalent conditions on a linear
functional u : V −→ R on a normed space V .
(1) u is continuous.
(2) u is continuous at 0.
(3) {u(f ) ∈ R ; f ∈ V , �f � ≤ 1} is bounded.
(4) ∃ C s.t. |u(f )| ≤ C�f � ∀ f ∈ V .
Proof. (1) =⇒ (2) by definition. Then (2) implies that u−1 (−1, 1) is
a neighborhood of 0 ∈ V , so for some � > 0, u({f ∈ V ; �f � < �}) ⊂
(−1, 1). By linearity of u, u({f ∈ V ; �f � < 1}) ⊂ (− 1� , 1� ) is bounded,
so (2) =⇒ (3). Then (3) implies that
|u(f )| ≤ C ∀ f ∈ V, �f � ≤ 1
for some C. Again using linearity of u, if f �= 0,
� �
f
|u(f )| ≤ �f �u ≤ C�f � ,
�f �
giving (4). Finally, assuming (4),
|u(f ) − u(g)| = |u(f − g)| ≤ C�f − g�
shows that u is continuous at any point g ∈ V . �
In view of this identification, continuous linear functionals are often
said to be bounded. One of the important ideas that we shall exploit
later is that of ‘duality’. In particular this suggests that it is a good
idea to examine the totality of bounded linear functionals on V . The
dual space is
V � = V ∗ = {u : V −→ K , linear and bounded} .
This is also a normed linear space where the linear operations are
(u + v)(f ) = u(f ) + v(f )
(1.4) ∀ f ∈ V.
(cu)(f ) = c(u(f ))
 LECTURE NOTES FOR 18.155, FALL 2004 5

The natural norm on V � is

�u� = sup |u(f )|.
�f �≤1

This is just the ‘best constant’ in the boundedness estimate,

�u� = inf {C; |u(f )| ≤ C�f � ∀ f ⊂ V } .
One of the basic questions I wish to pursue in the first part of the
course is: What is the dual of C0 (X) for a locally compact metric space
X? The answer is given by Riesz’ representation theorem, in terms of
(Borel) measures.
Let me give you a vague picture of ‘regularity of functions’ which
is what this course is about, even though I have not introduced most
of these spaces yet. Smooth functions (and small spaces) are towards
the top. Duality flips up and down and as we shall see L2 , the space
of Lebesgue square­integrable functions, is generally ‘in the middle’.
What I will discuss first is the right side of the diagramme, where we
have the space of continuous functions on Rn which vanish at infinity
and its dual space, Mfin (Rn ), the space of finite Borel measures. There
are many other spaces that you may encounter, here I only include test
functions, Schwartz functions, Sobolev spaces and their duals; k is a
general positive integer.
(1.5) S(R n )  Uw UU
_
UUUU
UUUU
UUUU
UUUU
 UUU*
n  / C0 (Rn )
H k (R _
n
) C (R
c  )
k _
ss K
ss sss
 sy ss
L2 (R b
 ) s _ KKK
KKK
KKK
 K% 
H −k (R n
) M (Rn ) o ? _ Mfin (Rn )
 i
_ iii Gg
iiiiiii
i
iiii
 tiiii
S � (Rn ).
I have set the goal of understanding the dual space Mfin (Rn ) of C0 (X),
where X is a locally compact metric space. This will force me to go
through the elements of measure theory and Lebesgue integration. It
does require a little forcing!
The basic case of interest is Rn . Then an obvious example of a con­
tinuous linear functional on C0 (Rn ) is given by Riemann integration,
6 RICHARD B. MELROSE

for instance over the unit cube [0, 1]n :

�
u(f ) = f (x) dx .
[0,1]n

In some sense we must show that all continuous linear functionals

on C0 (X) are given by integration. However, we have to interpret
integration somewhat widely since there are also evaluation functionals.
If z ∈ X consider the Dirac delta
δz (f ) = f (z) .
This is also called a point mass of z. So we need a theory of measure
and integration wide enough to include both of these cases.
One special feature of C0 (X), compared to general normed spaces, is
that there is a notion of positivity for its elements. Thus f ≥ 0 just
means f (x) ≥ 0 ∀ x ∈ X.
Lemma 1.4. Each f ∈ C0 (X) can be decomposed uniquely as the dif­
ference of its positive and negative parts
(1.6) f = f+ − f− , f± ∈ C0 (X) , f± (x) ≤ |f (x)| ∀ x ∈ X .
Proof. Simply define
�
±f (x) if ±f (x) ≥ 0
f± (x) =
0 if ±f (x) < 0
for the same sign throughout. Then (1.6) holds. Observe that f+ is
continuous at each y ∈ X since, with U an appropriate neighborhood
of y, in each case
f (y) > 0 =⇒ f (x) > 0 for x ∈ U =⇒ f+ = f in U
f (y) < 0 =⇒ f (x) < 0 for x ∈ U =⇒ f+ = 0 in U
f (y) = 0 =⇒ given � > 0 ∃ U s.t. |f (x)| < � in U
=⇒ |f+ (x)| < � in U .
Thus f− = f −f+ ∈ C0 (X), since both f+ and f− vanish at infinity. �
We can similarly split elements of the dual space into positive and
negative parts although it is a little bit more delicate. We say that
u ∈ (C0 (X))� is positive if
(1.7) u(f ) ≥ 0 ∀ 0 ≤ f ∈ C0 (X) .
For a general (real) u ∈ (C0 (X))� and for each 0 ≤ f ∈ C0 (X) set
(1.8) u+ (f ) = sup {u(g) ; g ∈ C0 (X) , 0 ≤ g(x) ≤ f (x) ∀ x ∈ X} .
 LECTURE NOTES FOR 18.155, FALL 2004 7

This is certainly finite since u(g) ≤ C�g�∞ ≤ C�f �∞ . Moreover, if

0 < c ∈ R then u+ (cf ) = cu+ (f ) by inspection. Suppose 0 ≤ fi ∈
C0 (X) for i = 1, 2. Then given � > 0 there exist gi ∈ C0 (X) with
0 ≤ gi (x) ≤ fi (x) and
u+ (fi ) ≤ u(gi ) + � .
It follows that 0 ≤ g(x) ≤ f1 (x) + f2 (x) if g = g1 + g2 so
u+ (f1 + f2 ) ≥ u(g) = u(g1 ) + u(g2 ) ≥ u+ (f1 ) + u+ (f2 ) − 2� .
Thus
u+ (f1 + f2 ) ≥ u+ (f1 ) + u+ (f2 ).
Conversely, if 0 ≤ g(x) ≤ f1 (x) + f2 (x) set g1 (x) = min(g, f1 ) ∈
C0 (X) and g2 = g − g1 . Then 0 ≤ gi ≤ fi and u+ (f1 ) + u+ (f2 ) ≥
u(g1 ) + u(g2 ) = u(g). Taking the supremum over g, u+ (f1 + f2 ) ≤
u+ (f1 ) + u+ (f2 ), so we find
(1.9) u+ (f1 + f2 ) = u+ (f1 ) + u+ (f2 ) .
Having shown this effective linearity on the positive functions we can
obtain a linear functional by setting
(1.10) u+ (f ) = u+ (f+ ) − u+ (f− ) ∀ f ∈ C0 (X) .
Note that (1.9) shows that u+ (f ) = u+ (f1 ) − u+ (f2 ) for any decom­
posiiton of f = f1 − f2 with fi ∈ C0 (X), both positive. [Since f1 + f− =
f2 + f+ so u+ (f1 ) + u+ (f− ) = u+ (f2 ) + u+ (f+ ).] Moreover,
|u+ (f )| ≤ max(u+ (f+ ), u(f− )) ≤ �u� �f �∞
=⇒ �u+ � ≤ �u� .
The functional
u− = u + − u
is also positive, since u+ (f ) ≥ u(f ) for all 0 ≤ f ∈ C0 (x). Thus we
have proved
Lemma 1.5. Any element u ∈ (C0 (X))� can be decomposed,
u = u+ − u−
into the difference of positive elements with
�u+ � , �u− � ≤ �u� .
The idea behind the definition of u+ is that u itself is, more or less,
“integration against a function” (even though we do not know how to
interpret this yet). In defining u+ from u we are effectively throwing
away the negative part of that ‘function.’ The next step is to show that
a positive functional corresponds to a ‘measure’ meaning a function
8 RICHARD B. MELROSE

measuring the size of sets. To define this we really want to evaluate u

on the characteristic function of a set
�
1 if x ∈ E
χE (x) =
0 if x ∈
/ E.
The problem is that χE is not continuous. Instead we use an idea
similar to (1.8).
If 0 ≤ u ∈ (C0 (X))� and U ⊂ X is open, set1
(1.11) µ(U ) = sup {u(f ) ; 0 ≤ f (x) ≤ 1, f ∈ C0 (X) , supp(f ) � U } .
Here the support of f , supp(f ), is the closure of the set of points where
� 0. Thus supp(f ) is always closed, in (1.11) we only admit f if
f (x) =
its support is a compact subset of U. The reason for this is that, only
then do we ‘really know’ that f ∈ C0 (X).
Suppose we try to measure general sets in this way. We can do this
by defining
(1.12) µ∗ (E) = inf {µ(U ) ; U ⊃ E , U open} .
Already with µ it may happen that µ(U ) = ∞, so we think of
(1.13) µ∗ : P(X) → [0, ∞]
as defined on the power set of X and taking values in the extended
positive real numbers.
Definition 1.6. A positive extended function, µ∗ , defined on the power
set of X is called an outer measure if µ∗ (∅) = 0, µ∗ (A) ≤ µ∗ (B)
whenever A ⊂ B and

�
(1.14) µ∗ ( Aj ) ≤ µ(Aj ) ∀ {Aj }∞j=1 ⊂ P(X) .
j j

Lemma 1.7. If u is a positive continuous linear functional on C0 (X)

then µ∗ , defined by (1.11), (1.12) is an outer measure.
To prove this we need to find enough continuous functions. I have
relegated the proof of the following result to Problem 2.
Lemma 1.8. Suppose Ui , i = 1, . . . , N is ,a
finite collection of open sets
in a locally compact metric space and K � N i=1 Ui is a compact subset,
then there exist continuous functions fi ∈ C(X) with 0 ≤ fi ≤ 1,
supp(fi ) � Ui and
�
(1.15) fi = 1 in a neighborhood of K .
i

1See [5] starting p.42 or [1] starting p.206.

 LECTURE NOTES FOR 18.155, FALL 2004 9

Proof of Lemma 1.7. We have  to prove (1.14). Suppose first that the
Ai are open, then so is A = i Ai . If f ∈ C(X) and supp(f ) � A then
supp(f ) is covered by a finite union of the Ai s. Applying Lemma 1.8 we
� fi ’s, all but a finite number identically zero, so supp(fi ) � Ai
can find
and i fi = � 1 in a neighborhood of supp(f ).
Since f = i fi f we conclude that
� �
u(f ) = u(fi f ) =⇒ µ∗ (A) ≤ µ∗ (Ai )
i i

since 0 ≤ fi f ≤ 1 and supp(fi f ) � Ai .

Thus (1.14) holds when the Ai are open. In the general case if
Ai ⊂ Bi with the Bi open then, from the definition,


�
µ∗ ( Ai ) ≤ µ∗ ( Bi ) ≤ µ∗ (Bi ) .
i i i

Taking the infimum over the Bi gives (1.14) in general. �

10 RICHARD B. MELROSE

2. Measures and σ­algebras

An outer measure such as µ∗ is a rather crude object since, even
if the Ai are disjoint, there is generally strict inequality in (1.14). It
turns out to be unreasonable to expect equality in (1.14), for disjoint
unions, for a function defined on all subsets of X. We therefore restrict
attention to smaller collections of subsets.
Definition 2.1. A collection of subsets M of a set X is a σ­algebra if
(1) φ, X ∈ M
(2) E ∈ M =⇒ E C =  X\E ∈ M
(3) {Ei }i=1 ⊂ M =⇒ ∞
∞
i=1 Ei ∈ M.

For a general outer measure µ∗ we define the notion of µ∗ ­measurability

of a set.
Definition 2.2. A set E ⊂ X is µ∗ ­measurable (for an outer measure
µ∗ on X) if
(2.1) µ∗ (A) = µ∗ (A ∩ E) + µ∗ (A ∩ E � ) ∀ A ⊂ X .
Proposition 2.3. The collection of µ∗ ­measurable sets for any outer
measure is a σ­algebra.
Proof. Suppose E is µ∗ ­measurable, then E C is µ∗ ­measurable by the
symmetry of (2.1).
Suppose A, E and F are any three sets. Then
A ∩ (E ∪ F ) = (A ∩ E ∩ F ) ∪ (A ∩ E ∩ F C ) ∪ (A ∩ E C ∩ F )
A ∩ (E ∪ F )C = A ∩ E C ∩ F C .
From the subadditivity of µ∗
µ∗ (A ∩ (E ∪ F )) + µ∗ (A ∩ (E ∪ F )C )
≤ µ∗ (A ∩ E ∩ F ) + µ∗ (A ∩ E ∪ F C )
+ µ∗ (A ∩ E C ∩ F ) + µ∗ (A ∩ E C ∩ F C ).
Now, if E and F are µ∗ ­measurable then applying the definition twice,
for any A,
µ∗ (A) = µ∗ (A ∩ E ∩ F ) + µ∗ (A ∩ E ∩ F C )
+ µ∗ (A ∩ E C ∩ F ) + µ∗ (A ∩ E C ∩ F C )
≥ µ∗ (A ∩ (E ∪ F )) + µ∗ (A ∩ (E ∪ F )C ) .
The reverse inequality follows from the subadditivity of µ∗ , so E ∪ F
is also µ∗ ­measurable.
 LECTURE NOTES FOR 18.155, FALL 2004 11

∞
nIf {Ei }i=1 is a sequence of disjoint µ∗ ­measurable sets, set Fn =
∞
i=1 Ei and F = i=1 Ei . Then for any A,

µ∗ (A ∩ Fn ) = µ∗ (A ∩ Fn ∩ En ) + µ∗ (A ∩ Fn ∩ EnC )
= µ∗ (A ∩ En ) + µ∗ (A ∩ Fn−1 ) .
Iterating this shows that
n
�
µ∗ (A ∩ Fn ) = µ∗ (A ∩ Ej ) .
j=1

From the µ ­measurability of Fn and the subadditivity of µ∗ ,

∗

µ∗ (A) = µ∗ (A ∩ Fn ) + µ∗ (A ∩ FnC )
�n
≥ µ∗ (A ∩ Ej ) + µ∗ (A ∩ F C ) .
j=1

Taking the limit as n → ∞ and using subadditivity,

∞
�
(2.2) µ∗ (A) ≥ µ∗ (A ∩ Ej ) + µ∗ (A ∩ F C )
j=1

≥ µ∗ (A ∩ F ) + µ∗ (A ∩ F C ) ≥ µ∗ (A)
proves that inequalities are equalities, so F is also µ∗ ­measurable.
In general, for any countable union of µ∗ ­measurable sets,
∞

∞

Aj = A
�j ,
j=1 j=1

j−1
�j−1 �C

�j = Aj \
A Ai = Aj ∩ Ai
i=1 i=1
∗
is µ ­measurable since the A
�j are disjoint. �
A measure (sometimes called a positive measure) is an extended func­
tion defined on the elements of a σ­algebra M:
µ : M → [0, ∞]
such that
(2.3) µ(∅) = 0 and
�∞ � ∞

�
µ Ai = µ(Ai )
(2.4) i=1 i=1
if {Ai }∞
i=1 � j.
⊂ M and Ai ∩ Aj = φ i =
12 RICHARD B. MELROSE

The elements of M with measure zero, i.e., E ∈ M, µ(E) = 0, are

supposed to be ‘ignorable’. The measure µ is said to be complete if
(2.5) E ⊂ X and ∃ F ∈ M , µ(F ) = 0 , E ⊂ F ⇒ E ∈ M .
See Problem 4.
The first part of the following important result due to Caratheodory
was shown above.
Theorem 2.4. If µ∗ is an outer measure on X then the collection of
µ∗ ­measurable subsets of X is a σ­algebra and µ∗ restricted to M is a
complete measure.
Proof. We have already shown that the collection of µ∗ ­measurable
subsets of X is a σ­algebra. To see the second part, observe that
taking A = F in (2.2) gives
� ∞

∗ ∗
µ (F ) = µ (Ej ) if F = Ej
j j=1

and the Ej are disjoint elements of M. This is (2.4).

Similarly if µ∗ (E) = 0 and F ⊂ E then µ∗ (F ) = 0. Thus it is enough
to show that for any subset E ⊂ X, µ∗ (E) = 0 implies E ∈ M. For
any A ⊂ X, using the fact that µ∗ (A ∩ E) = 0, and the ‘increasing’
property of µ∗
µ∗ (A) ≤ µ∗ (A ∩ E) + µ∗ (A ∩ E C )
= µ∗ (A ∩ E C ) ≤ µ∗ (A)
shows that these must always be equalities, so E ∈ M (i.e., is µ∗ ­
measurable). �

Going back to our primary concern, recall that we constructed the

outer measure µ∗ from 0 ≤ u ∈ (C0 (X))� using (1.11) and (1.12). For
the measure whose existence follows from Caratheodory’s theorem to
be much use we need
Proposition 2.5. If 0 ≤ u ∈ (C0 (X))� , for X a locally compact met­
ric space, then each open subset of X is µ∗ ­measurable for the outer
measure defined by (1.11) and (1.12) and µ in (1.11) is its measure.
Proof. Let U ⊂ X be open. We only need to prove (2.1) for all A ⊂ X
with µ∗ (A) < ∞.2

2Why?
 LECTURE NOTES FOR 18.155, FALL 2004 13

Suppose first that A ⊂ X is open and µ∗ (A) < ∞. Then A ∩ U is

open, so given � > 0 there exists f ∈ C(X) supp(f ) � A ∩ U with
0 ≤ f ≤ 1 and
µ∗ (A ∩ U ) = µ(A ∩ U ) ≤ u(f ) + � .
Now, A\ supp(f ) is also open, so we can find g ∈ C(X) , 0 ≤ g ≤
1 , supp(g) � A\ supp(f ) with
µ∗ (A\ supp(f )) = µ(A\ supp(f )) ≤ u(g) + � .
Since
A\ supp(f ) ⊃ A ∩ U C , 0 ≤ f + g ≤ 1 , supp(f + g) � A ,
µ(A) ≥ u(f + g) = u(f ) + u(g)
> µ∗ (A ∩ U ) + µ∗ (A ∩ U C ) − 2�
≥ µ∗ (A) − 2�
using subadditivity of µ∗ . Letting � ↓ 0 we conclude that
µ∗ (A) ≤ µ∗ (A ∩ U ) + µ∗ (A ∩ U C ) ≤ µ∗ (A) = µ(A) .
This gives (2.1) when A is open.
In general, if E ⊂ X and µ∗ (E) < ∞ then given � > 0 there exists
A ⊂ X open with µ∗ (E) > µ∗ (A) − �. Thus,
µ∗ (E) ≥ µ∗ (A ∩ U ) + µ∗ (A ∩ U C ) − �
≥ µ∗ (E ∩ U ) + µ∗ (E ∩ U C ) − �
≥ µ∗ (E) − � .
This shows that (2.1) always holds, so U is µ∗ ­measurable if it is open.
We have already observed that µ(U ) = µ∗ (U ) if U is open. �
Thus we have shown that the σ­algebra given by Caratheodory’s
theorem contains all open sets. You showed in Problem 3 that the
intersection of any collection of σ­algebras on a given set is a σ­algebra.
Since P(X) is always a σ­algebra it follows that for any collection
E ⊂ P(X) there is always a smallest σ­algebra containing E, namely
�
ME = {M ⊃ E ; M is a σ­algebra , M ⊂ P(X)} .
The elements of the smallest σ­algebra containing the open sets are
called ‘Borel sets’. A measure defined on the σ­algebra of all Borel sets
is called a Borel measure. This we have shown:
Proposition 2.6. The measure defined by (1.11), (1.12) from 0 ≤ u ∈
(C0 (X))� by Caratheodory’s theorem is a Borel measure.
Proof. This is what Proposition 2.5 says! See how easy proofs are. �
14 RICHARD B. MELROSE

We can even continue in the same vein. A Borel measure is said to

be outer regular on E ⊂ X if
(2.6) µ(E) = inf {µ(U ) ; U ⊃ E , U open} .
Thus the measure constructed in Proposition 2.5 is outer regular on all
Borel sets! A Borel measure is inner regular on E if
(2.7) µ(E) = sup {µ(K) ; K ⊂ E , K compact} .
Here we need to know that compact sets are Borel measurable. This
is Problem 5.
Definition 2.7. A Radon measure (on a metric space) is a Borel mea­
sure which is outer regular on all Borel sets, inner regular on open sets
and finite on compact sets.
Proposition 2.8. The measure defined by (1.11), (1.12) from 0 ≤ u ∈
(C0 (X))� using Caratheodory’s theorem is a Radon measure.
Proof. Suppose K ⊂ X is compact. Let χK be the characteristic func­
tion of K , χK = 1 on K , χK = 0 on K C . Suppose f ∈ C0 (X) , supp(f ) �
X and f ≥ χK . Set
U� = {x ∈ X ; f (x) > 1 − �}
where � > 0 is small. Thus U� is open, by the continuity of f and
contains K. Moreover, we can choose g ∈ C(X) , supp(g) � U� , 0 ≤
g ≤ 1 with g = 1 near3 K. Thus, g ≤ (1 − �)−1 f and hence
µ∗ (K) ≤ u(g) = (1 − �)−1 u(f ) .
Letting � ↓ 0, and using the measurability of K,
µ(K) ≤ u(f )
⇒ µ(K) = inf {u(f ) ; f ∈ C(X) , supp(f ) � X , f ≥ χK } .
In particular this implies that µ(K) < ∞ if K � X, but is also proves
(2.7). �

Let me now review a little of what we have done. We used the

positive functional u to define an outer measure µ∗ , hence a measure
µ and then checked the properties of the latter.
This is a pretty nice scheme; getting ahead of myself a little, let me
suggest that we try it on something else.

3Meaning in a neighborhood of K.
 LECTURE NOTES FOR 18.155, FALL 2004 15

Let us say that Q ⊂ Rn is ‘rectangular’ if it is a product of finite

intervals (open, closed or half­open)
�n
(2.8) Q= (or[ai , bi ]or) ai ≤ bi
i=1
we all agree on its standard volume:
�n
(2.9) v(Q) = (bi − ai ) ∈ [0, ∞) .
i=1

Clearly if we have two such sets, Q1 ⊂ Q2 , then v(Q1 ) ≤ v(Q2 ). Let

us try to define an outer measure on subsets of Rn by
�∞ ∞
�
�

(2.10) v ∗ (A) = inf v(Qi ) ; A ⊂ Qi , Qi rectangular .

i=1 i=1

We want to show that (2.10) does define an outer measure. This is

pretty easy; certainly v(∅) = 0. Similarly if {Ai }∞
i=1 are (disjoint) sets
∞
and {Qij }i=1 is a covering
 of Ai by open rectangles then all the Qij
together cover A = i Ai and
��
v ∗ (A) ≤ v(Qij )
i j
�
⇒ v ∗ (A) ≤ v ∗ (Ai ) .
i
So we have an outer measure. We also want
Lemma 2.9. If Q is rectangular then v ∗ (Q) = v(Q).
Assuming this, the measure defined from v ∗ using Caratheodory’s
theorem is called Lebesgue measure.
Proposition 2.10. Lebesgue measure is a Borel measure.
To prove this we just need to show that (open) rectangular sets are
v ∗ ­measurable.
16 RICHARD B. MELROSE

3. Measureability of functions
Suppose that M is a σ­algebra on a set X 4 and N is a σ­algebra on
another set Y. A map f : X → Y is said to be measurable with respect
to these given σ­algebras on X and Y if
(3.1) f −1 (E) ∈ M ∀ E ∈ N .
Notice how similar this is to one of the characterizations of continuity
for maps between metric spaces in terms of open sets. Indeed this
analogy yields a useful result.
Lemma 3.1. If G ⊂ N generates N , in the sense that
�
(3.2) N = {N � ; N � ⊃ G, N � a σ­algebra}

then f : X −→ Y is measurable iff f −1 (A) ∈ M for all A ∈ G.

Proof. The main point to note here is that f −1 as a map on power
sets, is very well behaved for any map. That is if f : X → Y then
f −1 : P(Y ) → P(X) satisfies:
f −1 (E C ) = (f −1 (E))C
�∞ � ∞

−1
f Ej = f −1 (Ej )
j=1 j=1
(3.3) �∞ �
� ∞
�
−1
f Ej = f −1 (Ej )
j=1 j=1
−1 −1
f (φ) = φ , f (Y ) = X .
Putting these things together one sees that if M is any σ­algebra on
X then
f∗ (M) = E ⊂ Y ; f −1 (E) ∈ M
� �
(3.4)
is always a σ­algebra on Y.
In particular if f −1 (A) ∈ M for all A ∈ G ⊂ N then f∗ (M) is a σ­
algebra containing G, hence containing N by the generating condition.
Thus f −1 (E) ∈ M for all E ∈ N so f is measurable. �
Proposition 3.2. Any continuous map f : X → Y between metric
spaces is measurable with respect to the Borel σ­algebras on X and Y.

4ThenX, or if you want to be pedantic (X, M), is often said to be a measure

space or even a measurable space.
 LECTURE NOTES FOR 18.155, FALL 2004 17

Proof. The continuity of f shows that f −1 (E) ⊂ X is open if E ⊂ Y

is open. By definition, the open sets generate the Borel σ­algebra on
Y so the preceeding Lemma shows that f is Borel measurable i.e.,
f −1 (B(Y )) ⊂ B(X).
�
We are mainly interested in functions on X. If M is a σ­algebra
on X then f : X → R is measurable if it is measurable with respect
to the Borel σ­algebra on R and M on X. More generally, for an
extended function f : X → [−∞, ∞] we take as the ‘Borel’ σ­algebra
in [−∞, ∞] the smallest σ­algebra containing all open subsets of R and
all sets (a, ∞] and [−∞, b); in fact it is generated by the sets (a, ∞].
(See Problem 6.)
Our main task is to define the integral of a measurable function: we
start with simple functions. Observe that the characteristic function
of a set
�
1 x∈E
χE =
0 x∈ /E
is measurable if and only if E ∈ M. More generally a simple function,
N
�
(3.5) f= ai χEi , ai ∈ R
i=1

is measurable if the Ei are measurable. The presentation, (3.5), of a

simple function is not unique. We can make it so, getting the minimal
presentation, by insisting that all the ai are non­zero and
Ei = {x ∈ E ; f (x) = ai }
then f in (3.5) is measurable iff all the Ei are measurable.
The Lebesgue integral is based on approximation of functions by
simple functions, so it is important to show that this is possible.
Proposition 3.3. For any non­negative µ­measurable extended func­
tion f : X −→ [0, ∞] there is an increasing sequence fn of simple
measurable functions such that limn→∞ fn (x) = f (x) for each x ∈ X
and this limit is uniform on any measurable set on which f is finite.
Proof. Folland [1] page 45 has a nice proof. For each integer n > 0 and
0 ≤ k ≤ 22n − 1, set
En,k = {x ∈ X; 2−n k ≤ f (x) < 2−n (k + 1)},
En� = {x ∈ X; f (x) ≥ 2n }.
18 RICHARD B. MELROSE

These are measurable sets. On increasing n by one, the interval in the

definition of En,k is divided into two. It follows that the sequence of
simple functions
�
(3.6) fn = 2−n kχEk,n + 2n χEn�
k
is increasing and has limit f and that this limit is uniform on any
measurable set where f is finite. �
 LECTURE NOTES FOR 18.155, FALL 2004 19

4. Integration
The (µ)­integral of a non­negative simple function is by definition
� �
(4.1) f dµ = ai µ(Y ∩ Ei ) , Y ∈ M .
Y i

Here the convention is that if µ(Y ∩ Ei ) = ∞ but ai = 0 then ai · µ(Y ∩

Ei ) = 0. Clearly this integral takes values in [0, ∞]. More significantly,
if c ≥ 0 is a constant and f and g are two non­negative (µ­measurable)
simple functions then
� �
cf dµ = c f dµ
Y
� � Y �
(4.2) (f + g)dµ = f dµ + gdµ
Y Y Y
� �
0≤f ≤g ⇒ f dµ ≤ g dµ .
Y Y
(See [1] Proposition 2.13 on page 48.)
To see this, observe that (4.1) holds for any presentation (3.5) of f
with all ai ≥ 0. Indeed, by restriction to Ei and division by ai (which
can be assumed non­zero) it is enough to consider the special case
�
χE = bj χ Fj .
j

The Fj can always be written as the union of a finite number, N � ,

of disjoint measurable sets, Fj = ∪l∈Sj Gl where j = 1, . . . , N and
Sj ⊂ {1, . . . , N � }. Thus
� � �
bj µ(Fj ) = bj µ(Gl ) = µ(E)
j j l∈Sj
�
since {j;l∈Sj } bj = 1 for each j.
From this all the statements follow easily.
Definition 4.1. For a non­negative µ­measurable extended function
f : X −→ [0, ∞] the integral (with respect to µ) over any measurable
set E ⊂ X is
� �
(4.3) f dµ = sup{ hdµ; 0 ≤ h ≤ f, h simple and measurable.}
E E
�
By taking suprema, E f dµ has the first and last properties in (4.2).
It also has the middle property, but this is less obvious. To see this, we
shall prove the basic ‘Monotone convergence theorem’ (of Lebesgue).
Before doing so however, note what the vanishing of the integral means.
20 RICHARD B. MELROSE
�
Lemma 4.2. If f : X −→ [0, ∞] is measurable then E
f dµ = 0 for a
measurable set E if and only if
(4.4) {x ∈ E; f (x) > 0} has measure zero.
Proof. If (4.4) holds, then any positive simple function bounded above
by f must also vanish
� outside a set of measure zero, so its integral must
be zero and hence E f dµ = 0. Conversely, observe that the set in (4.4)
can be written as

En = {x ∈ E; f (x) > 1/n}.

n

Since these sets increase with n, if (4.4) does not hold then one of these
must have positive measure.
� In that case the simple function n−1 χEn
has positive integral so E f dµ > 0. �
Notice the fundamental difference in approach here between Rie­
mann and Lebesgue integrals. The Lebesgue integral, (4.3), uses ap­
proximation by functions constant on possibly quite nasty measurable
sets, not just intervals as in the Riemann lower and upper integrals.
Theorem 4.3 (Monotone Convergence). Let fn be an increasing se­
quence of non­negative measurable (extended) functions, then f (x) =
limn→∞ fn (x) is measurable and
� �
(4.5) f dµ = lim fn dµ
E n→∞ E
for any measurable set E ⊂ X.
Proof. To see that f is measurable, observe that

(4.6) f −1 (a, ∞] = fn−1 (a, ∞].

n

Since the sets (a, ∞] generate the Borel σ­algebra this shows that f is
measurable.
So we proceed to prove the main part of the proposition, which
is (4.5). Rudin has quite a nice proof of this, [5] page 21. Here I
paraphrase it. We can easily see from (4.1) that
� � �
α = sup fn dµ = lim fn dµ ≤ f dµ.
E n→∞ E E
Given a simple measurable function g with 0 ≤ g ≤ f and 0 < c < 1
consider the sets En = {x ∈ E; fn (x)
≥ cg(x)}. These are measurable
and increase with n. Moreover E = n En . It follows that
� � � �
(4.7) fn dµ ≥ fn dµ ≥ c gdµ = ai µ(En ∩ Fi )
E En En i
 LECTURE NOTES FOR 18.155, FALL 2004 21
�
in terms of the natural presentation of g = i ai χFi . Now, the fact
that the En are measurable and increase to E shows that
µ(En ∩ Fi ) → µ(E ∩ Fi )
�
as n → ∞. Thus
� the right side of (4.7) tends to c E gdµ as n → ∞.
Hence α ≥ c E gdµ for all 0 < c < 1. Taking the supremum over c and
then over all such g shows that
� � �
α = lim fn dµ ≥ sup gdµ = f dµ.
n→∞ E E E
They must therefore be equal. �
Now for instance the additivity in (4.1) for f ≥ 0 and g ≥ 0 any
measurable functions follows from Proposition 3.3. Thus if f ≥ 0 is
measurable
� and fn is an �approximating sequence as in the Proposition
then E f dµ = limn→∞ E fn dµ. So if f and g are two non­negative
measurable functions then fn (x) + gn (x) ↑ f + g(x) which shows not
only that f + g is measurable by also that
� � �
(f + g)dµ = f dµ + gdµ.
E E E
As with the definition of u+ long ago, this allows us to extend the
definition of the integral to any integrable function.
Definition 4.4. A measurable extended function f : X −→ [−∞, ∞]
is said to be integrable on E if its positive and negative parts both have
finite integrals over E, and then
� � �
f dµ = f+ dµ − f− dµ.
E E E

Notice if f is µ­integrable then so is |f |. One of the objects we wish

to study is the space of integrable functions. The fact that the integral
of |f | can vanish encourages us to look at what at first seems a much
more complicated object. Namely we consider an equivalence relation
between integrable functions
(4.8) f1 ≡ f2 ⇐⇒ µ({x ∈ X; f1 (x) �= f2 (x)}) = 0.
That is we identify two such functions if they are equal ‘off a set of
measure zero.’ Clearly if f1 ≡ f2 in this sense then
� � � �
|f1 |dµ = |f2 |dµ = 0, f1 dµ = f2 dµ.
X X X X
A necessary condition for a measurable function f ≥ 0 to be inte­
grable is
µ{x ∈ X; f (x) = ∞} = 0.
22 RICHARD B. MELROSE

Let E be the (necessarily measureable) set where f = ∞. Indeed, if

this does not have measure zero, then the sequence of simple functions
nχE ≤ f has integral tending to infinity. It follows that each equiva­
lence class under (4.8) has a representative which is an honest function,
i.e. which is finite everywhere. Namely if f is one representative then
�
f (x) x ∈/E
f � (x) =
0 x∈E
is also a representative.
We shall denote by L1 (X, µ) the space consisting of such equivalence
classes of integrable functions. This is a normed linear space as I ask
you to show in Problem 11.
The monotone convergence theorem often occurrs in the slightly dis­
guised form of Fatou’s Lemma.
Lemma 4.5 (Fatou). If fk is a sequence of non­negative integrable
functions then
� �
lim inf fn dµ ≤ lim inf fn dµ .
n→∞ n→∞

Proof. Set Fk (x) = inf n≥k fn (x). Thus Fk is an increasing sequence of

non­negative functions with limiting function lim inf n→∞ fn and Fk (x) ≤
fn (x) ∀ n ≥ k. By the monotone convergence theorem
� � �
lim inf fn dµ = lim Fk (x) dµ ≤ lim inf fn dµ.
n→∞ k→∞ n→∞

�
We further extend the integral to complex­valued functions, just say­
ing that
f :X→C
is integrable if its real and imaginary parts are both integrable. Then,
by definition, � � �
f dµ = Re f dµ + i Im f dµ
E E E
for any E ⊂ X measurable. It follows that if f is integrable then so is
|f |. Furthermore �� � �
� �
� f dµ� ≤ |f | dµ .
� �
E E
�
This is obvious if E f dµ = 0, and if not then
�
f dµ = Reiθ R > 0 , θ ⊂ [0, 2π) .
E
 LECTURE NOTES FOR 18.155, FALL 2004 23

Then
�� � �
� f dµ� = e−iθ
� �
� � f dµ
E E
�
= e−iθ f dµ
�E
= Re(e−iθ f ) dµ
�E
�Re(e−iθ f )� dµ
� �
≤
�E �
� −iθ �
≤ � e f dµ =
� |f | dµ .
E E
The other important convergence result for integrals is Lebesgue’s
Dominated convergence theorem.
Theorem 4.6. If fn is a sequence of integrable functions, fk → f a.e.5
and |fn | ≤ g for some integrable g then f is integrable and
� �
f dµ = lim fn dµ .
n→∞

Proof. First we can make the sequence fn (x) converge by changing all
the fn (x)’s to zero on a set of measure zero outside which they converge.
This does not change the conclusions. Moreover, it suffices to suppose
that the fn are real­valued. Then consider
hk = g − fk ≥ 0 .
Now, lim inf k→∞ hk = g − f by the convergence of fn ; in particular f
is integrable. By monotone convergence and Fatou’s lemma
� � �
(g − f )dµ = lim inf hk dµ ≤ lim inf (g − fk ) dµ
k→∞ k→∞
� �
= g dµ − lim sup fk dµ .
k→∞

Similarly, if Hk = g + fk then
� � � �
(g + f )dµ = lim inf Hk dµ ≤ g dµ + lim inf fk dµ.
k→∞ k→∞

It follows that
� � �
lim sup fk dµ ≤ f dµ ≤ lim inf fk dµ.
k→∞ k→∞

5Means on the complement of a set of measure zero.

24 RICHARD B. MELROSE

Thus in fact � �
fk dµ → f dµ .
�
Having proved Lebesgue’s theorem of dominated convergence, let
me use it to show something important. As before, let µ be a positive
measure on X. We have defined L1 (X, µ); let me consider the more
general space Lp (X, µ). A measurable function
f :X→C
is said to be ‘Lp ’, for 1 ≤ p < ∞, if |f |p is integrable6, i.e.,
�
|f |p dµ < ∞ .
X
As before we consider equivalence classes of such functions under the
equivalence relation
(4.9) f ∼ g ⇔ µ {x; (f − g)(x) �= 0} = 0 .
p
We denote by L (X, µ) the space of such equivalence classes. It is a
linear space and the function
�� �1/p
p
(4.10) �f �p = |f | dµ
X

is a norm (we always assume 1 ≤ p < ∞, sometimes p = 1 is excluded

but later p = ∞ is allowed). It is straightforward to check everything
except the triangle inequality. For this we start with
Lemma 4.7. If a ≥ 0, b ≥ 0 and 0 < γ < 1 then
(4.11) aγ b1−γ ≤ γa + (1 − γ)b
with equality only when a = b.
Proof. If b = 0 this is easy. So assume b > 0 and divide by b. Taking
t = a/b we must show
(4.12) tγ ≤ γt + 1 − γ , 0 ≤ t , 0 < γ < 1 .
The function f (t) = tγ − γt is differentiable for t > 0 with derivative
γtγ−1 − γ, which is positive for t < 1 and negative for t > 1. Thus
f (t) ≤ f (1) with equality only for t = 1. Since f (1) = 1 − γ, this is
(4.12), proving the lemma. �
We use this to prove Hölder’s inequality
6Check p
that |f | is automatically measurable.
 LECTURE NOTES FOR 18.155, FALL 2004 25

Lemma 4.8. If f and g are measurable then

�� �
� �
(4.13) � f gdµ� ≤ �f �p �g�q
� �
1 1
for any 1 < p < ∞, with p
+ q
= 1.
Proof. If �f �p = 0 or �g�q = 0 the result is trivial, as it is if either is
infinite. Thus consider
� f (x) �p � g(x) �q
� � � �
a = �� � ,b=� �
�f �p � � �g�q �

and apply (4.11) with γ = p1 . This gives

|f (x)g(x)| |f (x)|p |g(x)|q
≤ + .
�f �p �g�q p�f �pp q�g�qq
Integrating over X we find
�
1
|f (x)g(x)| dµ
�f �p �g�q X
1 1
≤ + = 1.
p q
�� � �
Since � X f g dµ� ≤ X |f g| dµ this implies (4.13).
�
The final inequality we need is Minkowski’s inequality.
Proposition 4.9. If 1 < p < ∞ and f, g ∈ Lp (X, µ) then
(4.14) �f + g�p ≤ �f �p + �g�p .
Proof. The case p = 1 you have already done. It is also obvious if
f + g = 0 a.e.. If not we can write
|f + g|p ≤ (|f | + |g |) |f + g |p−1
and apply Hölder’s inequality, to the right side, expanded out,
� �� �1/q
p q(p−1)
|f + g | dµ ≤ (�f �p + �g�p ) , |f + g | dµ .

1
Since q(p − 1) = p and 1 − q
= 1/p this is just (4.14). �
So, now we know that Lp (X, µ) is a normed space for 1 ≤ p < ∞. In
particular it is a metric space. One important additional property that
a metric space may have is completeness, meaning that every Cauchy
sequence is convergent.
26 RICHARD B. MELROSE

Definition 4.10. A normed space in which the underlying metric space

is complete is called a Banach space.
Theorem 4.11. For any measure space (X, M, µ) the spaces Lp (X, µ),
1 ≤ p < ∞, are Banach spaces.
Proof. We need to show that a given Cauchy sequence {fn } converges
in Lp (X, µ). It suffices to show that it has a convergent subsequence.
By the Cauchy property, for each k ∃ n = n(k) s.t.
(4.15) �fn − f� �p ≤ 2−k ∀ � ≥ n .
Consider the sequence
g1 = f1 , gk = fn(k) − fn(k−1) , k > 1 .
By (4.15), �gk �p ≤ 2−k , for k > 1, so the series k �gk �p converges,
�
say to B < ∞. Now set
�n ∞
�
hn (x) = |gk (x)| , n ≥ 1 , h(x) = gk (x).
k=1 k=1

Then by the monotone convergence theorem

� �
p
h dµ = lim |hn |p dµ ≤ B p ,
X n→∞ X
where we have also used Minkowski’s inequality. Thus h ∈ Lp (X, µ),
so the series
∞
�
f (x) = gk (x)
k=1
converges (absolutely) almost everywhere. Since
� �p
��n �
p
|f (x)| = lim � gk � ≤ hp
� �
n→∞ � �
k=1
p �
with h ∈ L (X, µ), the dominated convergence theorem applies and
shows that f ∈ Lp (X, µ). Furthermore,
�
� � �p
gk (x) = fn(�) (x) and �f (x) − fn(�) (x)� ≤ (2h(x))p
k=1

so again by the dominated convergence theorem,

�
�f (x) − fn(�) (x)�p → 0 .
� �
X
Thus the subsequence fn(�) → f in Lp (X, µ), proving its completeness.
�
 LECTURE NOTES FOR 18.155, FALL 2004 27

Next I want to return to our starting point and discuss the Riesz
representation theorem. There are two important results in measure
theory that I have not covered — I will get you to do most of them
in the problems — namely the Hahn decomposition theorem and the
Radon­Nikodym theorem. For the moment we can do without the
latter, but I will use the former.
So, consider a locally compact metric space, X. By a Borel measure
on X, or a signed Borel measure, we shall mean a function on Borel
sets
µ : B (X) → R
which is given as the difference of two finite positive Borel measures
(4.16) µ(E) = µ1 (E) − µ2 (E) .
Similarly we shall say that µ is Radon, or a signed Radon measure, if
it can be written as such a difference, with both µ1 and µ2 finite Radon
measures. See the problems below for a discussion of this point.
Let Mfin (X) denote the set of finite Radon measures on X. This is
a normed space with
(4.17) �µ�1 = inf (µ1 (X) + µ2 (X))
with the infimum over all Radon decompositions (4.16). Each signed
Radon measure defines a continuous linear functional on C0 (X):
� �
(4.18) · dµ : C0 (X) � f �−→ f · dµ .
X

Theorem 4.12 (Riesz representation.). If X is a locally compact met­

ric space then every continuous linear functional on C0 (X) is given by
a unique finite Radon measure on X through (4.18).
Thus the dual space of C0 (X) is Mfin (X) – at least this is how such
a result is usually interpreted
(4.19) (C0 (X))� = Mfin (X),
see the remarks following the proof.

Proof. We have done half of this already. Let me remind you of the
steps.
We started with u ∈ (C0 (X))� and showed that u = u+ − u− where
u± are positive continuous linear functionals; this is Lemma 1.5. Then
we showed that u ≥ 0 defines a finite positive Radon measure µ. Here
µ is defined by (1.11) on open sets and µ(E) = µ∗ (E) is given by (1.12)
28 RICHARD B. MELROSE

on general Borel sets. It is finite because

(4.20) µ(X) = sup {u(f ) ; 0 ≤ f ≤ 1 , supp f � X , f ∈ C(X)}
≤ �u� .
From Proposition 2.8 we conclude that µ is a Radon measure. Since
this argument applies to u± we get two positive finite Radon measures
µ± and hence a signed Radon measure
(4.21) µ = µ+ − µ− ∈ Mfin (X).
In the problems you are supposed to prove the Hahn decomposition
theorem, in particular in Problem 14 I ask you to show that (4.21) is
the Hahn decomposition of µ — this means that there is a Borel set
E ⊂ X such that µ− (E) = 0 , µ+ (X \ E) = 0.
What we have defined is a linear map
(4.22) (C0 (X))� → M (X), u �−→ µ .
We want to show that this is an isomorphism, i.e., it is 1 − 1 and onto.
We first show that it is 1 − 1. That is, suppose µ = 0. Given the
uniqueness of the Hahn decomposition this implies that µ+ = µ− = 0.
So we can suppose that u ≥ 0 and µ = µ+ = 0 and we have to show
that u = 0; this is obvious since
µ(X) = sup {u(f ); supp u � X, 0 ≤ f ≤ 1 f ∈ C(X)} = 0
(4.23)
⇒ u(f ) = 0 for all such f .
If 0 ≤ f ∈ C(X) and supp f � X then f � = f /�f �∞ is of this type
so u(f ) = 0 for every 0 ≤ f ∈ C(X) of compact support. From
the decomposition of continuous functions into positive and negative
parts it follows that u(f ) = 0 for every f of compact support. Now, if
f ∈ Co (X), then given n ∈ N there exists K � X such that |f | < 1/n
on X \ K. As you showed in the problems, there exists χ ∈ C(X) with
supp(χ) � X and χ = 1 on K. Thus if fn = χf then supp(fn ) � X and
�f − fn � = sup(|f − fn | < 1/n. This shows that C0 (X) is the closure
of the subspace of continuous functions of compact support so by the
assumed continuity of u, u = 0.
So it remains to show that every finite Radon measure on X arises
from (4.22). We do this by starting from µ and constructing u. Again
we use the Hahn decomposition of µ, as in (4.21)7. Thus we assume
µ ≥ 0 and construct u. It is obvious what we want, namely
�
(4.24) u(f ) = f dµ , f ∈ Cc (X) .
X
7Actually
we can just take any decomposition (4.21) into a difference of positive
Radon measures.
 LECTURE NOTES FOR 18.155, FALL 2004 29

Here we need to recall from Proposition 3.2 that continuous functions

on X, a locally compact metric space, are (Borel) measurable. Further­
more, we know that there is an increasing sequence of simple functions
with limit f , so
�� �
� �
(4.25) � f dµ� ≤ µ(X) · �f �∞ .
� �
X
This shows that u in (4.24) is continuous and that its norm �u� ≤
µ(X). In fact
(4.26) �u� = µ(X) .
Indeed, the inner regularity of µ implies that there is a compact set
K � X with µ(K) ≥ µ(X)− n1 ; then there is f ∈ Cc (X) with 0 ≤ f ≤ 1
and f = 1 on K. It follows that µ(f ) ≥ µ(K) ≥ µ(X) − n1 , for any n.
This proves (4.26).
We still have to show that if u is defined by (4.24), with µ a finite
positive Radon measure, then the measure µ̃ defined from u via (4.24)
is precisely µ itself.
This is easy provided we keep things clear. Starting from µ ≥ 0 a
finite Radon measure, define u by (4.24) and, for U ⊂ X open
�� �
(4.27) µ̃(U ) = sup f dµ, 0 ≤ f ≤ 1, f ∈ C(X), supp(f ) � U .
X
By the properties of the integral, µ̃(U ) ≤ µ(U ). Conversely if K � U
there exists an element f ∈ Cc (X), 0 ≤ f ≤ 1, f = 1 on K and
supp(f ) ⊂ U. Then we know that
�
(4.28) µ̃(U ) ≥ f dµ ≥ µ(K).
X
By the inner regularity of µ, we can choose K � U such that µ(K) ≥
µ(U ) − �, given � > 0. Thus µ̃(U ) = µ(U ).
This proves the Riesz representation theorem, modulo the decompo­
sition of the measure ­ which I will do in class if the demand is there!
In my view this is quite enough measure theory. �
Notice that we have in fact proved something stronger than the state­
ment of the theorem. Namely we have shown that under the correspon­
dence u ←→ µ,
(4.29) �u� = |µ| (X) =: �µ�1 .
Thus the map is an isometry.
30 RICHARD B. MELROSE

5. Hilbert space
We have shown that Lp (X, µ) is a Banach space – a complete normed
space. I shall next discuss the class of Hilbert spaces, a special class of
Banach spaces, of which L2 (X, µ) is a standard example, in which the
norm arises from an inner product, just as it does in Euclidean space.
An inner product on a vector space V over C (one can do the real
case too, not much changes) is a sesquilinear form
V ×V →C
written (u, v), if u, v ∈ V . The ‘sesqui­’ part is just linearity in the first
variable
(5.1) (a1 u1 + a2 u2 , v) = a1 (u1 , v) + a2 (u2 , v),
anti­linearly in the second
(5.2) (u, a1 v1 + a2 v2 ) = a1 (u, v1 ) + a2 (u, v2 )
and the conjugacy condition
(5.3) (u, v) = (v, u) .
Notice that (5.2) follows from (5.1) and (5.3). If we assume in addition
the positivity condition8
(5.4) (u, u) ≥ 0 , (u, u) = 0 ⇒ u = 0 ,
then
(5.5) �u� = (u, u)1/2
is a norm on V , as we shall see.
Suppose that u, v ∈ V have �u� = �v� = 1. Then (u, v) = eiθ |(u, v)|
for some θ ∈ R. By choice of θ, e−iθ (u, v) = |(u, v)| is real, so expanding
out using linearity for s ∈ R,
0 ≤ (e−iθ u − sv , e−iθ u − sv)
= �u�2 − 2s Re e−iθ (u, v) + s2 �v�2 = 1 − 2s|(u, v)| + s2 .
The minimum of this occurs when s = |(u, v)| and this is negative
unless |(u, v)| ≤ 1. Using linearity, and checking the trivial cases u =
or v = 0 shows that
(5.6) |(u, v)| ≤ �u� �v�, ∀ u, v ∈ V .
This is called Schwarz’9 inequality.
8Notice that (u, u) is real by (5.3).
9No ‘t’ in this Schwarz.
 LECTURE NOTES FOR 18.155, FALL 2004 31

Using Schwarz’ inequality

�u + v�2 = �u�2 + (u, v) + (v, u) + �v�2
≤ (�u� + �v�)2
=⇒ �u + v� ≤ �u� + �v� ∀ u, v ∈ V
which is the triangle inequality.
Definition 5.1. A Hilbert space is a vector space V with an inner
product satisfying (5.1) ­ (5.4) which is complete as a normed space
(i.e., is a Banach space).
Thus we have already shown L2 (X, µ) to be a Hilbert space for any
positive measure µ. The inner product is
�
(5.7) (f, g) = f g dµ ,
X
since then (5.3) gives �f �2 .
Another important identity valid in any inner product spaces is the
parallelogram law:
(5.8) �u + v�2 + �u − v�2 = 2�u�2 + 2�v�2 .
This can be used to prove the basic ‘existence theorem’ in Hilbert space
theory.
Lemma 5.2. Let C ⊂ H, in a Hilbert space, be closed and convex (i.e.,
su + (1 − s)v ∈ C if u, v ∈ C and 0 < s < 1). Then C contains a
unique element of smallest norm.
Proof. We can certainly choose a sequence un ∈ C such that
�un � → δ = inf {�v� ; v ∈ C} .
By the parallelogram law,
�un − um �2 = 2�un �2 + 2�um �2 − �un + um �2
≤ 2(�un �2 + �um �2 ) − 4δ 2
where we use the fact that (un + um )/2 ∈ C so must have norm at least
δ. Thus {un } is a Cauchy sequence, hence convergent by the assumed
completeness of H. Thus lim un = u ∈ C (since it is assumed closed)
and by the triangle inequality
|�un � − �u�| ≤ �un − u� → 0
So �u� = δ. Uniqueness of u follows again from the parallelogram law
which shows that if �u� � = δ then
�u − u� � ≤ 2δ 2 − 4�(u + u� )/2�2 ≤ 0 .
32 RICHARD B. MELROSE

�
The fundamental fact about a Hilbert space is that each element
v ∈ H defines a continuous linear functional by
H � u �−→ (u, v) ∈ C
and conversely every continuous linear functional arises this way. This
is also called the Riesz representation theorem.
Proposition 5.3. If L : H → C is a continuous linear functional on
a Hilbert space then this is a unique element v ∈ H such that
(5.9) Lu = (u, v) ∀ u ∈ H ,
Proof. Consider the linear space
M = {u ∈ H ; Lu = 0}
the null space of L, a continuous linear functional on H. By the as­
sumed continuity, M is closed. We can suppose that L is not identically
zero (since then v = 0 in (5.9)). Thus there exists w ∈
/ M . Consider
w + M = {v ∈ H ; v = w + u , u ∈ M } .
This is a closed convex subset of H. Applying Lemma 5.2 it has a
unique smallest element, v ∈ w + M . Since v minimizes the norm on
w + M,
�v + su�2 = �v�2 + 2 Re(su, v) + �s�2 �u�2
is stationary at s = 0. Thus Re(u, v) = 0 ∀ u ∈ M , and the same
argument with s replaced by is shows that (v, u) = 0 ∀ u ∈ M .
Now v ∈ w + M , so Lv = Lw = � 0. Consider the element w� =
�
w/Lw ∈ H. Since Lw = 1, for any u ∈ H
L(u − (Lu)w� ) = Lu − Lu = 0 .
It follows that u − (Lu)w� ∈ M so if w�� = w� /�w� �2
(w� , w� )
(u, w�� ) = ((Lu)w� , w�� ) = Lu = Lu .
�w� �2
The uniqueness of v follows from the positivity of the norm. �
Corollary 5.4. For any positive measure µ, any continuous linear
functional
L : L2 (X, µ) → C
is of the form �
Lf = f g dµ , g ∈ L2 (X, µ) .
X
 LECTURE NOTES FOR 18.155, FALL 2004 33

Notice the apparent power of ‘abstract reasoning’ here! Although we

seem to have constructed g out of nowhere, its existence follows from
the completeness of L2 (X, µ), but it is very convenient to express the
argument abstractly for a general Hilbert space.
34 RICHARD B. MELROSE

6. Test functions
So far we have largely been dealing with integration. One thing we
have seen is that, by considering dual spaces, we can think of functions
as functionals. Let me briefly review this idea.
Consider the unit ball in Rn ,
n
B = {x ∈ Rn ; |x| ≤ 1} .
I take the closed unit ball because I want to deal with a compact metric
space. We have dealt with several Banach spaces of functions on Bn ,
for example
� �
C(Bn ) = u : Bn → C ; u continuous
� � �
2 2
L (Bn ) = u : Bn → C; Borel measurable with |u| dx < ∞ .

Here, as always below, dx is Lebesgue measure and functions are iden­

tified if they are equal almost everywhere.
Since Bn is compact we have a natural inclusion
(6.1) C(Bn ) �→ L2 (Bn ) .
This is also a topological inclusion, i.e., is a bounded linear map, since
(6.2) �u�L2 ≤ C�u||∞
where C 2 is the volume of the unit ball.
In general if we have such a set up then
Lemma 6.1. If V �→ U is a subspace with a stronger norm,
�ϕ�U ≤ C�ϕ�V ∀ ϕ ∈ V
then restriction gives a continuous linear map
(6.3) ˜ = L|V ∈ V � , �L�
U � → V � , U � � L �−→ L ˜ V � ≤ C�L�U � .
If V is dense in U then the map (6.3) is injective.
Proof. By definition of the dual norm
�� � �
˜
�L�V � = sup �L̃(v)� ; �v�V ≤ 1 , v ∈ V
� �
�� � �
≤ sup �L̃(v)� ; �v�U ≤ C , v ∈ V
� �

≤ sup {|L(u)| ; �u�U ≤ C , u ∈ U }

= C�L�U � .
If V ⊂ U is dense then the vanishing of L : U → C on V implies its
vanishing on U .
�
 LECTURE NOTES FOR 18.155, FALL 2004 35

Going back to the particular case (6.1) we do indeed get a continuous

map between the dual spaces
L2 (Bn ) ∼
= (L2 (Bn ))� → (C(Bn ))� = M (Bn ) .
Here we use the Riesz representation theorem and duality for Hilbert
spaces. The map use here is supposed to be linear not antilinear, i.e.,
�
(6.4) L (B ) � g �−→ ·g dx ∈ (C(Bn ))� .
2 n

So the idea is to make the space of ‘test functions’ as small as reasonably

possible, while still retaining density in reasonable spaces.
Recall that a function u : Rn → C is differentiable at x ∈ Rn if there
exists a ∈ Cn such that
(6.5) |u(x) − u(x) − a · (x − x)| = o(|x − x|) .
The ‘little oh’ notation here means that given � > 0 there exists δ > 0
s.t.
|x − x| < δ ⇒ |u(x) − u(x) − a(x − x)| < � |x − x| .
The coefficients of a = (a1 , . . . , an ) are the partial derivations of u at
x,
∂u
ai = (x)
∂xj
since
u(x + tei ) − u(x)
(6.6) ai = lim ,
t→0 t
ei = (0, . . . , 1, 0, . . . , 0) being the ith basis vector. The function u is
said to be continuously differentiable on Rn if it is differentiable at each
point x ∈ Rn and each of the n partial derivatives are continuous,
∂u
(6.7) : Rn → C .
∂xj
Definition 6.2. Let C01 (Rn ) be the subspace of C0 (Rn ) = C00 (Rn ) such
∂u
that each element u ∈ C01 (Rn ) is continuously differentiable and ∂x j
∈
n
C0 (R ), j = 1, . . . , n.
Proposition 6.3. The function
n
� ∂u
�u�C 1 = �u�∞ + � �∞
i=1
∂x 1

is a norm on C01 (Rn ) with respect to which it is a Banach space.

36 RICHARD B. MELROSE

Proof. That � �C 1 is a norm follows from the properties of � �∞ . Namely

�u�C 1 = 0 certainly implies u = 0, �au�C 1 = |a| �u�C 1 and the triangle
inequality follows from the same inequality for � �∞ .
Similarly, the main part of the completeness of C01 (Rn ) follows from
the completeness of C00 (Rn ). If {un } is a Cauchy sequence in C01 (Rn )
then un and the ∂u n
∂xj
are Cauchy in C00 (Rn ). It follows that there are
limits of these sequences,
∂un
un → v , → vj ∈ C00 (Rn ) .
∂xj
However we do have to check that v is continuously differentiable and
∂v
that ∂x j
= vj .
One way to do this is to use the Fundamental Theorem of Calculus
in each variable. Thus
� t
∂un
un (x + tei ) = (x + sei ) ds + un (x) .
0 ∂xj
As n → ∞ all terms converge and so, by the continuity of the integral,
� t
u(x + tei ) = vj (x + sei ) ds + u(x) .
0
This shows that the limit in (6.6) exists, so vi (x) is the partial deriva­
tion of u with respect to xi . It remains only to show that u is indeed
differentiable at each point and I leave this to you in Problem 17.
�
So, almost by definition, we have an example of Lemma 6.1,
C01 (Rn ) �→ C00 (Rn ).
It is in fact dense but I will not bother showing this (yet). So we know
that
(C00 (Rn ))� → (C01 (Rn ))�
and we expect it to be injective. Thus there are more functionals on
C01 (Rn ) including things that are ‘more singular than measures’.
An example is related to the Dirac delta
δ(x)(u) = u(x) , u ∈ C00 (Rn ) ,
namely
∂u
C01 (Rn ) � u �−→ (x) ∈ C .
∂xj
This is clearly a continuous linear functional which it is only just to
denote ∂x∂ j δ(x).
Of course, why stop at one derivative?
 LECTURE NOTES FOR 18.155, FALL 2004 37

Definition 6.4. The space C0k (Rn ) ⊂ C01 (Rn ) k ≥ 1 is defined induc­
tively by requiring that
∂u
∈ C0k−1 (Rn ) , j = 1, . . . , n .
∂xj
The norm on C0k (Rn ) is taken to be
n
� ∂u
(6.8) �u�C k = �u�C k−1 + � � k−1 .
j=1
∂xj C

These are all Banach spaces, since if {un } is Cauchy in C0k (Rn ), it is
Cauchy and hence convergent in C0k−1 (Rn ), as is ∂un /∂xj , j = 1, . . . , n−
1. Furthermore the limits of the ∂un /∂xj are the derivatives of the limits
by Proposition 6.3.
This gives us a sequence of spaces getting ‘smoother and smoother’
C00 (Rn ) ⊃ C01 (Rn ) ⊃ · · · ⊃ C0k (Rn ) ⊃ · · · ,
with norms getting larger and larger. The duals can also be expected
to get larger and larger as k increases.
As well as looking at functions getting smoother and smoother, we
need to think about ‘infinity’, since Rn is not compact. Observe that
an element g ∈ L1 (Rn ) (with respect to Lebesgue measure by default)
defines a functional on C00 (Rn ) — and hence all the C0k (Rn )s. However a
function such as the constant function 1 is not integrable on Rn . Since
we certainly want to talk about this, and polynomials, we consider a
second condition of smallness at infinity. Let us set
(6.9) �x� = (1 + |x|2 )1/2
a function which is the size of |x| for |x| large, but has the virtue of
being smooth10
Definition 6.5. For any k, l ∈ N = {1, 2, · · · } set
�x�−l C0k (Rn ) = u ∈ C0k (Rn ) ; u = �x�−l v , v ∈ C0k (Rn ) ,
� �

with norm, �u�k,l = �v�C k , v = �x�l u.

Notice that the definition just says that u = �x�−l v, with v ∈ C0k (Rn ).
It follows immediately that �x�−l C0k (Rn ) is a Banach space with this
norm.
Definition 6.6. Schwartz’ space11 of test functions on Rn is
S(Rn ) = u : Rn → C; u ∈ �x�−l C0k (Rn ) for all k and l ∈ N .
� �

10See Problem 18.

11Laurent Schwartz – this one with a ‘t’.
38 RICHARD B. MELROSE

It is not immediately apparent that this space is non­empty (well 0

is in there but...); that
exp(− |x|2 ) ∈ S(Rn )
is Problem 19. There are lots of other functions in there as we shall
see.
Schwartz’ idea is that the dual of S(Rn ) should contain all the ‘in­
teresting’ objects, at least those of ‘polynomial growth’. The problem
is that we do not have a good norm on S(Rn ). Rather we have a lot of
them. Observe that
� �
�x�−l C0k (Rn ) ⊂ �x�−l C0k (Rn ) if l ≥ l� and k ≥ k � .
Thus we see that as a linear space
�
(6.10) S(Rn ) = �x�−k C0k (Rn ).
k

Since these spaces are getting smaller, we have a countably infinite

number of norms. For this reason S(Rn ) is called a countably normed
space.
Proposition 6.7. For u ∈ S(Rn ), set
(6.11) �u�(k) = ��x�k u�C k
and define
∞
� �u − v�(k)
(6.12) d(u, v) = 2−k ,
k=0
1 + �u − v�(k)
then d is a distance function in S(Rn ) with respect to which it is a
complete metric space.
Proof. The series in (6.12) certainly converges, since
�u − v�(k)
≤ 1.
1 + �u − v�(k)
The first two conditions on a metric are clear,
d(u, v) = 0 ⇒ �u − v�C0 = 0 ⇒ u = v,
and symmetry is immediate. The triangle inequality is perhaps more
mysterious!
Certainly it is enough to show that
�u − v�
(6.13) d˜(u, v) =
1 + �u − v�
 LECTURE NOTES FOR 18.155, FALL 2004 39

is a metric on any normed space, since then we may sum over k. Thus
we consider

�u − v� �v − w�
+
1 + �u − v� 1 + �v − w�
�u − v�(1 + �v − w�) + �v − w�(1 + �u − v�)
= .
(1 + �u − v�)(1 + �v − w�)

Comparing this to d˜(v, w) we must show that

(1 + �u − v�)(1 + �v − w�)�u − w�
≤ (�u − v�(1 + �v − w�) + �v − w�(1 + �u − v�))(1 + �u − w�).

Starting from the LHS and using the triangle inequality,

LHS ≤ �u − w� + (�u − v� + �v − w� + �u − v��v − w�)�u − w�

≤ (�u − v� + �v − w� + �u − v��v − w�)(1 + �u − w�)
≤ RHS.

Thus, d is a metric.
Suppose un is a Cauchy sequence. Thus, d(un , um ) → 0 as n, m →
∞. In particular, given

� > 0 ∃ N s.t. n, m > N implies

d(un , um ) < �2−k ∀ n, m > N.

The terms in (6.12) are all positive, so this implies

�un − um �(k)
< � ∀ n, m > N.
1 + �un − um �(k)

If � < 1/2 this in turn implies that

�un − um �(k) < 2�,

so the sequence is Cauchy in �x�−k C0k (Rn ) for each k. From the com­
pleteness of these spaces it follows that un → u in �x�−k C0k (Rn )j for
each k. Given � > 0 choose k so large that 2−k < �/2. Then ∃ N s.t.
n>N
⇒ �u − un �(j) < �/2 n > N, j ≤ k.
40 RICHARD B. MELROSE

Hence
� �u − un �(j)
d(un , u) = 2−j
j≤k
1 + �u − un �(j)

� �u − un �(j)
+ 2−j
j>k
1 + �u − un �(j)

≤ �/4 + 2−k < �.

This un → u in S(Rn ). �
As well as the Schwartz space, S(Rn ), of functions of rapid decrease
with all derivatives, there is a smaller ‘standard’ space of test functions,
namely
(6.14) Cc∞ (Rn ) = {u ∈ S(Rn ); supp(u) � Rn } ,
the space of smooth functions of compact support. Again, it is not
quite obvious that this has any non­trivial elements, but it does as
we shall see. If we fix a compact subset of Rn and look at functions
with support in that set, for instance the closed ball of radius R > 0,
then we get a closed subspace of S(Rn ), hence a complete metric space.
One ‘problem’ with Cc∞ (Rn ) is that it does not have a complete metric
topology which restricts to this topology on the subsets. Rather we
must use an inductive limit procedure to get a decent topology.
Just to show that this is not really hard, I will discuss it briefly
here, but it is not used in the sequel. In particular I will not do this
in the lectures themselves. By definition our space Cc∞ (Rn ) (denoted
traditionally as D(Rn )) is a countable union of subspaces
(6.15)

Cc∞ (Rn ) = C˙c∞ (B(n)), C˙c∞ (B(n)) = {u ∈ S(Rn ); u = 0 in |x| > n}.
n∈N

Consider
(6.16)
T = {U ⊂ Cc∞ (Rn ); U ∩ C˙c∞ (B(n)) is open in C˙∞ (B(n)) for each n}.
This is a topology on Cc∞ (Rn ) – contains the empty set and the whole
space and is closed under finite intersections and arbitrary unions –
simply because the same is true for the open sets in C˙∞ (B(n)) for each
n. This is in fact the inductive limit topology. One obvious question
is:­ what does it mean for a linear functional u : Cc∞ (Rn ) −→ C to be
continuous? This just means that u−1 (O) is open for each open set in C.
Directly from the definition this in turn means that u−1 (O)∩C˙∞ (B(n))
 LECTURE NOTES FOR 18.155, FALL 2004 41

should be open in C˙∞ (B(n)) for each n. This however just means that,
restricted to each of these subspaces u is continuous. If you now go
forwards to Lemma 7.3 you can see what this means; see Problem 74.
Of course there is a lot more to be said about these spaces; you can
find plenty of it in the references.
42 RICHARD B. MELROSE

7. Tempered distributions
A good first reference for distributions is [2], [4] gives a more exhaus­
tive treatment.
The complete metric topology on S(Rn ) is described above. Next I
want to try to convice you that elements of its dual space S � (Rn ), have
enough of the properties of functions that we can work with them as
‘generalized functions’.
First let me develop some notation. A differentiable function ϕ :
n n
R → C has partial derivatives which we have denoted ∂ϕ/∂x √ j:R →
C. For reasons that will become clear later, we put a −1 into the
definition and write
1 ∂ϕ
(7.1) Dj ϕ = .
i ∂xj
We say ϕ is once continuously differentiable if each of these Dj ϕ is
continuous. Then we defined k times continuous differentiability in­
ductively by saying that ϕ and the Dj ϕ are (k − 1)­times continuously
differentiable. For k = 2 this means that
Dj Dk ϕ are continuous for j, k = 1, · · · , n .
Now, recall that, if continuous, these second derivatives are symmetric:
(7.2) Dj Dk ϕ = Dk Dj ϕ .
This means we can use a compact notation for higher derivatives. Put
N0 = {0, 1, . . .}; we call an element α ∈ Nn0 a ‘multi­index’ and if ϕ is
at least k times continuously differentiable, we set12
1 ∂ α1 ∂ αn
(7.3) Dα ϕ = · · · ϕ whenever |α| = α1 +α2 +· · · +αn ≤ k.
i|α| ∂x1 ∂xn
Now we have defined the spaces.
C0k (Rn ) = ϕ : Rn → C ; Dα ϕ ∈ C00 (Rn ) ∀ |α| ≤ k .
� �
(7.4)
Notice the convention is that Dα ϕ is asserted to exist if it is required
to be continuous! Using �x� = (1 + |x|2 ) we defined
�x�−k C0k (Rn ) = ϕ : Rn → C ; �x�k ϕ ∈ C0k (Rn ) ,
� �
(7.5)
and then our space of test functions is
�
S(Rn ) = �x�−k C0k (Rn ) .
k

12Periodically
there is the possibility of confusion between the two meanings of
|α| but it seldom arises.
 LECTURE NOTES FOR 18.155, FALL 2004 43

Thus,
(7.6) ϕ ∈ S(Rn ) ⇔ Dα (�x�k ϕ) ∈ C00 (Rn ) ∀ |α| ≤ k and all k .
Lemma 7.1. The condition ϕ ∈ S(Rn ) can be written
�x�k Dα ϕ ∈ C00 (Rn ) ∀ |α| ≤ k , ∀ k .
Proof. We first check that
ϕ ∈ C00 (Rn ) , Dj (�x�ϕ) ∈ C00 (Rn ) , j = 1, · · · , n
⇔ ϕ ∈ C00 (Rn ) , �x�Dj ϕ ∈ C00 (Rn ) , j = 1, · · · , n .
Since
Dj �x�ϕ = �x�Dj ϕ + (Dj �x�)ϕ
and Dj �x� = 1
x �x�−1 is a bounded continuous function,
i j
this is clear.
Then consider the same thing for a larger k:
(7.7) Dα �x�p ϕ ∈ C00 (Rn ) ∀ |α| = p , 0 ≤ p ≤ k
⇔ �x�p Dα ϕ ∈ C00 (Rn ) ∀ |α| = p , 0 ≤ p ≤ k .
�
I leave you to check this as Problem 7.1.
Corollary 7.2. For any k ∈ N the norms
�
��x�k ϕ�C k and �xα Dxβ ϕ�∞
|α|≤k,
|β|≤k

are equivalent.
Proof. Any reasonable proof of (7.2) shows that the norms
�
��x�k ϕ�C k and ��x�k Dβ ϕ�∞
|β|≤k

are equivalent. Since there are positive constants such that

⎛ ⎞ ⎛ ⎞
� �
C1 ⎝1 + |xα |⎠ ≤ �x�k ≤ C2 ⎝1 + |xα |⎠
|α|≤k |α|≤k

the equivalent of the norms follows.

�
Proposition 7.3. A linear functional u : S(Rn ) → C is continuous if
and only if there exist C, k such that
� � �
|u(ϕ)| ≤ C sup �xα Dxβ ϕ� .
Rn
|α|≤k,
|β|≤k
44 RICHARD B. MELROSE

Proof. This is just the equivalence of the norms, since we showed that
u ∈ S � (Rn ) if and only if
|u(ϕ)| ≤ C��x�k ϕ�C k
for some k.
�
Lemma 7.4. A linear map
T : S(Rn ) → S(Rn )
is continuous if and only if for each k there exist C and j such that if
|α| ≤ k and |β| ≤ k
� �
� � � � � � �
(7.8) sup �xα Dβ T ϕ� ≤ C sup �xα Dβ ϕ� ∀ ϕ ∈ S(Rn ).
Rn
|α� |≤j, |β � |≤j

Proof. This is Problem 7.2. �

All this messing about with norms shows that
xj : S(Rn ) → S(Rn ) and Dj : S(Rn ) → S(Rn )
are continuous.
So now we have some idea of what u ∈ S � (Rn ) means. Let’s notice
that u ∈ S � (Rn ) implies
(7.9) xj u ∈ S � (Rn ) ∀ j = 1, · · · , n
(7.10) Dj u ∈ S � (Rn ) ∀ j = 1, · · · , n
(7.11) ϕu ∈ S � (Rn ) ∀ ϕ ∈ S(Rn )
where we have to define these things in a reasonable way. Remem­
ber that u ∈ S � (Rn ) is “supposed” to be like an integral against a
“generalized function”
�
(7.12) u(ψ) = u(x)ψ(x) dx ∀ ψ ∈ S(Rn ).
Rn
Since it would be true if u were a function we define
(7.13) xj u(ψ) = u(xj ψ) ∀ ψ ∈ S(Rn ).
Then we check that xj u ∈ S � (Rn ):
|xj u(ψ)| = |u(xj ψ)|
� � �
≤C sup �xα Dβ (xj ψ)�
Rn
|α|≤k, |β|≤k

�
≤ C�
� �
sup �xα Dβ ψ � .
Rn
|α|≤k+1, |β|≤k
 LECTURE NOTES FOR 18.155, FALL 2004 45

Similarly we can define the partial derivatives by using the standard

integration by parts formula
� �
(7.14) (Dj u)(x)ϕ(x) dx = − u(x)(Dj ϕ(x)) dx
Rn Rn

if u ∈ C01 (Rn ). Thus if u ∈ S � (Rn ) again we define

Dj u(ψ) = −u(Dj ψ) ∀ ψ ∈ S(Rn ).
Then it is clear that Dj u ∈ S � (Rn ).
Iterating these definition we find that Dα , for any multi­index α,
defines a linear map
(7.15) Dα : S � (Rn ) → S � (Rn ) .
In general a linear differential operator with constant coefficients is a
sum of such “monomials”. For example Laplace’s operator is
∂2 ∂2 ∂2
Δ = − 2 − 2 − · · · − 2 = D12 + D22 + · · · + Dn2 .
∂x1 ∂x2 ∂xn
We will be interested in trying to solve differential equations such as
Δu = f ∈ S � (Rn ) .
We can also multiply u ∈ S � (Rn ) by ϕ ∈ S(Rn ), simply defining
(7.16) ϕu(ψ) = u(ϕψ) ∀ ψ ∈ S(Rn ).
For this to make sense it suffices to check that
� � � � � �
(7.17) sup �xα Dβ (ϕψ)� ≤ C sup �xα Dβ ψ � .
Rn Rn
|α|≤k, |α|≤k,
|β|≤k |β|≤k

This follows easily from Leibniz’ formula.

Now, to start thinking of u ∈ S � (Rn ) as a generalized function we
first define its support. Recall that
(7.18) supp(ψ) = clos {x ∈ Rn ; ψ(x) =
� 0} .
We can write this in another ‘weak’ way which is easier to generalize.
Namely
(7.19) / supp(u) ⇔ ∃ϕ ∈ S(Rn ) , ϕ(p) =
p∈ � 0 , ϕu = 0 .
In fact this definition makes sense for any u ∈ S � (Rn ).
Lemma 7.5. The set supp(u) defined by (7.19) is a closed subset of
Rn and reduces to (7.18) if u ∈ S(Rn ).
46 RICHARD B. MELROSE

Proof. The set defined by (7.19) is closed, since

(7.20) supp(u)� = {p ∈ Rn ; ∃ ϕ ∈ S(Rn ), ϕ(p) =
� 0, ϕu = 0}
is clearly open — the same ϕ works for nearby points. If ψ ∈ S(Rn )
we define uψ ∈ S � (Rn ), which we will again identify with ψ, by
�
(7.21) uψ (ϕ) = ϕ(x)ψ(x) dx .

Obviously uψ = 0 =⇒ ψ = 0, simply set ϕ = ψ in (7.21). Thus the

map
(7.22) S(Rn ) � ψ �−→ uψ ∈ S � (Rn )
is injective. We want to show that
(7.23) supp(uψ ) = supp(ψ)
on the left given by (7.19) and on the right by (7.18). We show first
that
supp(uψ ) ⊂ supp(ψ).
Thus, we need to see that p ∈ / supp(ψ) ⇒ p ∈ / supp(uψ ). The first
condition is that ψ(x) = 0 in a neighbourhood, U of p, hence there
is a C ∞ function ϕ with support in U and ϕ(p) = � 0. Then ϕψ ≡ 0.
Conversely suppose p ∈ / supp(uψ ). Then there exists ϕ ∈ S(Rn ) with
� 0 and ϕuψ = 0, i.e., ϕuψ (η) = 0 ∀ η ∈ S(Rn ). By the injectivity
ϕ(p) =
of S(Rn ) �→ S � (Rn ) this means ϕψ = 0, so ψ ≡ 0 in a neighborhood of
p and p ∈/ supp(ψ). �
Consider the simplest examples of distribution which are not func­
tions, namely those with support at a given point p. The obvious one
is the Dirac delta ‘function’
(7.24) δp (ϕ) = ϕ(p) ∀ ϕ ∈ S(Rn ) .
We can make many more, because Dα is local
(7.25) supp(Dα u) ⊂ supp(u) ∀ u ∈ S � (Rn ) .
Indeed, p ∈/ supp(u) ⇒ ∃ ϕ ∈ S(Rn ), ϕu ≡ 0, ϕ(p) = � 0. Thus each of
α
the distributions D δp also has support contained in {p}. In fact none
of them vanish, and they are all linearly independent.
 LECTURE NOTES FOR 18.155, FALL 2004 47

8. Convolution and density

We have defined an inclusion map
(8.1) �
n � n
S(R ) � ϕ �−→ uϕ ∈ S (R ), uϕ (ψ) = ϕ(x)ψ(x) dx ∀ ψ ∈ S(Rn ).
Rn

This allows us to ‘think of’ S(Rn ) as a subspace of S � (Rn ); that is we

habitually identify uϕ with ϕ. We can do this because we know (8.1)
to be injective. We can extend the map (8.1) to include bigger spaces
C00 (Rn ) � ϕ �−→ uϕ ∈ S � (Rn )
Lp (Rn ) � ϕ �−→ uϕ ∈ S � (Rn )
(8.2) M (Rn ) � µ �−→ uµ ∈ S � (Rn )
�
uµ (ψ) = ψ dµ ,
Rn
but we need to know that these maps are injective before we can forget
about them.
We can see this using convolution. This is a sort of ‘product’ of
functions. To begin with, suppose v ∈ C00 (Rn ) and ψ ∈ S(Rn ). We
define a new function by ‘averaging v with respect to ψ:’
�
(8.3) v ∗ ψ(x) = v(x − y)ψ(y) dy .
Rn

The integral converges by dominated convergence, namely ψ(y) is in­

tegrable and v is bounded,
|v(x − y)ψ(y)| ≤ �v�C00 |ψ(y)| .
We can use the same sort of estimates to show that v ∗ ψ is continuous.
Fix x ∈ Rn ,

(8.4) v ∗ ψ(x + x� ) − v ∗ ψ(x)

�
= (v(x + x� − y) − v(x − y))ψ(y) dy .

To see that this is small for x� small, we split the integral into two
pieces. Since ψ is very small near infinity, given � > 0 we can choose
R so large that
�
(8.5) �v�∞ · |ψ(y)| dy ≤ �/4 .
|y]|≥R

The set |y | ≤ R is compact and if |x| ≤ R� , |x� | ≤ 1 then |x + x� − y| ≤

R + R� + 1. A continuous function is uniformly continuous on any
48 RICHARD B. MELROSE

compact set, so we can chose δ > 0 such that

�
�
(8.6) sup |v(x + x − y) − v(x − y)| · |ψ(y)| dy < �/2 .
|x� |<δ |y|≤R
|y|≤R

Combining (8.5) and (8.6) we conclude that v∗ψ is continuous. Finally,

we conclude that
(8.7) v ∈ C00 (Rn ) ⇒ v ∗ ψ ∈ C00 (Rn ) .
For this we need to show that v ∗ ψ is small at infinity, which follows
from the fact that v is small at infinity. Namely given � > 0 there exists
R > 0 such that |v(y)| ≤ � if |y | ≥ R. Divide the integral defining the
convolution into two
� �
|v ∗ ψ(x)| ≤ u(y)ψ(x − y)dy + |u(y)ψ(x − y)|dy
|y |>R y<R
≤ �/2�ψ�∞ + �u�∞ sup |ψ|.
B(x,R)

Since ψ ∈ S(Rn ) the last constant tends to 0 as |x| → ∞.

We can do much better than this! Assuming |x� | ≤ 1 we can use
Taylor’s formula with remainder to write
� � n
� d �
(8.8) ψ(z + x ) − ψ(z) = ψ(z + tx� ) dt = xj · ψ̃j (z, x� ) .
0 dt j=1

As Problem 23 I ask you to check carefully that

(8.9) ψj (z; x� ) ∈ S(Rn ) depends continuously on x� in |x� | ≤ 1 .
Going back to (8.3))we can use the translation and reflection­invariance
of Lebesgue measure to rewrite the integral (by changing variable) as
�
(8.10) v ∗ ψ(x) = v(y)ψ(x − y) dy .
Rn

This reverses the role of v and ψ and shows that if both v and ψ are in
S(Rn ) then v ∗ ψ = ψ ∗ v.
Using this formula on (8.4) we find
(8.11) �
�
v ∗ ψ(x + x ) − v ∗ ψ(x) = v(y)(ψ(x + x� − y) − ψ(x − y)) dy
n
� � n
�
�
= xj v(y)ψ̃j (x − y, x ) dy = xj (v ∗ ψj (·; x� )(x) .
j=1 Rn j=1
 LECTURE NOTES FOR 18.155, FALL 2004 49

From (8.9) and what we have already shown, v ∗ ψ(·; x� ) is continuous

in both variables, and is in C00 (Rn ) in the first. Thus
(8.12) v ∈ C00 (Rn ) , ψ ∈ S(Rn ) ⇒ v ∗ ψ ∈ C01 (Rn ) .
In fact we also see that
∂ ∂ψ
(8.13) v∗ψ =v∗ .
∂xj ∂xj
Thus v ∗ ψ inherits its regularity from ψ.
Proposition 8.1. If v ∈ C00 (Rn ) and ψ ∈ S(Rn ) then
�
(8.14) v ∗ ψ ∈ C0∞ (Rn ) = C0k (Rn ) .
k≥0

Proof. This follows from (8.12), (8.13) and induction. �

Now, let us make a more special choice of ψ. We have shown the
existence of
(8.15) ϕ ∈ Cc∞ (Rn ) , ϕ ≥ 0 , supp(ϕ) ⊂ {|x| ≤ 1} .
�
We can also assume Rn ϕ dx = 1, by multiplying by a positive constant.
Now consider
�x�
−n
(8.16) ϕt (x) = t ϕ 1 ≥ t > 0.
t
This has all the same properties, except that
�
(8.17) supp ϕt ⊂ {|x| ≤ t} , ϕt dx = 1 .

Proposition 8.2. If v ∈ C00 (Rn ) then as t → 0, vt = v ∗ ϕt → v in

C00 (Rn ).
Proof. using (8.17) we can write the difference as
�
(8.18) |vt (x) − v(x)| = | (v(x − y) − v(x))ϕt (y) dy |
Rn
≤ sup |v(x − y) − v(x)| → 0.
|y|≤t

Here we have used the fact that ϕt ≥ 0 has support in |y | ≤ t and has
integral 1. Thus vt → v uniformly on any set on which v is uniformly
continuous, namel Rn ! �
Corollary 8.3. C0k (Rn ) is dense in C0p (Rn ) for any k ≥ p.
Proposition 8.4. S(Rn ) is dense in C0k (Rn ) for any k ≥ 0.
50 RICHARD B. MELROSE

Proof. Take k = 0 first. The subspace Cc0 (Rn ) is dense in C00 (Rn ), by
cutting off outside a large ball. If v ∈ Cc0 (Rn ) has support in {|x| ≤ R}
then
v ∗ ϕt ∈ Cc∞ (Rn ) ⊂ S(Rn )
has support in {|x| ≤ R + 1}. Since v ∗ ϕt → v the result follows for
k = 0.
For k ≥ 1 the same argument works, since Dα (v ∗ ϕt ) = (Dα V ) ∗
ϕt . �
Corollary 8.5. The map from finite Radon measures
(8.19) Mfin (Rn ) � µ �−→ uµ ∈ S � (Rn )
is injective.
Now, we want the same result for L2 (Rn ) (and maybe for Lp (Rn ),
1 ≤ p < ∞). I leave the measure­theoretic part of the argument to
you.
Proposition 8.6. Elements of L2 (Rn ) are “continuous in the mean”
i.e.,
�
(8.20) lim |u(x + t) − u(x)|2 dx = 0 .
|t|→0 Rn

This is Problem 24.

Using this we conclude that
(8.21) S(Rn ) �→ L2 (Rn ) is dense
as before. First observe that the space of L2 functions of compact
support is dense in L2 (Rn ), since
�
lim |u(x)|2 dx = 0 ∀ u ∈ L2 (Rn ) .
R→∞ |x|≥R

Then look back at the discussion of v ∗ ϕ, now v is replaced by u ∈

L2c (Rn ). The compactness of the support means that u ∈ L1 (Rn ) so in
�
(8.22) u ∗ ϕ(x) = u(x − y)ϕ(y)dy
Rn
the integral is absolutely convergent. Moreover
|u ∗ ϕ(x + x� ) − u ∗ ϕ(x)|
�� �
�
� �
= �� u(y)(ϕ(x + x − y) − ϕ(x − y)) dy ��

≤ C�u� sup |ϕ(x + x� − y) − ϕ(x − y)| → 0

|y|≤R
 LECTURE NOTES FOR 18.155, FALL 2004 51

when {|x| ≤ R} large enough. Thus u ∗ ϕ is continuous and the same

argument as before shows that
u ∗ ϕt ∈ S(Rn ) .
Now to see that u ∗ ϕt → u, assuming u has compact support (or not)
we estimate the integral
�� �
� �
|u ∗ ϕt (x) − u(x)| = � (u(x − y) − u(x))ϕt (y) dy ��
�
�
≤ |u(x − y) − u(x)| ϕt (y) dy .

Using the same argument twice

�
|u ∗ ϕt (x) − u(x)|2 dx
���
≤ |u(x − y) − u(x)| ϕt (y) |u(x − y � ) − u(x)| ϕt (y � ) dx dy dy �
�� �
2 � �
≤ |u(x − y) − u(x)| ϕt (y)ϕt (y )dx dy dy
�
≤ sup |u(x − y) − u(x)|2 dx .
|y|≤t

Note that at the second step here I have used Schwarz’s inequality with
the integrand written as the product
1/2 1/2 1/2 1/2
|u(x − y) − u(x)| ϕt (y)ϕt (y � ) · |u(x − y � ) − u(x)| ϕt (y)ϕt (y � ) .
Thus we now know that
L2 (Rn ) �→ S � (Rn ) is injective.
This means that all our usual spaces of functions ‘sit inside’ S � (Rn ).
Finally we can use convolution with ϕt to show the existence of
smooth partitions of unity. If K � U ⊂ Rn is a compact set in an
open set then we have shown the existence of ξ ∈ Cc0 (Rn ), with ξ = 1
in some neighborhood of K and ξ = 1 in some neighborhood of K and
supp(ξ) � U .
Then consider ξ ∗ ϕt for t small. In fact
supp(ξ ∗ ϕt ) ⊂ {p ∈ Rn ; dist(p, supp ξ) ≤ 2t}
and similarly, 0 ≤ ξ ∗ ϕt ≤ 1 and
ξ ∗ ϕt = 1 at p if ξ = 1 on B(p, 2t) .
Using this we get:
52 RICHARD B. MELROSE

Proposition 8.7. If Ua ⊂ Rn are open for a ∈ A and K � a∈A Ua


then there exist�finitely many ϕi ∈ Cc∞ (Rn ), with 0 ≤ ϕi ≤ 1, supp(ϕi ) ⊂
Uai such that ϕi = 1 in a neighbourhood of K.
i

Proof. By the compactness of K we may choose a finite open subcover.

Using Lemma 1.8 we may choose a continuous partition, φ�i , of unity
subordinate to this cover. Using the convolution argument above we
can replace φ�i by φ�i ∗ ϕt for t > 0. If t is sufficiently small then this is
again a partition of unity subordinate to the cover, but now smooth.
�
Next we can make a simple ‘cut off argument’ to show
Lemma 8.8. The space Cc∞ (Rn ) of C ∞ functions of compact support
is dense in S(Rn ).
Proof. Choose ϕ ∈ Cc∞ (Rn ) with ϕ(x) = 1 in |x| ≤ 1. Then given
ψ ∈ S(Rn ) consider the sequence
ψn (x) = ϕ(x/n)ψ(x) .
Clearly ψn = ψ on |x| ≤ n, so if it converges in S(Rn ) it must converge
to ψ. Suppose m ≥ n then by Leibniz’s formula13
Dxα (ψn (x) − ψm (x))
� �α � � x x �
= Dxβ ϕ( ) − ϕ( ) · Dxα−β ψ(x) .
β≤α
β n m

All derivatives of ϕ(x/n) are bounded, independent of n and ψn = ψm

in |x| ≤ n so for any p
�
α 0 |x| ≤ n
|Dx (ψn (x) − ψm (x))| ≤ −2p .
Cα,p �x� |x| ≥ n
Hence ψn is Cauchy in S(Rn ). �
Thus every element of S � (Rn ) is determined by its restriction to
Cc∞ (Rn ). The support of a tempered distribution was defined above to
be
(8.23) � 0 , ϕu = 0}� .
supp(u) = {x ∈ Rn ; ∃ ϕ ∈ S(Rn ) , ϕ(x) =
Using the preceding lemma and the construction of smooth partitions
of unity we find
Proposition 8.9. f u ∈ S � (Rn ) and supp(u) = ∅ then u = 0.
13Problem 25.
 LECTURE NOTES FOR 18.155, FALL 2004 53

Proof. From (8.23), if ψ ∈ S(Rn ), supp(ψu) ⊂ supp(u). If x � supp(u)

then, by definition, ϕu = 0 for some ϕ ∈ S(Rn ) with ϕ(x) = � 0. Thus
∞ n
ϕ=� 0 on B(x, �) for � > 0 sufficiently small. If ψ ∈ Cc (R ) has support
˜ = 0, where ψ̃ ∈ C ∞ (Rn ):
in B(x, �) then ψu = ψϕu c
�
ψ/ϕ in B(x, �)
ψ̃ =
0 elsewhere .
n ∞ n
� K � R we can find ϕj ∈ Cc (R ), supported in
Thus, given
∞
such balls,
so that j ϕj ≡ 1 on K but ϕj u = 0. For given µ ∈ Cc (Rn ) apply
this to supp(µ). Then
� �
µ= ϕj µ ⇒ u(µ) = (φj u)(µ) = 0 .
j j

Thus u = 0 on Cc∞ (Rn ), so u = 0. �

The linear space of distributions of compact support will be denoted
Cc−∞ (Rn ); it is often written E � (Rn ).
Now let us give a characterization of the ‘delta function’
δ(ϕ) = ϕ(0) ∀ ϕ ∈ S(Rn ) ,
or at least the one­dimensional subspace of S � (Rn ) it spans. This is
based on the simple observation that (xj ϕ)(0) = 0 if ϕ ∈ S(Rn )!
Proposition 8.10. If u ∈ S � (Rn ) satisfies xj u = 0, j = 1, · · · , n then
u = cδ.
Proof. The main work is in characterizing the null space of δ as a linear
functional, namely in showing that
(8.24) H = {ϕ ∈ S(Rn ); ϕ(0) = 0}
can also be written as
� n
�
�
n n
(8.25) H= ϕ ∈ S(R ); ϕ = xj ψj , ϕj ∈ S(R ) .
j=1

Clearly the right side of (8.25) is contained in the left. To see the
converse, suppose first that
(8.26) ϕ ∈ S(Rn ) , ϕ = 0 in |x| < 1 .
Then define
�
0 |x| < 1
ψ= 2
ϕ/ |x| |x| ≥ 1 .
54 RICHARD B. MELROSE

All the derivatives of 1/ |x|2 are bounded in |x| ≥ 1, so from Leibniz’s

formula it follows that ψ ∈ S(Rn ). Since
�
ϕ= xj (xj ψ)
j

this shows that ϕ of the form (8.26) is in the right side of (8.25). In
general suppose ϕ ∈ S(Rn ). Then
� t
d
ϕ(x) − ϕ(0) = ϕ(tx) dt
0 dt
(8.27) n � t
� ∂ϕ
= xj (tx) dt .
j=1 0 ∂xj

Certainly these integrals are C ∞ , but they may not decay rapidly at
infinity. However, choose µ ∈ Cc∞ (Rn ) with µ = 1 in |x| ≤ 1. Then
(8.27) becomes, if ϕ(0) = 0,
ϕ = µϕ + (1 − µ)ϕ
n � t
� ∂ϕ
= xj ψj + (1 − µ)ϕ , ψj = µ (tx) dt ∈ S(Rn ) .
j=1 0 ∂xj

Since (1 − µ)ϕ is of the form (8.26), this proves (8.25).

Our assumption on u is that xj u = 0, thus
u(ϕ) = 0 ∀ ϕ ∈ H
by (8.25). Choosing µ as above, a general ϕ ∈ S(Rn ) can be written
ϕ = ϕ(0) · µ + ϕ� , ϕ� ∈ H .
Then
u(ϕ) = ϕ(0)u(µ) ⇒ u = cδ , c = u(µ) .
�

This result is quite powerful, as we shall soon see. The Fourier

transform of an element ϕ ∈ S(Rn ) is14
�
(8.28) ϕ(ξ)
ˆ = e−ix·ξ ϕ(x) dx , ξ ∈ Rn .

14Normalizations vary, but it doesn’t matter much.

 LECTURE NOTES FOR 18.155, FALL 2004 55

The integral certainly converges, since |ϕ| ≤ C�x�−n−1 . In fact it fol­

lows easily that ϕˆ is continuous, since
� � �
��
|ϕ(ξ)
ˆ − ϕ(ξ ˆ )| ∈ �eix−ξ − e−x·ξ � |ϕ| dx
� �

→ 0 as ξ � → ξ .
In fact
Proposition 8.11. Fourier transformation, (8.28), defines a continu­
ous linear map
(8.29) F : S(Rn ) → S(Rn ) , Fϕ = ϕˆ .
Proof. Differentiating under the integral15 sign shows that
�
∂ξj ϕ(ξ)
ˆ = −i e−ix·ξ xj ϕ(x) dx .

Since the integral on the right is absolutely convergent that shows that
(remember the i’s)
(8.30) xj ϕ , ∀ ϕ ∈ S(Rn ) .
Dξj ϕˆ = −�
Similarly, if we multiply by ξj and observe that ξj e−ix·ξ = i ∂x∂ j e−ix·ξ
then integration by parts shows
�
∂ −ix·ξ
(8.31) ξj ϕ̂ = i ( e )ϕ(x) dx
∂xj
�
∂ϕ
= −i e−ix·ξ dx
∂xj
n
D j ϕ = ξj ϕ̂ , ∀ ϕ ∈ S(R ) .
�

Since xj ϕ, Dj ϕ ∈ S(Rn ) these results can be iterated, showing that

ξ α Dξβ ϕ̂ = F (−1)|β | Dα x xβ ϕ .
� �
(8.32)
� �
Thus �ξ α Dξβ ϕ̂� ≤ Cαβ sup ��x�+n+1 Dα x xβ ϕ� ≤ C��x�n+1+|β| ϕ�C |α| , which
� � � �

shows that F is continuous as a map (8.32).

�
Suppose ϕ ∈ S(Rn ). Since ϕˆ ∈ S(Rn ) we can consider the distribu­
tion u ∈ S � (Rn )
�
(8.33) u(ϕ) = ϕ(ξ)
ˆ dξ .
Rn

15See [5]
56 RICHARD B. MELROSE

The continuity of u follows from the fact that integration is continuous

and (8.29). Now observe that
�
u(xj ϕ) = x�
j ϕ(ξ) dξ
Rn
�
=− Dξj ϕˆ dξ = 0
Rn

where we use (8.30). Applying Proposition 8.10 we conclude that u =

cδ for some (universal) constant c. By definition this means
�
(8.34) ϕ(ξ)
ˆ dξ = cϕ(0) .
Rn

So what is the constant? To find it we need to work out an example.

The simplest one is
ϕ = exp(− |x|2 /2) .
Lemma 8.12. The Fourier transform of the Gaussian exp(− |x|2 /2)
is the Gaussian (2π)n/2 exp(− |ξ |2 /2).
Proof. There are two obvious methods — one uses complex analysis
(Cauchy’s theorem) the other, which I shall follow, uses the uniqueness
of solutions to ordinary differential equations.
First observe that exp(− |x|2 /2) = j exp(−x2j /2). Thus16
�

n
ˆ j ) , ψ(x) = e−x2 /2 ,
�
ϕ(ξ)
ˆ = ψ(ξ
j=1

being a function of one variable. Now ψ satisfies the differential equa­

tion
(∂x + x) ψ = 0 ,
and is the only solution of this equation up to a constant multiple. By
(8.30) and (8.31) its Fourier transform satisfies
d
∂�
x ψ + xψ = iξψ̂ + i
� ϕˆ = 0 .
dξ
2
This is the same equation, but in the ξ variable. Thus ψ̂ = ce−|ξ| /2
.
Again we need to find the constant. However,
�
2
ψ̂(0) = c = e−x /2 dx = (2π)1/2

16Really by Fubini’s theorem, but here one can use Riemann integrals.
 LECTURE NOTES FOR 18.155, FALL 2004 57

by the standard use of polar coordinates:

� � ∞ � 2π
−(x2 +y 2 )/2 2
2
c = e dx dy = e−r /2 r dr dθ = 2π .
Rn 0 0
This proves the lemma.
�
Thus we have shown that for any ϕ ∈ S(Rn )
�
(8.35) ˆ dξ = (2π)n ϕ(0) .
ϕ(ξ)
Rn

Since this is true for ϕ = exp(− |x|2 /2). The identity allows us to
invert the Fourier transform.
58 RICHARD B. MELROSE

9. Fourier inversion
It is shown above that the Fourier transform satisfies the identity
�
−n
(9.1) ϕ(0) = (2π) ϕ̂(ξ) dξ ∀ ϕ ∈ S(Rn ) .
Rn

If y ∈ Rn and ϕ ∈ S(Rn ) set ψ(x) = ϕ(x + y). The translation­

invariance of Lebesgue measure shows that
�
ψ̂(ξ) = e−ix·ξ ϕ(x + y) dx

= eiy·ξ ϕ(ξ)
ˆ .

Applied to ψ the inversion formula (9.1) becomes

�
−n
(9.2) ϕ(y) = ψ(0) = (2π) ψ̂(ξ) dξ
�
−n
= (2π) eiy·ξ ϕ(ξ)
ˆ dξ .
Rn

Theorem 9.1. Fourier transform F : S(Rn ) → S(Rn ) is an isomor­

phism with inverse
�
n n −n
(9.3) G : S(R ) → S(R ) , Gψ(y) = (2π) eiy·ξ ψ(ξ) dξ .

Proof. The identity (9.2) shows that F is 1 − 1, i.e., injective, since we

can remove ϕ from ϕ.ˆ Moreover,

(9.4) Gψ(y) = (2π)−n Fψ(−y)

So G is also a continuous linear map, G : S(Rn ) → S(Rn ). Indeed

the argument above shows that G ◦ F = Id and the same argument,
with some changes of sign, shows that F · G = Id. Thus F and G are
isomorphisms.
�

Lemma 9.2. For all ϕ, ψ ∈ S(Rn ), Paseval’s identity holds:

� �
(9.5) ϕψ dx = (2π)−n
ϕˆψˆ dξ .
Rn Rn
 LECTURE NOTES FOR 18.155, FALL 2004 59

Proof. Using the inversion formula on ϕ,

� �
−n
� ix·ξ �
ϕψ dx = (2π) e ϕ(ξ)ˆ dξ ψ(x) dx
� �
−n
= (2π) ϕ(ξ)
ˆ e−ix·ξ ψ(x) dx dξ
�
−n
= (2π) ϕ(ξ)
ˆ ϕ(ξ)ˆ dξ .

Here the integrals are absolutely convergent, justifying the exchange of

orders.
�
Proposition 9.3. Fourier transform extends to an isomorphism
(9.6) F : L2 (Rn ) → L2 (Rn ) .
Proof. Setting ϕ = ψ in (9.5) shows that
(9.7) �Fϕ�L2 = (2π)n/2 �ϕ�L2 .
In particular this proves, given the known density of S(Rn ) in L2 (Rn ),
that F is an isomorphism, with inverse G, as in (9.6).
�
For any m ∈ R
�x�m L2 (Rn ) = u ∈ S � (Rn ) ; �x�−m û ∈ L2 (Rn )
� �

is a well­defined subspace. We define the Sobolev spaces on Rn by, for

m≥0
H m (Rn ) = u ∈ L2 (Rn ) ; û = Fu ∈ �ξ�−m L2 (Rn ) .
� �
(9.8)
�
Thus H m (Rn ) ⊂ H m (Rn ) if m ≥ m� , H 0 (Rn ) = L2 (Rn ) .
Lemma 9.4. If m ∈ N is an integer, then
(9.9) u ∈ H m (Rn ) ⇔ Dα u ∈ L2 (Rn ) ∀ |α| ≤ m .
Proof. By definition, u ∈ H m (Rn ) implies that �ξ�−m û ∈ L2 (Rn ). Since
D�αu = ξαuˆ this certainly implies that Dα u ∈ L2 (Rn ) for |α| ≤ m.
Conversely if Dα u ∈ L2 (Rn ) for all |α| ≤ m then ξ α û ∈ L2 (Rn ) for all
|α| ≤ m and since
�
�ξ�m ≤ Cm |ξ α | .
|α|≤m

this in turn implies that �ξ� û ∈ L (Rn ).

m 2

�
60 RICHARD B. MELROSE

Now that we have considered the Fourier transform of Schwartz test

functions we can use the usual method, of duality, to extend it to
tempered distributions. If we set η = ψˆ then ψˆ = η and ψ = Gψ̂ = Gη
so
�
−n
ψ(x) = (2π) e−ix·ξ ψ̂(ξ) dξ
�
−n
= (2π) e−ix·ξ η(ξ) dξ = (2π)−n η̂(x).

Substituting in (9.5) we find that

� �
ϕˆη dx = ϕη
ˆ dξ .

Now, recalling how we embed S(Rn ) �→ S � (Rn ) we see that

(9.10) η) ∀ η ∈ S(Rn ) .
uϕˆ (η) = uϕ (ˆ
Definition 9.5. If u ∈ S � (Rn ) we define its Fourier transform by
(9.11) u(ϕ)
ˆ ˆ ∀ ϕ ∈ S(Rn ) .
= u(ϕ)
As a composite map, uˆ = u · F, with each term continuous, uˆ is
continuous, i.e., û ∈ S � (Rn ).
Proposition 9.6. The definition (9.7) gives an isomorphism
F : S � (Rn ) → S � (Rn ) , Fu = uˆ
satisfying the identities
(9.12) D
� αu = ξαu , x
� α u = (−1)|α| D α û .

Proof. Since û = u ◦ F and G is the 2­sided inverse of F,

(9.13) u = û ◦ G
gives the inverse to F : S � (Rn ) → S � (Rn ), showing it to be an isomor­
phism. The identities (9.12) follow from their counterparts on S(Rn ):
D
� ˆ = u((−1)|α| Dα ϕ)
α u(ϕ) = D α u(ϕ) ˆ
ˆ α ϕ) = ξ α u(ϕ)
α ϕ) = u(ξ
= u(ξ� ˆ ∀ ϕ ∈ S(Rn ) .
�
We can also define Sobolev spaces of negative order:
H m (Rn ) = u ∈ S � (Rn ) ; uˆ ∈ �ξ�−m L2 (Rn ) .
� �
(9.14)
 LECTURE NOTES FOR 18.155, FALL 2004 61

Proposition 9.7. If m ≤ 0 is an integer then u ∈ H m (Rn ) if and only

if it can be written in the form
�
(9.15) u= Dα vα , vα ∈ L2 (Rn ) .
|α|≤−m

Proof. If u ∈ S � (Rn ) is of the form (9.15) then

�
(9.16) uˆ = ξ α v̂α with vα
ˆ ∈ L2 (Rn ) .
|α|≤−m

Thus �ξ�m û = α m α m
�
|α|≤−m ξ �ξ� v̂α . Since all the factors ξ �ξ� are
2 n m 2 n
bounded, each term here is in L (R ), so �ξ� û ∈ L (R ) which is the
definition, u ∈ �ξ�−m L2 (Rn ).
Conversely, suppose u ∈ H m (Rn ), i.e., �ξ�m uˆ ∈ L2 (Rn ). The func­
tion ⎛ ⎞
�
⎝ |ξ α |⎠ · �ξ�m ∈ L2 (Rn ) (m < 0)
|α|≤−m

is bounded below by a positive constant. Thus

⎛ ⎞−1
�
v=⎝ |ξ α |⎠ uˆ ∈ L2 (Rn ) .
|α|≤−m

Each of the functions vˆα = sgn(ξ α )ˆ

v ∈ L2 (Rn ) so the identity (9.16),
and hence (9.15), follows with these choices.
�
Proposition 9.8. Each of the Sobolev spaces H m (Rn ) is a Hilbert space
with the norm and inner product
�� �1/2
2 2m
(9.17) �u�H m = |u(ξ)|
ˆ �ξ� dξ ,
Rn
�
2m
�u, v� = u(ξ)v̂(ξ)�ξ�
ˆ dξ .
Rn

The Schwartz space S(Rn ) �→ H m (Rn ) is dense for each m and the
pairing
(9.18) H m (Rn ) × H −m (Rn ) � (u, u� ) �−→
�
�
((u, u )) = uˆ� (ξ)û� (·ξ) dξ ∈ C
Rn

gives an identification (H m (Rn ))� = H −m (Rn ).

62 RICHARD B. MELROSE

Proof. The Hilbert space property follows essentially directly from the
definition (9.14) since �ξ�−m L2 (Rn ) is a Hilbert space with the norm
(9.17). Similarly the density of S in H m (Rn ) follows, since S(Rn ) dense
in L2 (Rn ) (Problem L11.P3) implies �ξ�−m S(Rn ) = S(Rn ) is dense in
�ξ�−m L2 (Rn ) and so, since F is an isomorphism in S(Rn ), S(Rn ) is
dense in H m (Rn ).
Finally observe that the pairing in (9.18) makes sense, since �ξ�−m û(ξ),
�ξ�m uˆ� (ξ) ∈ L2 (Rn ) implies
û(ξ))û� (−ξ) ∈ L1 (Rn ) .
Furthermore, by the self­duality of L2 (Rn ) each continuous linear func­
tional
U : H m (Rn ) → C , U (u) ≤ C�u�H m
can be written uniquely in the form
U (u) = ((u, u� )) for some u� ∈ H −m (Rn ) .
�
Notice that if u, u� ∈ S(Rn ) then
�
�
((u, u )) = u(x)u� (x) dx .
Rn
This is always how we “pair” functions — it is the natural pairing on
L2 (Rn ). Thus in (9.18) what we have shown is that this pairing on test
function �
S(Rn ) × S(Rn ) � (u, u� ) �−→ ((u, u� )) = u(x)u� (x) dx
Rn
extends by continuity to H m (Rn ) × H −m (Rn ) (for each fixed m) when
it identifies H −m (Rn ) as the dual of H m (Rn ). This was our ‘picture’
at the beginning.
For m > 0 the spaces H m (Rn ) represents elements of L2 (Rn ) that
have “m” derivatives in L2 (Rn ). For m < 0 the elements are ?? of “up
to −m” derivatives of L2 functions. For integers this is precisely ??.
 LECTURE NOTES FOR 18.155, FALL 2004 63

10. Sobolev embedding

The properties of Sobolev spaces are briefly discussed above. If m
is a positive integer then u ∈ H m (Rn ) ‘means’ that u has up to m
derivatives in L2 (Rn ). The question naturally arises as to the sense
in which these ‘weak’ derivatives correspond to old­fashioned ‘strong’
derivatives. Of course when m is not an integer it is a little harder
to imagine what these ‘fractional derivatives’ are. However the main
result is:
Theorem 10.1 (Sobolev embedding). If u ∈ H m (Rn ) where m > n/2
then u ∈ C00 (Rn ), i.e.,
(10.1) H m (Rn ) ⊂ C00 (Rn ) , m > n/2 .
Proof. By definition, u ∈ H m (Rn ) means v ∈ S � (Rn ) and �ξ�m uˆ(ξ) ∈
L2 (Rn ). Suppose first that u ∈ S(Rn ). The Fourier inversion formula
shows that
�� �
n
� ix·ξ
�
(2π) |u(x)| = � e û(ξ) dξ ��
�

�� �1/2 �� �1/2
≤ �ξ�2m |uˆ(ξ)|2 dξ · �ξ�−2m dξ .
Rn Rn

Now, if m > n/2 then the second integral is finite. Since the first
integral is the norm on H m (Rn ) we see that
(10.2) sup |u(x)| = �u�L∞ ≤ (2π)−n �u�H m , m > n/2 .
Rn

This is all for u ∈ S(Rn ), but S(Rn ) �→ H m (Rn ) is dense. The

estimate (10.2) shows that if uj → u in H m (Rn ), with uj ∈ S(Rn ),
then uj → u� in C00 (Rn ). In fact u� = u in S � (R n
� ) since uj → u in
2 n � 0 n
L (R ) and uj → u in C0 (R ) both imply that uj ϕ converges, so
� � �
uj ϕ → uϕ = u� ϕ ∀ ϕ ∈ S(Rn ).
Rn Rn Rn

Notice here the precise meaning of u = u� , u ∈ H m (Rn ) ⊂ L2 (Rn ),

u� ∈ C00 (Rn ). When identifying u ∈ L2 (Rn ) with the corresponding
tempered distribution, the values on any set of measure zero ‘are lost’.
Thus as functions (10.1) means that each u ∈ H m (Rn ) has a represen­
tative u� ∈ C00 (Rn ).
We can extend this to higher derivatives by noting that
64 RICHARD B. MELROSE

Proposition 10.2. If u ∈ H m (Rn ), m ∈ R, then Dα u ∈ H m−|α| (Rn )

and
(10.3) Dα : H m (Rn ) → H m−|α| (Rn )
is continuous.
Proof. First it is enough to show that each Dj defines a continuous
linear map
(10.4) Dj : H m (Rn ) → H m−1 (Rn ) ∀ j
since then (10.3) follows by composition.
If m ∈ R then u ∈ H m (Rn ) means uˆ ∈ �ξ�−m L2 (Rn ). Since D
� ju =
ξj · û, and
|ξj | �ξ�−m ≤ Cm �ξ�−m+1 ∀ m
we conclude that Dj u ∈ H m−1 (Rn ) and
�Dj u�H m−1 ≤ Cm �u�H m .
�
Applying this result we see
n
Corollary 10.3. If k ∈ N0 and m > 2
+ k then
(10.5) H m (Rn ) ⊂ C0k (Rn ) .
Proof. If |α| ≤ k, then Dα u ∈ H m−k (Rn ) ⊂ C00 (Rn ). Thus the ‘weak
derivatives’ Dα u are continuous. Still we have to check that this means
that u is itself k times continuously differentiable. In fact this again
follows from the density of S(Rn ) in H m (Rn ). The continuity in (10.3)
implies that if uj → u in H m (Rn ), m > n2 + k, then uj → u� in C0k (Rn )
(using its completeness). However u = u� as before, so u ∈ C0k (Rn ).
�
In particular we see that
�
(10.6) H ∞ (Rn ) = H m (Rn ) ⊂ C ∞ (Rn ) .
m

These functions are not in general Schwartz test functions.

Proposition 10.4. Schwartz space can be written in terms of weighted
Sobolev spaces
�
(10.7) S(Rn ) = �x�−k H k (Rn ) .
k
 LECTURE NOTES FOR 18.155, FALL 2004 65

Proof. This follows directly from (10.5) since the left side is contained
in �
�x�−k C0k−n (Rn ) ⊂ S(Rn ).
k
�
Theorem 10.5 (Schwartz representation). Any tempered distribution
can be written in the form of a finite sum
�
(10.8) u= xα Dxβ uαβ , uαβ ∈ C00 (Rn ).
|α|≤m
|β|≤m

or in the form
�
(10.9) u= Dxβ (xα vαβ ), vαβ ∈ C00 (Rn ).
|α|≤m
|β|≤m

Thus every tempered distribution is a finite sum of derivatives of

continuous functions of poynomial growth.
Proof. Essentially by definition any u ∈ S � (Rn ) is continuous with re­
spect to one of the norms ��x�k ϕ�C k . From the Sobolev embedding
theorem we deduce that, with m > k + n/2,
|u(ϕ)| ≤ C��x�k ϕ�H m ∀ ϕ ∈ S(Rn ).
This is the same as
� −k �
��x� u(ϕ)� ≤ C�ϕ�H m ∀ ϕ ∈ S(Rn ).
which shows that �x�−k u ∈ H −m (Rn ), i.e., from Proposition 9.8,
�
�x�−k u = Dα uα , uα ∈ L2 (Rn ) .
|α|≤m

In fact, choose j > n/2 and consider vα ∈ H j (Rn ) defined by v̂α =

�ξ�−j uˆα . As in the proof of Proposition 9.14 we conclude that
�
uα = Dβ u�α,β , u�α,β ∈ H j (Rn ) ⊂ C00 (Rn ) .
|β|≤j
17
Thus,
�
(10.10) u = �x�k Dαγ vγ , vγ ∈ C00 (Rn ) .
|γ|≤M

To get (10.9) we ‘commute’ the factor �x�k to the inside; since I have
not done such an argument carefully so far, let me do it as a lemma.
17This is probably the most useful form of the representation theorem!
66 RICHARD B. MELROSE

Lemma 10.6. For any γ ∈ Nn0 there are polynomials pα,γ (x) of degrees
at most |γ − α| such that
�
�x�k Dγ v = Dγ−α pα,γ �x�k−2|γ−α| v .
� �
α≤γ

Proof. In fact it is convenient to prove a more general result. Suppose

p is a polynomial of a degree at most j then there exist polynomials of
degrees at most j + |γ − α| such that
�
(10.11) p�x�k Dγ v = Dγ−α (pα,γ �x�k−2|γ−α| v) .
α≤γ

The lemma follows from this by taking p = 1.

Furthermore, the identity (10.11) is trivial when γ = 0, and proceed­
ing by induction we can suppose it is known whenever |γ| ≤ L. Taking
|γ | = L + 1,
�
Dγ = Dj Dγ |γ � | = L.
Writing the identity for γ � as
� � � � �
�
p�x�k Dγ = Dγ −α (pα� ,γ � �x�k−2|γ −α | v)
α� ≤γ �

we may differentiate with respect to xj . This gives

�
p�x�k Dγ = −Dj (p�x�k ) · Dγ v
�
�
+ Dγ−α (p�α� ,γ � �x�k−2|γ−α|+2 v) .
|α� |≤γ

The first term on the right expands to

� 1 �
(−(Dj p) · �x�k Dγ v − kpxj �x�k−2 Dγ v) .
i
We may apply the inductive hypothesis to each of these terms and
rewrite the result in the form (10.11); it is only necessary to check the
order of the polynomials, and recall that �x�2 is a polynomial of degree
2. �
Applying Lemma 10.6 to (10.10) gives (10.9), once negative powers
of �x� are absorbed into the continuous functions. Then (10.8) follows
from (10.9) and Leibniz’s formula. �
 LECTURE NOTES FOR 18.155, FALL 2004 67

11. Differential operators.

In the last third of the course we will apply what we have learned
about distributions, and a little more, to understand properties of dif­
ferential operators with constant coefficients. Before I start talking
about these, I want to prove another density result.
So far we have not defined a topology on S � (Rn ) – I will leave this
as an optional exercise.18 However we shall consider a notion of con­
vergence. Suppose uj ∈ S � (Rn ) is a sequence in S � (Rn ). It is said to
converge weakly to u ∈ S � (Rn ) if
(11.1) uj (ϕ) → u(ϕ) ∀ ϕ ∈ S(Rn ) .
There is no ‘uniformity’ assumed here, it is rather like pointwise con­
vergence (except the linearity of the functions makes it seem stronger).
Proposition 11.1. The subspace S(Rn ) ⊂ S � (Rn ) is weakly dense,
i.e., each u ∈ S � (Rn ) is the weak limit of a subspace uj ∈ S(Rn ).
Proof. We can use Schwartz representation theorem to write, for some
m depending on u,
�
u = �x�m Dα uα , uα ∈ L2 (Rn ) .
|α|≤m

We know that S(Rn ) is dense in L2 (Rn ), in the sense of metric spaces

so we can find uα,j ∈ S(Rn ), uα,j → uα in L2 (Rn ). The density result
then follows from the basic properties of weak convergence. �
Proposition 11.2. If uj → u and u�j → u� weakly in S � (Rn ) then
cuj → cu, uj + u�j → u + u� , Dα uj → Dα u and �x�m uj → �x�m u weakly
in S � (Rn ).
Proof. This follows by writing everyting in terms of pairings, for exam­
ple if ϕ ∈ S(Rn )
Dα uj (ϕ) = uj ((−1)(α) Dα ϕ) → u((−1)(α) Dα ϕ) = Dα u(ϕ) .
�
This weak density shows that our definition of Dj , and xj × are
unique if we require Proposition 11.2 to hold.
We have discussed differentiation as an operator (meaning just a
linear map between spaces of function­like objects)
Dj : S � (Rn ) → S � (Rn ) .
18Problem 34.
68 RICHARD B. MELROSE

Any polynomial on Rn
�
p(ξ) = pα ξ α , pα ∈ C
|α|≤m

defines a differential operator19

�
(11.2) p(D)u = pα D α u .
|α|≤m

Before discussing any general theorems let me consider some exam­

ples.
(11.3) On R2 , ∂ = ∂x + i∂y “d­bar operator”
� n
n
(11.4) on R , Δ = Dj2 “Laplacian”
j=1
n n+1
(11.5) on R × R = R , Dt2 − Δ“Wave operator”
(11.6) onR × Rn = Rn+1 , ∂t + Δ“Heat operator”
(11.7) on R × Rn = Rn+1 , Dt + Δ“Schr¨odinger operator”
Functions, or distributions, satisfying ∂u = 0 are said to be holo­
morphic, those satisfying Δu = 0 are said to be harmonic.
Definition 11.3. An element E ∈ S � (Rn ) satisfying
(11.8) P (D)E = δ
is said to be a (tempered) fundamental solution of P (D).
Theorem 11.4 (without proof). Every non­zero constant coefficient
differential operator has a tempered fundamental solution.
This is quite hard to prove and not as interetsing as it might seem.
We will however give lots of examples, starting with ∂. Consider the
function
1
(11.9) E(x, y) = (x + iy)−1 , (x, y) �= 0 .
2π
Lemma 11.5. E(x, y) is locally integrable and so defines E ∈ S � (R2 )
by
�
1
(11.10) E(ϕ) = (x + iy)−1 ϕ(x, y) dx dy ,
2π R2
and E so defined is a tempered fundamental solution of ∂.
19More correctly a partial differential operator with constant coefficients.
 LECTURE NOTES FOR 18.155, FALL 2004 69

Proof. Since (x+iy)−1 is smooth and bounded away from the origin the
local integrability follows from the estimate, using polar coordinates,
� � 2π � 1
dx dy r dr dθ
(11.11) = = 2π .
|(x,y)|≤1 |x + iy| 0 0 r
Differentiating directly in the region where it is smooth,
∂x (x + iy)−1 = −(x + iy)−2 , ∂y (x + iy)−1 = −i(x ∈ iy)−2
� 0.20
so indeed, ∂E = 0 in (x, y) =
The derivative is really defined by
(11.12) (∂E)(ϕ) = E(−∂ϕ)
�
1
= lim − (x + iy)−1 ∂ϕ dx dy .
�↓0 2π |x|≥�
|y|≥�

Here I have cut the space {|x| ≤ � , |y | ≤ �} out of the integral and used
the local integrability in taking the limit as � ↓ 0. Integrating by parts
in x we find
� �
−1
− (x + iy) ∂x ϕ dx dy = (∂x (x + iy)−1 )ϕ dx dy
|x|≥� |x|≥�
|y|≥� |y|≥�
� �
−1
+ (x + iy) ϕ(x, y) dy − (x + iy)−1 ϕ(x, y) dy .
|y|≤� |y|≤�
x=� x=−�

There is a corrsponding formula for integration by parts in y so,

recalling that ∂E = 0 away from (0, 0),

(11.13) 2π∂E(ϕ) =
�
lim [(� + iy)−1 ϕ(�, y) − (−� + iy)−1 ϕ(−�, y)] dy
�↓0 |y|≤�
�
+ i lim [(x + i�)−1 ϕ(x, �) − (x − i�)−1 ϕ(x, �)] dx ,
�↓0 |x|≤�

assuming that both limits exist. Now, we can write

ϕ(x, y) = ϕ(0, 0) + xψ1 (x1 y) + yψ2 (x, y) .
Replacing ϕ by either xψ1 or yψ2 in (11.13) both limits are zero. For
example
�� � �
−1
� �
�
� (� + iy) �ψ1 (�, y) dy �� ≤ |ψ1 | → 0 .
|y|≤� |y|≤�

20Thus at this stage we know ∂E must be a sum of derivatives of δ.

70 RICHARD B. MELROSE

Thus we get the same result in (11.13) by replacing ϕ(x, y) by ϕ(0, 0).
Then 2π∂E(ϕ) = cϕ(0),
� �
dy dy
c = lim 2� 2 2
= lim < 2
= 2π .
�↓0 |y|≤� � + y �↓0 |y|≤1 1 + y

�
Let me remind you that we have already discussed the convolution
of functions
�
u ∗ v(x) = u(x − y)v(y) dy = v ∗ u(x) .

This makes sense provided u is of slow growth and s ∈ S(Rn ). In fact

we can rewrite the definition in terms of pairing
(11.14) (u ∗ ϕ)(x) = �u, ϕ(x − ·)�
where the · indicates the variable in the pairing.
Theorem 11.6 (Hörmander, Theorem 4.1.1). If u ∈ S � (Rn ) and ϕ ∈
S(Rn ) then u ∗ ϕ ∈ S � (Rn ) ∩ C ∞ (Rn ) and if supp(ϕ) � Rn
supp(u ∗ ϕ) ⊂ supp(u) + supp(ϕ) .
For any multi­index α
Dα (u ∗ ϕ) = Dα u ∗ ϕ = u ∗ Dα ϕ .
Proof. If ϕ ∈ S(Rn ) then for any fixed x ∈ Rn ,
ϕ(x − ·) ∈ S(Rn ) .
Indeed the seminorm estimates required are
sup(1 + |y|2 )k/2 |Dα y ϕ(x − y)| < ∞ ∀ α, k > 0 .
y

Since Dα y ϕ(x − y) = (−1)|α| (Dα ϕ)(x − y) and

(1 + |y|2 ) ≤ (1 + |x − y |2 )(1 + |x|2 )
we conclude that
�(1 + |y |2 )k/2 Dα y (x − y)�L∞ ≤ (1 + |x|2 )k/2 ��y�k Dα y ϕ(y)�L∞ .
The continuity of u ∈ S � (Rn ) means that for some k
|u(ϕ)| ≤ C sup �(y)k Dα ϕ�L∞
|α|≤k

so it follows that
(11.15) |u ∗ ϕ(x)| = |�u, ϕ(x − ·)�| ≤ C(1 + |x|2 )k/2 .
 LECTURE NOTES FOR 18.155, FALL 2004 71

The argument above shows that x �→ ϕ(x− ·) is a continuous function

of x ∈ Rn with values in S(Rn ), so u ∗ ϕ is continuous and satisfies
(11.15). It is therefore an element of S � (Rn ).
Differentiability follows in the same way since for each j, with ej the
jth unit vector
ϕ(x + sej − y) − ϕ(x − y)
∈ S(Rn )
s
is continuous in x ∈ Rn , s ∈ R. Thus, u ∗ ϕ has continuous partial
derivatives and
Dj u ∗ ϕ = u ∗ Dj ϕ .
The same argument then shows that u∗ϕ ∈ C ∞ (Rn ). That Dj (u∗ϕ) =
Dj u ∗ ϕ follows from the definition of derivative of distributions
Dj (u ∗ ϕ(x)) = (u ∗ Dj ϕ)(x)
= �u, Dxj ϕ(x − y)� = −�u(y), Dyj ϕ(x − y)�y
= (Dj u) ∗ ϕ .
Finally consider the support property. Here we are assuming that
supp(ϕ) is compact; we also know that supp(u) is a closed set. We
have to show that
(11.16) x∈
/ supp(u) + supp(ϕ)
implies u ∗ ϕ(x ) = 0 for x� near x. Now (11.16) just means that
�

(11.17) supp ϕ(x − ·) ∩ supp(u) = φ ,

Since supp ϕ(x − ·) = {y ∈ Rn ; x − y ∈ supp(ϕ)}, so both statements
mean that there is no y ∈ supp(ϕ) with x − y ∈ supp(u). This can also
be written
supp(ϕ) ∩ supp u(x − ·) = φ
and as we showed when discussing supports implies
u ∗ ϕ(x� ) = �u(x� − ·), ϕ� = 0 .
From (11.17) this is an open condition on x� , so the support property
follows.
�
Now suppose ϕ, ψ ∈ S(Rn ) and u ∈ S � (Rn ). Then
(11.18) (u ∗ ϕ) ∗ ψ = u ∗ (ϕ ∗ ψ) .
This is really Hörmander’s Lemma 4.1.3 and Theorem 4.1.2; I ask you
to prove it as Problem 35.
We have shown that u ∗ ϕ is C ∞ if u ∈ S � (Rn ) and ϕ ∈ S(Rn ),
i.e., the regularity of u ∗ ϕ follows from the regularity of one of the
72 RICHARD B. MELROSE

factors. This makes it reasonable to expect that u ∗ v can be defined

when u ∈ S � (Rn ), v ∈ S � (Rn ) and one of them has compact support.
If v ∈ Cc∞ (Rn ) and ϕ ∈ S(Rn ) then
� �
u ∗ v(ϕ) = �u(·), v(x − ·)�ϕ(x) dx = �u(·), v(x − ·)�v̌ϕ(−x) dx

where ϕ(z)
ˇ = ϕ(−z). In fact using Problem 35,
(11.19) u ∗ v(ϕ) = ((u ∗ v) ∗ ϕ)(0)
ˇ = (u ∗ (v ∗ ϕ))(0)
ˇ .
Here, v, ϕ are both smooth, but notice
Lemma 11.7. If v ∈ S � (Rn ) has compact support and ϕ ∈ S(Rn ) then
v ∗ ϕ ∈ S(Rn ).
Proof. Since v ∈ S � (Rn ) has compact support there exists χ ∈ Cc∞ (Rn )
such that χv = v. Then
v ∗ ϕ(x) = (χv) ∗ ϕ(x) = �χv(y), ϕ(x − y)�y
= �u(y), χ(y)ϕ(x − y)�y .
Thus, for some k,
|v ∗ ϕ(x)| ≤ C�χ(y)ϕ(x − y)�(k)
where � �(k) is one of our norms on S(Rn ). Since χ is supported in
some large ball,
�χ(y)ϕ(x − y)�(k)
� �
≤ sup ��y�k Dα y (χ(y)ϕ(x − y))�
|α|≤k

≤ C sup sup |(Dα ϕ)(x − y)|

|y|≤R |α|≤k

≤ CN sup (1 + |x − y |2 )−N/2
|y|≤R

≤ CN (1 + |x|2 )−N/2 .
Thus (1 + |x|2 )N/2 |v ∗ ϕ| is bounded for each N . The same argument
applies to the derivative using Theorem 11.6, so
v ∗ ϕ ∈ S(Rn ) .
�
In fact we get a little more, since we see that for each k there exists
�
k and C (depending on k and v) such that
�v ∗ ϕ�(k) ≤ C�ϕ�(k� ) .
 LECTURE NOTES FOR 18.155, FALL 2004 73

This means that

v∗ : S(Rn ) → S(Rn )
is a continuous linear map.
Now (11.19) allows us to define u∗v when u ∈ S � (Rn ) and v ∈ S � (Rn )
has compact support by
u ∗ v(ϕ) = u ∗ (v ∗ ϕ)(0)
ˇ .
Using the continuity above, I ask you to check that u ∗ v ∈ S � (Rn ) in
Problem 36. For the moment let me assume that this convolution has
the same properties as before – I ask you to check the main parts of
this in Problem 37.
Recall that E ∈ S � (Rn ) is a fundamental situation for P (D), a con­
stant coefficient differential operator, if P (D)E = δ. We also use a
weaker notion.
Definition 11.8. A parametrix for a constant coefficient differential
operator P (D) is a distribution F ∈ S � (Rn ) such that
(11.20) P (D)F = δ + ψ , ψ ∈ C ∞ (Rn ) .
An operator P (D) is said to be hypoelliptic if it has a parametrix sat­
isfying
(11.21) sing supp(F ) ⊂ {0} ,
� n
where for any u ∈ S (R )

(11.22) (sing supp(u))� = {x ∈ Rn ; ∃ ϕ ∈ Cc∞ (Rn ) ,

� 0, ϕu ∈ Cc∞ (Rn )} .
ϕ(x) =
Since the same ϕ must work for nearby points in (11.22), the set
sing supp(u) is closed. Furthermore
(11.23) sing supp(u) ⊂ supp(u) .
As Problem 37 I ask you to show that if K � Rn and K ∩sing supp(u) =
φ the ∃ ϕ ∈ Cc∞ (Rn ) with ϕ(x) = 1 in a neighbourhood of K such that
ϕu ∈ Cc∞ (Rn ). In particular
(11.24) sing supp(u) = φ ⇒ u ∈ S � (Rn ) ∩ C ∞ (Rn ) .
Theorem 11.9. If P (D) is hypoelliptic then
(11.25) sing supp(u) = sing supp(P (D)u) ∀ u ∈ S � (Rn ) .
Proof. One half of this is true for any differential operator:
Lemma 11.10. If u ∈ S � (Rn ) then for any polynomial
(11.26) sing supp(P (D)u) ⊂ sing supp(u) ∀ u ∈ S � (Rn ) .
74 RICHARD B. MELROSE

�
Proof. We must show that x ∈ / sing supp(u) ⇒ x ∈/ sing supp(P (D)u).
/ sing supp(u) we can find ϕ ∈ Cc∞ (Rn ), ϕ ≡ 1 near x, such
Now, if x ∈
that ϕu ∈ Cc∞ (Rn ). Then
P (D)u = P (D)(ϕu + (1 − ϕ)u)
= P (D)(ϕu) + P (D)((1 − ϕ)u) .
The first term is C ∞ and x ∈
/ supp(P (D)((1−ϕ)u)), so x ∈
/ sing supp(P (D)u).
�
It remains to show the converse of (11.26) where P (D) is assumed to
be hypoelliptic. Take F , a parametrix for P (D) with sing supp u ⊂ {0}
and assume, or rather arrange, that F have compact support. In fact
if x ∈
/ sing supp(P (D)u) we can arrange that
(supp(F ) + x) ∩ sing supp(P (D)u) = φ .
Now P (D)F = δψ with ψ ∈ Cc∞ (Rn ) so
u = δ ∗ u = (P (D)F ) ∗ u − ψ ∗ u.
Since ψ ∗ u ∈ C ∞ it suffices to show that x¯ ∈
/ sing supp ((P (D)u) ∗ f ) .
Take ϕ ∈ Cc∞ (Rn ) with ϕf ∈ C ∞ , f = P (D)u but
(supp F + x) ∩ supp(ϕ) = 0 .
Then f = f1 + f2 , f1 = ϕf ∈ Cc∞ (Rn ) so
f ∗ F = f1 ∗ F + f2 ∗ F
where f1 ∗ F ∈ C ∞ (Rn ) and x ∈
/ supp(f2 ∗ F ). It follows that x ∈
/
sing supp(u).
Example 11.1. If u is holomorphic on Rn , ∂u = 0, then u ∈ C ∞ (Rn ).
Recall from last time that a differential operator P (D) is said to be
hypoelliptic if there exists F ∈ S � (Rn ) with
(11.27) P (D)F − δ ∈ C ∞ (Rn ) and sing supp(F ) ⊂ {0} .
The second condition here means that if ϕ ∈ Cc∞ (Rn ) and ϕ(x) = 1 in
|x| < � for some � > 0 then (1−ϕ)F ∈ C ∞ (Rn ). Since P (D)((1−ϕ)F ) ∈
C ∞ (Rn ) we conclude that
P (D)(ϕF ) − δ ∈ Cc∞ (Rn )
and we may well suppose that F , replaced now by ϕF , has compact
support. Last time I showed that
If P (D) is hypoelliptic and u ∈ S � (Rn ) then
sing supp(u) = sing supp(P (D)u) .
 LECTURE NOTES FOR 18.155, FALL 2004 75

I will remind you of the proof later.

First however I want to discuss the important notion of ellipticity.
Remember that P (D) is ‘really’ just a polynomial, called the charac­
teristic polynomial
�
P (ξ) = Cα ξ α .
|α|≤m

It has the property

P� u(ξ) ∀ u ∈ S � (Rn ) .
(D)u(ξ) = P (ξ)ˆ
This shows (if it isn’t already obvious) that we can remove P (ξ) from
P (D) thought of as an operator on S � (Rn ).
We can think of inverting P (D) by dividing by P (ξ). This works
well provided P (ξ) �= 0, for all ξ ∈ Rn . An example of this is
n
�
2
P (ξ) = |ξ| + 1 = +1 .
j=1

However even the Laplacian, Δ = nj=1 Dj2 , does not satisfy this rather
�
stringent condition.
It is reasonable to expect the top order derivatives to be the most
important. We therefore consider
�
Pm (ξ) = Cα ξ α
|α|=m

the leading part, or principal symbol, of P (D).

Definition 11.11. A polynomial P (ξ), or P (D), is said to be elliptic
� ξ ∈ Rn .
of order m provided Pm (ξ) �= 0 for all 0 =
So what I want to show today is
Theorem 11.12. Every elliptic differential operator P (D) is hypoel­
liptic.
We want to find a parametrix for P (D); we already know that we
might as well suppose that F has compact support. Taking the Fourier
transform of (11.27) we see that F� should satisfy
(11.28) � ψ� ∈ S(Rn ) .
P (ξ)F�(ξ) = 1 + ψ,
Here we use the fact that ψ ∈ Cc∞ (Rn ) ⊂ S(Rn ), so ψ� ∈ S(Rn ) too.
First suppose that P (ξ) = Pm (ξ) is actually homogeneous of degree
m. Thus
Pm (ξ) = |ξ|m Pm (ξ�), ξ� = ξ/ |ξ | , ξ =
� 0.
76 RICHARD B. MELROSE

The assumption at ellipticity means that

(11.29) � 0 ∀ ξ� ∈ S n−1 = {ξ ∈ Rn ; |ξ | = 1} .
Pm (ξ�) =
Since S n−1 is compact and Pm is continuous
� �
(11.30) �Pm (ξ�)� ≥ C > 0 ∀ ξ� ∈ S n−1 ,
� �

for some constant C. Using homogeneity

� �
m
(11.31) �Pm (ξ�)� ≥ C |ξ | , C > 0 ∀ ξ ∈ Rn .
� �

Now, to get F� from (11.28) we want to divide by Pm (ξ) or multiply

by 1/Pm (ξ). The only problem with defining 1/Pm (ξ) is at ξ = 0. We
shall simply avoid this unfortunate point by choosing P ∈ Cc∞ (Rn ) as
before, with ϕ(ξ) = 1 in |ξ | ≤ 1.
Lemma 11.13. If Pm (ξ) is homogeneous of degree m and elliptic then
(1 − ϕ(ξ))
(11.32) Q(ξ) = ∈ S � (Rn )
Pm (ξ)
is the Fourier transform of a parametrix for Pm (D), satisfying (11.27).
Proof. Clearly Q(ξ) is a continuous function and |Q(ξ)| ≤ C(1+|ξ |)−m ∀ ξ ∈
Rn , so Q ∈ S � (Rn ). It therefore is the Fourier transform of some
F ∈ S � (Rn ). Furthermore
�
Pm (D)F (ξ) = Pm (ξ)F� = Pm (ξ)Q(ξ)
= 1 − ϕ(ξ) ,
⇒ Pm (D)F = δ + ψ , ψ�(ξ) = −ϕ(ξ) .
Since ϕ ∈ Cc∞ (Rn ) ⊂ S(Rn ), ψ ∈ S(Rn ) ⊂ C ∞ (Rn ). Thus F is a
parametrix for Pm (D). We still need to show the ‘hard part’ that
(11.33) sing supp(F ) ⊂ {0} .
�
We can show (11.33) by considering the distributions xα F . The idea
is that for |α| large, xα vanishes rather rapidly at the origin and this
should ‘weaken’ the singularity of F there. In fact we shall show that
(11.34) xα F ∈ H |α|+m−n−1 (Rn ) , |α| > n + 1 − m .
If you recall, these Sobolev spaces are defined in terms of the Fourier
transform, namely we must show that
α F ∈ �ξ�−|α|−m+n+1 L2 (Rn ) .
x�
 LECTURE NOTES FOR 18.155, FALL 2004 77

Now x� α F = (−1)|α| D α F
ξ , so what we need to cinsider is the behaviour
�
of the derivatives of F�, which is just Q(ξ) in (11.32).
Lemma 11.14. Let P (ξ) be a polynomial of degree m satisfying
(11.35) |P (ξ)| ≥ C |ξ|m in |ξ | > 1/C for some C > 0 ,
then for some constants Cα
� �
� α 1 � −m−|α|
(11.36) � P (ξ) � ≤ Cα |ξ | in |ξ | > 1/C .
�D �

Proof. The estimate in (11.36) for α = 0 is just (11.35). To prove the

higher estimates that for each α there is a polynomial of degree at most
(m − 1) |α| such that
1 Lα (ξ)
(11.37) Dα = .
P (ξ) (P (ξ))1+|α|
Once we know (11.37) we get (11.36) straight away since
� � � (m−1)|α|
� ≤ Cα |ξ |
� α 1 �
�D
� P (ξ) � m(1+|α|)
≤ Cα |ξ |−m−|α| .
C 1+ | α | |ξ|
We can prove (11.37) by induction, since it is certainly true for α = 0.
Suppose it is true for |α| ≤ k. To get the same identity for each β with
|β | = k +1 it is enough to differentiate one of the identities with |α| = k
once. Thus
1 1 Dj Lα (ξ) (1 + |α|)Lα Dj P (ξ)
Dβ = Dj Dα = 1+|α|
− .
P (ξ) P (ξ) P (ξ) (P (ξ))2+|α|
Since Lβ (ξ) = P (ξ)Dj Lα (ξ) − (1 + |α|)Lα (ξ)Dj P (ξ) is a polynomial of
degree at most (m − 1) |α| + m − 1 = (m − 1) |β | this proves the lemma.
�
1−ϕ
Going backwards, observe that Q(ξ) = Pm (ξ)
is smooth in |ξ | ≤ 1/C,
so (11.36) implies that
(11.38) |Dα Q(ξ)| ≤ Cα (1 + |ξ |)−m−|α|
n
⇒ �ξ�� Dα Q ∈ L2 (Rn ) if � − m − |α| < − ,
2
which certainly holds if � = |α| + m − n − 1, giving (11.34). Now, by
Sobolev’s embedding theorem
n
xα F ∈ C k if |α| > n + 1 − m + k + .
2
78 RICHARD B. MELROSE

In particular this means that if we choose µ ∈ Cc∞ (Rn ) with 0 ∈

/ supp(µ)
2k
then for every k, µ/ |x| is smooth and
µ
µF = 2k |x|2k F ∈ C 2�−2n , � > n .
|x|
Thus µF ∈ Cc∞ (Rn ) and this is what we wanted to show, sing supp(F ) ⊂
{0}.
So now we have actually proved that Pm (D) is hypoelliptic if it is
elliptic. Rather than go through the proof again to make sure, let me
go on to the general case and in doing so review it.
Proof. Proof of theorem. We need to show that if P (ξ) is elliptic then
P (D) has a parametrix F as in (11.27). From the discussion above the
ellipticity of P (ξ) implies (and is equivalent to)
|Pm (ξ)| ≥ c |ξ |m , c > 0 .
On the other hand
�
P (ξ) − Pm (ξ) = Cα ξ α
|α|<m

is a polynomial of degree at most m − 1, so

|P (ξ) − Pm (ξ)| 2 ≤ C � (1 + |ξ |)m−1 .
This means that id C > 0 is large enough then in |ξ| > C, C � (1 +
|ξ|)m−1 < 2c |ξ |m , so
|P (ξ)| ≥ |Pm (ξ)| − |P (ξ) − Pm (ξ)|
c
≥ c |ξ|m − C � (1 + |ξ|)m−1 ≥ |ξ |m .
2
This means that P (ξ) itself satisfies the conditions of Lemma 11.14.
Thus if ϕ ∈ Cc∞ (Rn ) is equal to 1 in a large enough ball then Q(xi) =
(1 − ϕ(ξ))/P (ξ) in C ∞ and satisfies (11.36) which can be written
|Dα Q(ξ)| ≤ Cα (1 + |ξ |)m−|α| .
The discussion above now shows that defining F ∈ S � (Rn ) by F�(ξ) =
Q(ξ) gives a solution to (11.27).
�
The last step in the proof is to show that if F ∈ S � (Rn ) has compact
support, and satisfies (11.27), then
u ∈ S(Rn ) , P (D)u ∈ S � (Rn ) ∩ C ∞ (Rn )
⇒ u = F ∗ (P (D)u) − ψ ∗ u ∈ C ∞ (Rn ) .
Let me refine this result a little bit.
 LECTURE NOTES FOR 18.155, FALL 2004 79

Proposition 11.15. If f ∈ S � (Rn ) and µ ∈ S � (Rn ) has compact sup­

port then
sing supp(u ∗ f ) ⊂ sing supp(u) + sing supp(f ).
Proof. We need to show that p ∈ / sing supp(u) ∈ sing supp(f ) then
p∈/ sing supp(u ∗ f ). Once we can fix p, we might as well suppose that
f has compact support too. Indeed, choose a large ball B(R, 0) so that
z∈
/ B(0, R) ⇒ p ∈
/ supp(u) + B(0, R) .
This is possible by the assumed boundedness of supp(u). Then choose
ϕ ∈ Cc∞ (Rn ) with ϕ = 1 on B(0, R); it follows from Theorem L16.2, or
rather its extension to distributions, that φ ∈ / supp(u(1 − ϕ)f ), so we
can replace f by ϕf , noting that sing supp(ϕf ) ⊂ sing supp(f ). Now if
f has compact support we can choose compact neighbourhoods K1 , K2
of sing supp(u) and sing supp(f ) such that p ∈ / K1 + K2 . Furthermore
we an decompose u = u1 + u2 , f = f1 + f2 so that supp(u1 ) ⊂ K1 ,
supp(f2 ) ⊂ K2 and u2 , f2 ∈ C ∞ (Rn ). It follows that
u ∗ f = u1 ∗ f1 + u2 ∗ f2 + u1 ∗ f2 + u2 ∗ f2 .
Now, p ∈/ supp(u1 ∗ f1 ), by the support property of convolution and the
three other terms are C ∞ , since at least one of the factors is C ∞ . Thus
p∈/ sing supp(u ∗ f ). �

The most important example of a differential operator which is hy­

poelliptic, but not elliptic, is the heat operator
n
�
(11.39) ∂t + Δ = ∂t − ∂x2j .
j=1

In fact the distribution

� � 2
�
1
(4πt)n/2
exp − |x4|t t≥0
(11.40) E(t, x) =
0 t≤0
is a fundamental solution. First we need to check that E is a distri­
bution. Certainly E is C ∞ in t > 0. Moreover as t ↓ 0 in x = � 0 it
∞
vanishes with all derivatives, so it is C except at t = 0, x = 0. Since
it is clearly measurable we will check that it is locally integrable near
the origin, i.e.,
�
(11.41) E(t, x) dx dt < ∞ ,
0≤t≤1
|x|≤1
80 RICHARD B. MELROSE

since E ≥ 0. We can change variables, setting X = x/t1/2 , so dx =

tn/2 dX and the integral becomes
|X|2
� ��
1
exp(− ) dx dt < ∞ .
(4π)n/2 0 |X|≤t−1/2 4
Since E is actually bounded near infinity, it follows that E ∈ S � Rn ,
�
E(ϕ) = E(t, x)ϕ(t, x) dx dt ∀ ϕ ∈ S(Rn+1 ) .
t≥0

As before we want to compute

(11.42) (∂t + Δ)E(ϕ) = E(−∂t ϕ + Δϕ)
� ∞�
= lim E(t, x)(−∂t ϕ + Δϕ) dx dt .
E↓0 E Rn
First we check that (∂t + Δ)E = 0 in t > 0, where it is a C ∞ function.
This is a straightforward computation:
n |x|2
∂t E = − E + 2 E
2t 4t
xj 2 1 x2j
∂xj E = − E , ∂xj E = − E + 2 E
2t 2t 4t
2
n |x|
⇒ ΔE = E + 2 E .
2t 4t
Now we can integrate by parts in (11.42) to get
2
e−|x| /4E
�
(∂t + Δ)E(ϕ) = lim ϕ(E, x) dx .
E ↓0 Rn (4πE)n/2
Making the same change of variables as before, X = x/2E 1/2 ,
2
e−|x|
�
1/2
(∂t + Δ)E(ϕ) = lim ϕ(E, E X) n/2 dX .
E ↓0 Rn π
As E ↓ 0 the integral here is bounded by the integrable function
C exp(− |X |2 ), for some C > 0, so by Lebesgue’s theorem of domi­
nated convergence, conveys to the integral of the limit. This is
�
2 dx
ϕ(0, 0) · e−|x| n/2 = ϕ(0, 0) .
Rn π
Thus
(∂t + Δ)E(ϕ) = ϕ(0, 0) ⇒ (∂t + Δ)E = δt δx ,
so E is indeed a fundamental solution. Since it vanishes in t < 0 it is
canned a forward fundamrntal solution.
Let’s see what we can use it for.
 LECTURE NOTES FOR 18.155, FALL 2004 81

Proposition 11.16. If f ∈ S � Rn has compact support ∃ !u ∈ S � Rn

with supp(m) ⊂ {t ≥ −T } for some T and
(11.43) (∂t + Δ)u = f in Rn+1 .
Proof. Naturally we try u = E ∗ f . That it satisfies (11.43)follows from
the properties of convolution. Similarly if T is such that supp(f ) ⊂
{t ≥ T } then
supp(u) ⊂ supp(f ) + supp(E) ⊂ {t ≥ T ] .
So we need to show uniqueness. If u1 , u2 ∈ S � Rn in two solutions of
(11.43) then their difference v = u1 − u2 satisfies the ‘homogeneous’
equation (∂t + Δ)v = 0. Furthermore, v = 0 in t < T � for some T � .
Given any E ∈ R choose ϕ(t) ∈ C ∞ (R) with ϕ(t) = 0 in t > t + 1,
ϕ(t) = 1 in t < t and consider
Et = ϕ(t)E = F1 + F2 ,
where F1 = ψEt for some ψ ∈ Cc∞ Rn+1 ), ψ = 1 near 0. Thus F1 has
comapct support and in fact F2 ∈ SRn . I ask you to check this last
statement as Problem L18.P1.
Anyway,
(∂t + Δ)(F1 + F2 ) = δ + ψ ∈ SRn , ψt = 0 t ≤ t .
Now,
(∂t + Δ)(Et ∗ u) = 0 = u + ψt ∗ u .
�
Since supp(ψt ) ⊂ t ≥ t], the second tier here is supported in t ≥ t ≥
T � . Thus u = 0 in t < t + T � , but t is arbitrary, so u = 0. �

Notice that the assumption that u ∈ S � Rn is not redundant in the

statement of the Proposition, if we allow “large” solutions they be­
come non­unique. Problem L18.P2 asks you to apply the fundamental
solution to solve the initial value problem for the heat operator.
Next we make similar use of the fundamental solution for Laplace’s
operator. If n ≥ 3 the
(11.44) E = Cn |x|−n+2
is a fundamental solution. You should check that ΔEn = 0 in x �= 0
directly, I will show later that ΔEn = δ, for the appropriate choice of
Cn , but you can do it directly, as in the case n = 3.
Theorem 11.17. If f ∈ SRn ∃ !u ∈ C0∞ Rn such that Δu = f.
82 RICHARD B. MELROSE

Proof. Since convolution u = E ∗ f ∈ S � Rn ∩ C ∞ Rn is defined we

certainly get a solution to Δu = f this way. We need to check that
u ∈ C0∞ Rn . First we know that Δ is hypoelliptic so we can decompose
E = F1 + F2 , F1 ∈ S � Rn , supp F, � Rn
and then F2 ∈ C ∞ Rn . In fact we can see from (11.44) that
|Dα F2 (x)| ≤ Cα (1 + |x|)−n+2−|α| .
Now, F1 ∗ f ∈ SRn , as we showed before, and continuing the integral
we see that
|Dα u| ≤ |Dα F2 ∗ f | + CN (1 + |x|)−N ∀ N
≤ Cα� (1 + |x|)−n+2−|α| .
Since n > 2 it follows that u ∈ C0∞ Rn .
So only the uniqueness remains. If there are two solutions, u1 , u2 for
a given f then v = u1 − u2 ∈ C0∞ Rn satisfies Δv = 0. Since v ∈ S � Rn
we can take the Fourier transform and see that
|χ|2 v�(χ) = 0 ⇒ supp(�
v ) ⊂ {0} .
an earlier problem was to conclude from this that v� = |α|≤m Cα Dα δ
�
for some constants Cα . This in turn implies that v is a polynomial.
However the only polynomials in C00 Rn are identically 0. Thus v = 0
and uniqueness follows. �
 LECTURE NOTES FOR 18.155, FALL 2004 83

12. Cone support and wavefront set

In discussing the singular support of a tempered distibution above,
notice that
singsupp(u) = ∅
only implies that u ∈ C ∞ (Rn ), not as one might want, that u ∈ S(Rn ).
We can however ‘refine’ the concept of singular support a little to get
this.
Let us think of the sphere Sn−1 as the set of ‘asymptotic directions’ in
R . That is, we identify a point in Sn−1 with a half­line {ax̄; a ∈ (0, ∞)}
n

for 0 �= x̄ ∈ Rn . Since two points give the same half­line if and only if
they are positive multiples of each other, this means we think of the
sphere as the quotient
(12.1) Sn−1 = (Rn \ {0})/R+ .
Of course if we have a metric on Rn , for instance the usual Euclidean
metric, then we can identify Sn−1 with the unit sphere. However (12.1)
does not require a choice of metric.
Now, suppose we consider functions on Rn \{0} which are (positively)
homogeneous of degree 0. That is f (a¯ x) = f (¯
x), for all a > 0, and
they are just functions on Sn−1 . Smooth functions on Sn−1 correspond
(if you like by definition) with smooth functions on Rn \ {0} which are
homogeneous of degree 0. Let us take such a function ψ ∈ C ∞ (Rn \{0}),
ψ(ax) = ψ(x) for all a > 0. Now, to make this smooth on Rn we need
to cut it off near 0. So choose a cutoff function χ ∈ Cc∞ (Rn ), with
χ(x) = 1 in |x| < 1. Then
(12.2) ψR (x) = ψ(x)(1 − χ(x/R)) ∈ C ∞ (Rn ),
for any R > 0. This function is supported in |x| ≥ R. Now, if ψ has
support near some point ω ∈ Sn−1 then for R large the corresponding
function ψR will ‘localize near ω as a point at infinity of Rn .’ Rather
than try to understand this directly, let us consider a corresponding
analytic construction.
First of all, a function of the form ψR is a multiplier on S(Rn ). That
is,
(12.3) ψR · : S(Rn ) −→ S(Rn ).
To see this, the main problem is to estimate the derivatives at infinity,
since the product of smooth functions is smooth. This in turn amounts
to estimating the deriviatives of ψ in |x| ≥ 1. This we can do using the
homogeneity.
84 RICHARD B. MELROSE

Lemma 12.1. If ψ ∈ C ∞ (Rn \ {0}) is homogeneous of degree 0 then

(12.4) |Dα ψ | ≤ Cα |x|−|α| .
Proof. I should not have even called this a lemma. By the chain rule,
the derivative of order α is a homogeneous function of degree −|α| from
which (12.4) follows. �
For the smoothed versio, ψR , of ψ this gives the estimates
(12.5) |Dα ψR (x)| ≤ Cα �x�−|α| .
This allows us to estimate the derivatives of the product of a Schwartz
function and ψR :
(12.6) xβ Dα (ψR f )
� �α �
= Dα−γ ψR xβ Dγ f =⇒ sup |xβ Dα (ψR f )| ≤ C sup �f �k
γ≤α
γ |x|≥1

for some seminorm on S(Rn ). Thus the map (12.3) is actually continu­
ous. This continuity means that ψR is a multiplier on S � (Rn ), defined
as usual by duality:
(12.7) ψR u(f ) = u(ψR f ) ∀ f ∈ S(Rn ).
Definition 12.2. The cone­support and cone­singular­support of a tem­
pered distribution are the subsets Csp(u) ⊂ Rn ∪ Sn−1 and Css(u) ⊂
Rn ∪ Sn−1 defined by the conditions
(12.8)
Csp(u) ∩ Rn = supp(u)
(Csp(u))� ∩ Sn−1 ={ω ∈ Sn−1 ;
∃ R > 0, ψ ∈ C ∞ (Sn−1 ), ψ(ω) =
� 0, ψR u = 0},
Css(u) ∩ Rn = singsupp(u)
(Css(u))� ∩ Sn−1 ={ω ∈ Sn−1 ;
∃ R > 0, ψ ∈ C ∞ (Sn−1 ), ψ(ω) =
� 0, ψR u ∈ S(Rn )}.
That is, on the Rn part these are the same sets as before but ‘at
infinity’ they are defined by conic localization on Sn−1 .
In considering Csp(u) and Css(u) it is convenient to combine Rn
and Sn−1 into a compactification of Rn . To do so (topologically) let
us identify Rn with the interior of the unit ball with respect to the
Euclidean metric using the map
x
(12.9) Rn � x �−→ ∈ {y ∈ Rn ; |y| ≤ 1} = Bn .
�x�
 LECTURE NOTES FOR 18.155, FALL 2004 85

Clearly |x| < �x� and for 0 ≤ a < 1, |x| = a�x� has only the solution
1
|x| = a/(1 − a2 ) 2 . Thus if we combine (12.9) with the identification of
Sn with the unit sphere we get an identification
(12.10) Rn ∪ Sn−1 � Bn .
Using this identification we can, and will, regard Csp(u) and Css(u) as
subsets of Bn .21
Lemma 12.3. For any u ∈ S � (Rn ), Csp(u) and Css(u) are closed
subsets of Bn and if ψ˜ ∈ C ∞ (Sn ) has supp(ψ̃) ∩ Css(u) = ∅ then for R
sufficiently large ψ˜R u ∈ S(Rn ).
Proof. Directly from the definition we know that Csp(u) ∩ Rn is closed,
as is Css(u) ∩ Rn . Thus, in each case, we need to show that if ω ∈ Sn−1
and ω ∈/ Csp(u) then Csp(u) is disjoint from some neighbourhood of ω
n
in B . However, by definition,
� 0} ∪ {ω � ∈ Sn−1 ; ψ(ω � ) =
U = {x ∈ Rn ; ψR (x) = � 0}
is such a neighbourhood. Thus the fact that Csp(u) is closed follows
directly from the definition. The argument for Css(u) is essentially the
same.
The second result follows by the use of a partition of unity on Sn−1 .
Thus, for each point in supp(ψ) ⊂ Sn−1 there exists a conic localizer for
which ψR u ∈ S(Rn ). By compactness we may choose a finite number of
these functions ψj such that the open sets {ψj (ω) > 0} cover supp(ψ̃).
By assumption (ψj )Rj u ∈ S(Rn ) for some Rj > 0. However this will
remain true if Rj is increased, so we may suppose that Rj = R is
independent of j. Then for function
�
µ= |ψj |2 ∈ C ∞ (Sn−1 )
j

we have µR u ∈ S(Rn ). Since ψ˜ = ψ � µ for some µ ∈ C ∞ (Sn−1 ) it follows

that ψ˜R+1 u ∈ S(Rn ) as claimed. �
Corollary 12.4. If u ∈ S � (Rn ) then Css(u) = ∅ if and only if u ∈
S(Rn ).
Proof. Certainly Css(u) = ∅ if u ∈ S(Rn ). If u ∈ S � (Rn ) and Css(u) =
∅ then from Lemma 12.3, ψR u ∈ S(Rn ) where ψ = 1. Thus v =
(1 − ψR )u ∈ Cc−∞ (Rn ) has singsupp(v) = ∅ so v ∈ Cc∞ (Rn ) and hence
u ∈ S(Rn ). �
21In fact while the topology here is correct the smooth structure on Bn is not
the right one —– see Problem?? For our purposes here this issue is irrelevant.
86 RICHARD B. MELROSE

Of course the analogous result for Csp(u), that Csp(u) = ∅ if and

only if u = 0 follows from the fact that this is true if supp(u) = ∅. I
will treat a few other properties as self­evident. For instance
(12.11)
Csp(φu) ⊂ Csp(u), Css(φu) ⊂ Css(u) ∀ u ∈ S � (Rn ), φ ∈ S(Rn )
and

(12.12) Csp(c1 u1 + c2 u2 ) ⊂ Csp(u1 ) ∪ Csp(u2 ),

Css(c1 u1 + c2 u2 ) ⊂ Css(u1 ) ∪ Css(u2 )
∀ u1 , u2 ∈ S � (Rn ), c1 , c2 ∈ C.
One useful consequence of having the cone support at our disposal
is that we can discuss sufficient conditions to allow us to multiply dis­
tributions; we will get better conditions below using the same idea but
applied to the wavefront set but this preliminary discussion is used
there. In general the product of two distributions is not defined, and
indeed not definable, as a distribution. However, we can always multi­
ply an element of S � (Rn ) and an element of S(Rn ).
To try to understand multiplication look at the question of pairing
between two distributions.
Lemma 12.5. If Ki ⊂ Bn , i = 1, 2, are two disjoint closed (hence
compact) subsets then we can define an unambiguous pairing

(12.13)
{u ∈ S � (Rn ); Css(u) ⊂ K1 } × {u ∈ S � (Rn ); Css(u) ⊂ K2 } � (u1 , u2 )
−→ u1 (u2 ) ∈ C.

Proof. To define the pairing, choose a function ψ ∈ C ∞ (Sn−1 ) which is

identically equal to 1 in a neighbourhood of K1 ∩Sn−1 and with support
disjoint from K2 ∩ Sn−1 . Then extend it to be homogeneous, as above,
and cut off to get ψR . If R is large enough Csp(ψR ) is disjoint from K2 .
Then ψR + (1 − ψ)R = 1 + ν where ν ∈ Cc∞ (Rn ). We can find another
function µ ∈ Cc∞ (Rn ) such that ψ1 = ψR + µ = 1 in a neighbourhood of
K1 and with Csp(ψ1 ) disjoint from K2 . Once we have this, for u1 and
u2 as in (12.13),
(12.14) ψ1 u2 ∈ S(Rn ) and (1 − ψ1 )u1 ∈ S(Rn )
since in both cases Css is empty from the definition. Thus we can define
the desired pairing between u1 and u2 by
(12.15) u1 (u2 ) = u1 (ψ1 u2 ) + u2 ((1 − ψ1 )u1 ).
 LECTURE NOTES FOR 18.155, FALL 2004 87

Of course we should check that this definition is independent of the

cut­off function used in it. However, if we go through the definition and
choose a different function ψ � to start with, extend it homogeneoulsy
and cut off (probably at a different R) and then find a correction term
µ� then the 1­parameter linear homotopy between them
(12.16) ψ1 (t) = tψ1 + (1 − t)ψ1� , t ∈ [0, 1]
satisfies all the conditions required of ψ1 in formula (12.14). Thus in
fact we get a smooth family of pairings, which we can write for the
moment as
(12.17) (u1 , u2 )t = u1 (ψ1 (t)u2 ) + u2 ((1 − ψ1 (t))u1 ).
By inspection, this is an affine­linear function of t with derivative
(12.18) u1 ((ψ1 − ψ1� )u2 ) + u2 ((ψ1� − ψ1 ))u1 ).
Now, we just have to justify moving the smooth function in (12.18) to
see that this gives zero. This should be possible since Csp(ψ1� − ψ1 ) is
disjoint from both K1 and K2 .
In fact, to be very careful for once, we should construct another
function χ in the same way as we constructed ψ1 to be homogenous
near infinity and smooth and such that Csp(χ) is also disjoint from both
K1 and K2 but χ = 1 on Csp(ψ1� − ψ1 ). Then χ(ψ1� − ψ1 ) = ψ1� − ψ1 so
we can insert it in (12.18) and justify
(12.19) u1 ((ψ1 − ψ1� )u2 ) = u1 (χ2 (ψ1 − ψ1� )u2 ) = (χu1 )((ψ1 − ψ1� )χu2 )
= (χu2 )(ψ1 − ψ1� )χu1 ) = u2 (ψ1 − ψ1� )χu1 ).
Here the second equality is just the identity for χ as a (multiplica­
tive) linear map on S(Rn ) and hence S � (Rn ) and the operation to give
the crucial, third, equality is permissible because both elements are in
S(Rn ). �
Once we have defined the pairing between tempered distibutions with
disjoint conic singular supports, in the sense of (12.14), (12.15), we can
define the product under the same conditions. Namely to define the
product of say u1 and u2 we simply set
(12.20) u1 u2 (φ) = u1 (φu2 ) = u2 (φu1 ) ∀ φ ∈ S(Rn ),
provided Css(u1 ) ∩ Css(u2 ) = ∅.
Indeed, this would be true if one of u1 or u2 was itself in S(Rn ) and
makes sense in general. I leave it to you to check the continuity state­
ment required to prove that the product is actually a tempered disti­
bution (Problem 78).
88 RICHARD B. MELROSE

One can also give a similar discussion of the convolution of two tem­
pered distributions. Once again we do not have a definition of u∗v as a
tempered distribution for all u, v ∈ S � (Rn ). We do know how to define
the convolution if either u or v is compactly supported, or if either is
in S(Rn ). This leads directly to
Lemma 12.6. If Css(u) ∩ Sn−1 = ∅ then u ∗ v is defined unambiguously
by
x
(12.21) u ∗ v = u1 ∗ v + u2 ∗ v , u1 = (1 − χ( ))u, u2 = u − u1
r
∞ n
where χ ∈ Cc (R ) has χ(x) = 1 in |x| ≤ 1 and R is sufficiently large;
there is a similar definition if Css(v) ∩ Sn−1 = ∅.
Proof. Since Css(u) ∩ Sn−1 = ∅, we know that Css(u1 ) = ∅ if R is large
enough, so then both terms on the right in (12.21) are well­defined. To
see that the result is independent of R just observe that the difference
of the right­hand side for two values of R is of the form w ∗ v − w ∗ v
with w compactly supported. �
Now, we can go even further using a slightly more sophisticated
decomposition based on
Lemma 12.7. If u ∈ S � (Rn ) and Css(u) ∩ Γ = ∅ where Γ ⊂ Sn−1 is a
closed set, then u = u1 + u2 where Csp(u1 ) ∩ Γ = ∅ and u2 ∈ S(Rn ); in
fact
(12.22) u = u�1 + u��1 + u2 where u�1 ∈ Cc−∞ (Rn ) and
/ supp(u��1 ), x ∈ Rn \ {0}, x/|x| ∈ Γ =⇒ x ∈
0∈ / supp(u��1 ).
Proof. A covering argument which you should provide. �
Let Γi ⊂ Rn , i = 1, 2, be closed cones. That is they are closed sets
such that if x ∈ Γi and a > 0 then ax ∈ Γi . Suppose in addition that
(12.23) Γ1 ∩ (−Γ2 ) = {0}.
That is, if x ∈ Γ1 and −x ∈ Γ2 then x = 0. Then it follows that for
some c > 0,
(12.24) x ∈ Γ1 , y ∈ Γ2 =⇒ |x + y| ≥ c(|x| + |y |).
To see this consider x + y where x ∈ Γ1 , y ∈ Γ2 and |y | ≤ |x|. We
can assume that x �= 0, otherwise the estimate is trivially true with
c = 1, and then Y = y/|x| ∈ Γ1 and X = x/|x| ∈ Γ2 have |Y | ≤ 1 and
|X| = 1. However X + Y �= 0, since |X | = 1, so by the continuity of the
sum, |X + Y | ≥ 2c > 0 for some c > 0. Thus |X + Y | ≥ c(|X| + |Y |)
and the result follows by scaling back. The other case, of |x| ≤ |y|
 LECTURE NOTES FOR 18.155, FALL 2004 89

follows by the same argument with x and y interchanged, so (12.24) is

a consequence of (12.23).
Lemma 12.8. For any u ∈ S � (Rn ) and φ ∈ S(Rn ),
(12.25) Css(φ ∗ u) ⊂ Css(u) ∩ Sn−1 .
Proof. We already know that φ∗u is smooth, so Css(φ∗u) ⊂ Sn−1 . Thus,
we need to show that if ω ∈ Sn−1 and ω ∈ / Css(u) then ω ∈ / Css(φ ∗ u).
n−1
Fix such a point ω ∈ S \ Css(u) and take a closed set Γ ⊂ Sn−1
which is a neighbourhood of ω but which is still disjoint from Css(u) and
then apply Lemma 12.7. The two terms φ ∗ u2 , where u2 ∈ S(Rn ) and
φ ∗ u�1 where u1� ∈ Cc−∞ (Rn ) are both in S(Rn ) so we can assume that u
has the support properties of u��1 . In particular there is a smaller closed
subset Γ1 ⊂ Sn−1 which is still a neighbourhood of ω but which does
not meet Γ2 , which is the closure of the complement of Γ. If we replace
these Γi by the closed cones of which they are the ‘cross­sections’ then
we are in the situation of (12.23) and (12.23), except for the signs.
That is, there is a constant c > 0 such that
(12.26) |x − y | ≥ c(|x| + |y |).
Now, we can assume that there is a cutoff function ψR which has
support in Γ2 and is such that u = ψR u. For any conic cutoff, ψR� , with
support in Γ1
(12.27) ψR� (φ ∗ u) = �ψR u, φ(x − ·)� = �u(y), ψR (y)ψR� (x)φ(x − y)�.
The continuity of u means that this is estimated by some Schwartz
seminorm
(12.28) sup |Dyα (ψR (y)ψR� (x)φ(x − y))|(1 + |y |)k
y,|α|≤k

≤ CN �φ� sup(1 + |x| + |y |)−N (1 + |y |)k ≤ CN �φ�(1 + |x|)−N +k

y

for some Schwartz seminorm on φ. Here we have used the estimate

(12.24), in the form (12.26), using the properties of the supports of
ψR� and ψR . Since this is true for any N and similar estimates hold
for the derivatives, it follows that ψR� (u ∗ φ) ∈ S(Rn ) and hence that
ω∈ / Css(u ∗ φ). �
Corollary 12.9. Under the conditions of Lemma 12.6
(12.29) Css(u ∗ v) ⊂ (singsupp(u) + singsupp(v)) ∪ (Css(v) ∩ Sn−1 ).
Proof. We can apply Lemma 12.8 to the first term in (12.21) to con­
clude that it has conic singular support contained in the second term
in (12.29). Thus it is enough to show that (12.29) holds when u ∈
90 RICHARD B. MELROSE

Cc−∞ (Rn ). In that case we know that the singular support of the con­
volution is contained in the first term in (12.29), so it is enough to
consider the conic singular support in the sphere at infinity. Thus, if
ω ∈/ Css(v) we need to show that ω ∈ / Css(u ∗ v). Using Lemma 12.7
we can decompose v = v1 + v2 + v3 as a sum of a Schwartz term, a
compact supported term and a term which does not have ω in its conic
support. Then u ∗ v1 is Schwartz, u ∗ v2 has compact support and sat­
isfies (12.29) and ω is not in the cone support of u ∗ v3 . Thus (12.29)
holds in general. �
Lemma 12.10. If u, v ∈ S � (Rn ) and ω ∈ Css(u) ∩ Sn−1 =⇒ −ω ∈ /
Css(v) then their convolution is defined unambiguously, using the pair­
ing in Lemma 12.5, by
(12.30) v ∗ φ) ∀ φ ∈ S(Rn ).
u ∗ v(φ) = u(ˇ
Proof. Since v(x)
ˇ v) = − Css(v) so applying Lemma 12.8
= v(−x), Css(ˇ
we know that
(12.31) Css(v̌ ∗ φ) ⊂ − Css(v) ∩ Sn−1 .
Thus, Css(v) ∩ Css(v̌ ∗ φ) = ∅ and the pairing on the right in (12.30)
is well­defined by Lemma 12.5. Continuity follows from your work in
Problem 78. �
In Problem 79 I ask you to get a bound on Css(u ∗ v) ∩ Sn−1 under
the conditions in Lemma 12.10.
Let me do what is actually a fundamental computation.
Lemma 12.11. For a conic cutoff, ψR , where ψ ∈ C ∞ (Sn−1 ),
(12.32) �R ) ⊂ {0}.
Css(ψ
Proof. This is actually much easier than it seems. Namely we already
know that Dα (ψR ) is smooth and homogeneous of degree −|α| near
infinity. From the same argument it follows that
(12.33) Dα (xβ ψR ) ∈ L2 (Rn ) if |α| > |β | + n/2
since this is a smooth function homogeneous of degree less than −n/2
near infinity, hence square­integrable. Now, taking the Fourier trans­
form gives
(12.34) ξ α Dβ (ψ
�R ) ∈ L2 (Rn ) ∀ |α| > |β | + n/2.
If we localize in a cone near infinity, using a (completely unrelated)
cutoff ψR� � (ξ) then we must get a Schwartz function since
(12.35)
�R ) ∈ L2 (Rn ) ∀ |α| > |β | + n/2 =⇒ ψ � � (ξ)ψ
|ξ||α| ψR� � (ξ)Dβ (ψ R
�R ∈ S(Rn ).
 LECTURE NOTES FOR 18.155, FALL 2004 91

� 0 and so shows that

Indeed this argument applies anywhere that ξ =
(12.32) holds. �
Now, we have obtained some reasonable looking conditions under
which the product uv or the convolution u∗v of two elements of S � (Rn )
is defined. However, reasonable as they might be there is clearly a flaw,
or at least a deficiency, in the discussion. We know that in the simplest
of cases,
(12.36) ∗ v = u�
u� �v.
Thus, it is very natural to expect a relationship between the conditions
under which the product of the Fourier transforms is defined and the
conditions under which the convolution is defined. Is there? Well, not
much it would seem, since on the one hand we are considering the rela­
tionship between Css(� u) and Css(v�) and on the other the relationship
n−1
between Css(u) ∩ S and Css(v) ∩ Sn−1 . If these are to be related,
we would have to find a relationship of some sort between Css(u) and
Css(�u). As we shall see, there is one but it is not very strong as can
be guessed from Lemma 12.11. This is not so much a bad thing as a
sign that we should look for another notion which combines aspects of
both Css(u) and Css(� u). This we will do through the notion of wave­
front set. In fact we define two related objects. The first is the more
conventional, the second is more natural in our present discussion.
Definition 12.12. If u ∈ S � (Rn ) we define the wavefront set of u to
be
(12.37) WF(u) = {(x, ω) ∈ Rn × Sn−1 ;
∃ φ ∈ Cc∞ (Rn ), φ(x) =
� 0, ω ∈ � �
/ Css(φu)}
and more generally the scattering wavefront set by
(12.38) WFsc (u) = WF(u) ∪ {(ω, p) ∈ Sn−1 × Bn ;
∃ ψ ∈ C ∞ (Sn ), ψ(ω) =
� 0, R > 0 such that p ∈
/ Css(ψ� �
R u)} .

So, the definition is really always the same. To show that (p, q) ∈/
WFsc (u) we need to find ‘a cutoff Φ near p’ – depending on whether
p ∈ Rn or p ∈ Sn−1 this is either Φ = φ ∈ Cc∞ (Rn ) with F = φ(p) =
� 0
∞ n−1
or a ψR where ψ ∈ C (S ) has ψ(p) = � 0 – such that q ∈/ Css(Φu).
�
One crucial property is
Lemma 12.13. If (p, q) ∈ / WFsc (u) then if p ∈ Rn there exists a
neighbourhood U ⊂ R of p and a neighbourhood U ⊂ Bn of q such
n

that for all φ ∈ Cc∞ (Rn ) with support in U, U � ∩ Css(φu)

� = ∅; similarly
92 RICHARD B. MELROSE

if p ∈ Sn−1 then there exists a neigbourhood Ũ ⊂ Bn of p such that

U � ∩ Css(ψ� ˜
R u) = ∅ if Csp(ωR ) ⊂ U .

Proof. First suppose p ∈ Rn . From the definition of conic singular sup­

port, (12.37) means precisely that there exists ψ ∈ C ∞ (Sn−1 ), ψ(ω) =
� 0
and R such that
(12.39) � ∈ S(Rn ).
ψR (φu)
Since we know that φu � ∈ C ∞ (Rn ), this is actually true for all R > 0
as soon as it is true for one value. Furthermore, if φ� ∈ Cc∞ (Rn ) has
supp(φ� ) ⊂ {φ �= 0} then ω ∈ / Css(φ� � u) follows from ω ∈ / Css(φu).
�
� ∞ n
Indeed we can then write φ = µφ where µ ∈ Cc (R ) so it suffices
to show that if v ∈ Cc−∞ (Rn ) has ω ∈ / Css(v�) then ω ∈ / Css(� µv) if
µ ∈ Cc∞ (Rn ). Since µv
� = (2π)−n υ ∗ u � where υ̌ = µ � ∈ S(Rn ), applying
Lemma 12.8 we see that Css(υ ∗ v�) ⊂ Css(� v ), so indeed ω ∈ � u).
/ Css(φ�
n−1
The case that p ∈ S is similar. Namely we have one cut­off ψR
with ψ(p) =� 0 and q ∈/ Css(ω� R u). We can take U = {ψR+10 = � 0} since if
� �
ψR� has conic support in U then ψR� = ψ R ψR for some ψ ∈ C ∞ (Sn−1 ).
�� � ��

Thus
� ��
� u = v ∗ ψR u, v̌ = ω �� .
(12.40) �
ψ � �
R R

From Lemma 12.11 and Corollary12.9 we deduce that

�
R� u) ⊂ Css(ω
(12.41) �
Css(ψ �R u)

and hence the result follows with U � a small neighourhood of q. �

Proposition 12.14. For any u ∈ S � (Rn ),
(12.42) WFsc (u) ⊂ ∂(Bn × Bn ) = (Bn × Sn−1 ) ∪ (Sn−1 × Bn )
= (Rn × Sn−1 ) ∪ (Sn−1 × Sn−1 ) ∪ (Sn−1 × Rn )
and WF(u) ⊂ Rn are closed sets and under projection onto the first
variable
(12.43)
π1 (WF(u)) = singsupp(u) ⊂ Rn , π1 (WFsc (u)) = Css(u) ⊂ Bn .
Proof. To prove the first part of (12.43) we need to show that if (x̄, ω) ∈/
WF(u) for all ω ∈ Sn−1 with x¯ ∈ Rn fixed, then x¯ ∈ / singsupp(u). The
definition (12.37) means that for each ω ∈ Sn−1 there exists φω ∈
Cc∞ (Rn ) with φω (¯
x) =� 0 such that ω ∈ / Css(φ�ω u). Since Css(φu) is
n−1
closed and S is compact, a finite number of these cutoffs, φj ∈
Cc∞ (Rn ), can be chosen so that φj (¯ � 0 with the Sn−1 \ Css(φ
x) = �j u)
n−1
covering S . Now applying Lemma 12.13 above, we can find one
 LECTURE NOTES FOR 18.155, FALL 2004 93

φ ∈ Cc∞ (Rn ), with support in j {φj (x) =

�
� 0} and φ(x̄) =� 0, such that
� ⊂ Css(φ n
Css(φu) j u) for each j and hence φu ∈ S(R ) (since it is
�
already smooth). Thus indeed it follows that x¯ ∈ / singsupp(u). The
converse, that x¯ ∈
/ singsupp(u) implies (¯ / WF(u) for all ω ∈ Sn−1
x, ω) ∈
is immediate.
The argument to prove the second part of (12.43) is similar. Since, by
definition, WFsc (u)∩(Rn ×Bn ) = WF(u) and Css(u)∩Rn = singsupp(u)
we only need consider points in Css(u) ∩ Sn−1 . Now, we first check that
/ Css(u) then {θ} × Bn ∩ WFsc (u) = ∅. By definition of Css(u)
if θ ∈
there is a cut­off ψR , where ψ ∈ C ∞ (Sn−1 ) and ψ(θ) = � 0, such that
n
ψR u ∈ S(R ). From (12.38) this implies that (θ, p) ∈ / WFsc (u) for all
p ∈ Bn .
Now, Lemma 12.13 allows us to apply the same argument as used
above for WF . Namely we are given that (θ, p) ∈ / WFsc (u) for all
n
p ∈ B . Thus, for each p we may find ψR , depending on p, such that
n
ψ(θ) �= 0 and p ∈ / Css(ψ� R u). Since B is compact, we may choose a
(j)
finite subset of these conic localizers, ψRj such that the intersection
�(j)
of the corresponding sets Css(ψRj u), is empty, i.e. their complements
cover Bn . Now, using Lemma 12.13 we may choose one ψ with support
in the intersection of the sets {ψ (j) = � 0} with ψ(θ) = � 0 and one R
such that Css(ψR u) = ∅, but this just means that ψR u ∈ S(Rn ) and so
�
θ∈ / Css(u) as desired.
The fact that these sets are closed (in the appropriate sets) follows
directly from Lemma12.13. �
Corollary 12.15. For u ∈ S � (Rn ),
(12.44) WFsc (u) = ∅ ⇐⇒ u ∈ S(Rn ).
Let me return to the definition of WFsc (u) and rewrite it, using what
we have learned so far, in terms of a decomposition of u.
Proposition 12.16. For any u ∈ S � (Rn ) and (p, q) ∈ ∂(Bn × Bn ),
(12.45) (p, q) ∈
/ WFsc (u) ⇐⇒
u = u1 + u2 , u1 , u2 ∈ S � (Rn ), p ∈
/ Css(u1 ), q ∈
/ Css(u�2 ).
Proof. For given (p, q) ∈/ WFsc (u), take Φ = φ ∈ Cc∞ (Rn ) with φ ≡ 1
near p, if p ∈ Rn or Φ = ψR with ψ ∈ C ∞ (Sn−1 ) and ψ ≡ 1 near p, if
p ∈ Sn−1 . In either case p ∈
/ Css(u1 ) if u1 = (1 − Φ)u directly from the
definition. So u2 = u − u1 = Φu. If the support of Φ is small enough it
follows as in the discussion in the proof of Proposition 12.14 that
(12.46) q∈
/ Css(u�2 ).
94 RICHARD B. MELROSE

Thus we have (12.45) in the forward direction.

For reverse implication it follows directly that (p, q) ∈
/ WFsc (u1 ) and
that (p, q) ∈
/ WFsc (u2 ). �
This restatement of the definition makes it clear that there a high
degree of symmetry under the Fourier transform
Corollary 12.17. For any u ∈ S � (Rn ),
(12.47) (p, q) ∈ WFsc (u)) ⇐⇒ (q, −p) ∈ WFsc (û).
Proof. I suppose a corollary should not need a proof, but still . . . . The
statement (12.47) is equivalent to
(12.48) (p, q) ∈
/ WFsc (u)) =⇒ (q, −p) ∈
/ WFsc (ˆ
u)
since the reverse is the same by Fourier inversion. By (12.45) the
condition on the left is equivalent to u = u1 + u2 with p ∈
/ Css(u1 ),
q∈/ Css(u�2 ). Hence equivalent to
(12.49) � = v1 + v2 , v1 = u�2 , v�2 = (2π)−n uˇ1
u
so q ∈
/ Css(v1 ), −p ∈
/ Css(v�2 ) which proves (12.47). �
Now, we can exploit these notions to refine our conditions under
which pairing, the product and convolution can be defined.
Theorem 12.18. For u, v ∈ S � (Rn )
(12.50) uv ∈ S � (Rn ) is unambiguously defined provided
(p, ω) ∈ WFsc (u) ∩ (Bn × Sn−1 ) =⇒ (p, −ω) ∈
/ WFsc (v)
and
(12.51) u ∗ v ∈ S � (Rn ) is unambiguously defined provided
(θ, q) ∈ WFsc (u) ∩ (Sn−1 × Bn ) =⇒ (−θ, q) ∈
/ WFsc (v).
Proof. Let us consider convolution first. The hypothesis, (12.51) means
that for each θ ∈ Sn−1
(12.52)
{q ∈ Bn−1 ; (θ, q) ∈ WFsc (u)} ∩ {q ∈ Bn−1 ; (−θ, q) ∈ WFsc (v)} = ∅.
Now, the fact that WFsc is always a closed set means that (12.52)
remains true near θ in the sense that if U ⊂ Sn−1 is a sufficiently small
neighbourhood of θ then
(12.53) {q ∈ Bn−1 ; ∃ θ� ∈ U, (θ� , q) ∈ WFsc (u)}
∩ {q ∈ Bn−1 ; ∃ θ�� ∈ U, (−θ�� , q) ∈ WFsc (v)} = ∅.
 LECTURE NOTES FOR 18.155, FALL 2004 95

The compactness of Sn−1 means that there is a finite cover of Sn−1 by

such sets Uj . Now select a partition of unity ψi of Sn−1 which is not
only subordinate to this open cover, so each ψi is supported in one of
the Uj but satisfies the additional condition that
(12.54) supp(ψi ) ∩ (− supp(ψi� )) =
� ∅ =⇒
supp(ψi ) ∪ (− supp(ψi� )) ⊂ Uj for some j.
�
Now, if we set ui = (ψi )R u, and vi� = (ψi� )R v, we know that u − ui
i
has compact support and similarly for v. Since convolution is already
known to be possible if (at least) one factor has compact support, it
suffices to define ui ∗ vi� for every i, i� . So, first suppose that supp(ψi ) ∩
� ∅. In this case we conclude from (12.54) that
(− supp(ψi� )) =
(12.55) Css(u�i ) ∩ Css(v�i� ) = ∅.
Thus we may define
(12.56) i ∗ vi� = u
u� �i v�i�
using (12.20). On the other hand if supp ψi ∩ (− supp(ψi� )) = ∅ then
(12.57) Css(ui ) ∩ (− Css(vi� )) ∩ Sn−1 = ∅
and in this case we can define ui ∗ vi� using Lemma 12.10.
Thus with such a decomposition of u and v all terms in the convolu­
tion are well­defined. Of course we should check that this definition is
independent of choices made in the decomposition. I leave this to you.
That the product is well­defined under condition (12.50) now follows
if we define it using convolution, i.e. as
(12.58) � = f ∗ g, f = u,
uv � ǧ = v.
�
Indeed, using (12.47), (12.50) for u and v becomes (12.51) for f and
g. �
96 RICHARD B. MELROSE

13. Homogeneous distributions

Next time I will talk about homogeneous distributions. On R the
functions
� s
s x x>0
xt =
0 x<0
where S ∈ R, is locally integrable (and hence a tempered distribution)
precisely when S > −1. As a function it is homogeneous of degree s.
Thus if a > 0 then
(ax)st = as xst .
Thinking of xts = µs as a distribution we can set this as
�
µs (ax)(ϕ) = µs (ax)ϕ(x) dx
�
dx
= µs (x)ϕ(x/a)
a
s
= a µs (ϕ) .
Thus if we define ϕa (x) = a1 ϕ( xa ), for any a > 0, ϕ ∈ S(R) we can ask
whether a distribution is homogeneous:
µ(ϕa ) = as µ(ϕ) ∀ ϕ ∈ S(R).
 LECTURE NOTES FOR 18.155, FALL 2004 99

16. Spectral theorem

For a bounded operator T on a Hilbert space we define the spectrum
as the set
(16.1) spec(T ) = {z ∈ C; T − z Id is not invertible}.
Proposition 16.1. For any bounded linear operator on a Hilbert space
spec(T ) ⊂ C is a compact subset of {|z | ≤ �T �}.
Proof. We show that the set C \ spec(T ) (generally called the resolvent
set of T ) is open and contains the complement of a sufficiently large
ball. This is based on the convergence of the Neumann series. Namely
if T is bounded and �T � < 1 then
�∞
−1
(16.2) (Id −T ) = Tj
j=0

converges to a bounded operator which is a two­sided inverse of Id −T.

Indeed, �T j � ≤ �T �j so the series is convergent and composing with
Id −T on either side gives a telescoping series reducing to the identity.
Applying this result, we first see that
(16.3) (T − z) = −z(Id −T /z)
is invertible if |z | > �T �. Similarly, if (T − z0 )−1 exists for some z0 ∈ C
then
(16.4) (T −z) = (T −z0 )−(z−z0 ) = (T −z0 )−1 (Id −(z−z0 )(T −z0 )−1 )
exists for |z − z0 |�(T − z0 )−1 � < 1. �
In general it is rather difficult to precisely locate spec(T ).
However for a bounded self­adjoint operator it is easier. One sign of
this is the the norm of the operator has an alternative, simple, charac­
terization. Namely
(16.5) if A∗ = A then sup �Aφ, φ�| = �A�.
�φ�=1

If a is this supermum, then clearly a ≤ �A�. To see the converse, choose

any φ, ψ ∈ H with norm 1 and then replace ψ by eiθ ψ with θ chosen
so that �Aφ, ψ� is real. Then use the polarization identity to write
(16.6) 4�Aφ, ψ� = �A(φ + ψ), (φ + ψ)� − �A(φ − ψ), (φ − ψ)�
+ i�A(φ + iψ), (φ + iψ)� − i�A(φ − iψ), (φ − iψ)�.
Now, by the assumed reality we may drop the last two terms and see
that
(16.7) 4|�Aφ, ψ�| ≤ a(�φ + ψ�2 + �φ − ψ�2 ) = 2a(�φ�2 + �ψ�2 ) = 4a.
100 RICHARD B. MELROSE

Thus indeed �A� = sup�φ�=�ψ�=1 |�Aφ, ψ�| = a.

We can always subtract a real constant from A so that A� = A − t
satisfies
(16.8) − inf �A� φ, φ� = sup �A� φ, φ� = �A� �.
�φ�=1 �φ�=1

Then, it follows that A� ± �A� � is not invertible. Indeed, there exists a

sequence φn , with �φn � = 1 such that �(A� − �A� �)φn , φn � → 0. Thus
(16.9)
�(A� −�A� �)φn �2 = −2�A� φn , φn �+�A� φn �2 +�A� �2 ≤ −2�A� φn , φn �+2�A� �2 → 0.
This shows that A� − �A� � cannot be invertible and the same argument
works for A� + �A� �. For the original operator A if we set
(16.10) m = inf �Aφ, φ� M = sup �Aφ, φ�
�φ�=1 �φ�=1

then we conclude that neither A − m Id nor A − M Id is invertible and

�A� = max(−m, M ).
Proposition 16.2. If A is a bounded self­adjoint operator then, with
m and M defined by (16.10),
(16.11) {m} ∪ {M } ⊂ spec(A) ⊂ [m, M ].
Proof. We have already shown the first part, that m and M are in
the spectrum so it remains to show that A − z is invertible for all
z ∈ C \ [m, M ].
Using the self­adjointness
(16.12) Im�(A − z)φ, φ� = − Im z�φ�2 .
This implies that A − z is invertible if z ∈ C \ R. First it shows that
(A − z)φ = 0 implies φ = 0, so A − z is injective. Secondly, the range is
closed. Indeed, if (A − z)φn → ψ then applying (16.12) directly shows
that �φn � is bounded and so can be replaced by a weakly convergent
subsequence. Applying (16.12) again to φn − φm shows that the se­
quence is actually Cauchy, hence convergens to φ so (A − z)φ = ψ is in
the range. Finally, the orthocomplement to this range is the null space
of A∗ − z̄, which is also trivial, so A − z is an isomorphism and (16.12)
also shows that the inverse is bounded, in fact
1
(16.13) �(A − z)−1 � ≤ .
| Im z|
When z ∈ R we can replace A by A� satisfying (16.8). Then we have
to show that A� − z is inverible for |z | > �A�, but that is shown in the
proof of Proposition 16.1. �
 LECTURE NOTES FOR 18.155, FALL 2004 101

The basic estimate leading to the spectral theorem is:

Proposition 16.3. If A is a bounded self­adjoint operator and p is a
real polynomial in one variable,
N
�
(16.14) p(t) = ci ti , cN �= 0,
i=0
N
ci Ai satisfies
�
then p(A) =
i=0

(16.15) �p(A)� ≤ sup |p(t)|.

t∈[m,M ]

Proof. Clearly, p(A) is a bounded self­adjoint operator. If s ∈/ p([m, M ])

then p(A) − s is invertible. Indeed, the roots of p(t) − s must cannot
lie in [m.M ], since otherwise s ∈ p([m, M ]). Thus, factorizing p(s) − t
we have
(16.16)
�N
p(t) − s = cN / [m, M ] =⇒ (p(A) − s)−1 exists
(t − ti (s)), ti (s) ∈
i=1
�
since p(A) = cN (A − ti (s)) and each of the factors is invertible.
i
Thus spec(p(A)) ⊂ p([m, M ]), which is an interval (or a point), and
from Proposition 16.3 we conclude that �p(A)� ≤ sup p([m, M ]) which
is (16.15). �
Now, reinterpreting (16.15) we have a linear map
(16.17) P(R) � p �−→ p(A) ∈ B(H)
from the real polynomials to the bounded self­adjoint operators which
is continuous with respect to the supremum norm on [m, M ]. Since
polynomials are dense in continuous functions on finite intervals, we
see that (16.17) extends by continuity to a linear map
(16.18)
C([m, M ]) � f �−→ f (A) ∈ B(H), �f (A)� ≤ �f �[m,M ] , f g(A) = f (A)g(A)
where the multiplicativity follows by continuity together with the fact
that it is true for polynomials.
Now, consider any two elements φ, ψ ∈ H. Evaluating f (A) on φ and
pairing with ψ gives a linear map
(16.19) C([m, M ]) � f �−→ �f (A)φ, ψ� ∈ C.
This is a linear functional on C([m, M ]) to which we can apply the Riesz
representatin theorem and conclude that it is defined by integration
102 RICHARD B. MELROSE

against a unique Radon measure µφ,ψ :

�
(16.20) �f (A)φ, ψ� = f dµφ,ψ .
[m,M ]

The total mass |µφ,ψ | of this measure is the norm of the functional.
Since it is a Borel measure, we can take the integral on −∞, b] for any
b ∈ R ad, with the uniqueness, this shows that we have a continuous
sesquilinear map
(16.21) �
Pb (φ, ψ) : H×H � (φ, ψ) �−→ dµφ,ψ ∈ R, |Pb (φ, ψ)| ≤ �A��φ��ψ�.
[m,b]

From the Hilbert space Riesz representation theorem it follows that

this sesquilinear form defines, and is determined by, a bounded linear
operator
(16.22) Pb (φ, ψ) = �Pb φ, ψ�, �Pb � ≤ �A�.
In fact, from the functional calculus (the multiplicativity in (16.18))
we see that
(16.23) Pb∗ = Pb , Pb2 = Pb , �Pb � ≤ 1,
so Pb is a projection.
Thus the spectral theorem gives us an increasing (with b) family of
commuting self­adjoint projections such that µφ,ψ ((−∞, b]) = �Pb φ, ψ�
determines the Radon measure for which (16.20) holds. One can go
further and think of Pb itself as determining a measure
(16.24) µ((−∞, b]) = Pb
which takes values in the projections on H and which allows the func­
tions of A to be written as integrals in the form
�
(16.25) f (A) = f dµ
[m,M ]

of which (16.20) becomes the ‘weak form’. To do so one needs to

develop the theory of such measures and the corresponding integrals.
This is not so hard but I shall not do it.
 LECTURE NOTES FOR 18.155, FALL 2004 103

17. Problems
Problem 1. Prove that u+ , defined by (1.10) is linear.
Problem 2. Prove Lemma 1.8.
Hint(s). All functions here are supposed to be continuous, I just
don’t bother to keep on saying it.
(1) Recall, or check, that the local compactness of a metric space
X means that for each point x ∈ X there is an � > 0 such that
the ball {y ∈ X; d(x, y) ≤ δ} is compact for δ ≤ �.
(2) First do the case n = 1, so K � U is a compact set in an open
subset.
(a) Given δ > 0, use the local compactness of X, to cover K
with a finite number of compact closed balls of radius at
most δ.
(b) Deduce that if � > 0 is small enough then the set {x ∈
X; d(x, K) ≤ �}, where
d(x, K) = inf d(x, y),
y∈K

is compact.
(c) Show that d(x, K), for K compact, is continuous.
(d) Given � > 0 show that there is a continuous function g� :
R −→ [0, 1] such that g� (t) = 1 for t ≤ �/2 and g� (t) = 0
for t > 3�/4.
(e) Show that f = g� ◦ d(·, K) satisfies the conditions for n = 1
if � > 0 is small enough.
(3) Prove the general case by induction over n.
(a) In the general case, set K � = K ∩ U1� and show that the
inductive hypothesis applies to K � and the Uj for j > 1; let
fj� , j = 2, . . . , n be the functions supplied by the inductive
assumption and put f = j≥2 fj� .
�
�

(b) Show that K1 = K ∩ {f � ≤ 21 } is a compact subset of U1 .

(c) Using the case n = 1 construct a function F for K1 and
U1 .
(d) Use the case n = 1 again to find G such that G = 1 on K
and supp(G) � {f � + F > 21 }.
(e) Make sense of the functions
G G
f1 = F , fj = fj� � , j≥2
f� +F f +F
and show that they satisfies the inductive assumptions.
104 RICHARD B. MELROSE

Problem 3. Show that σ­algebras are closed under countable intersec­

tions.
Problem 4. (Easy) Show that if µ is a complete measure and E ⊂ F
where F is measurable and has measure 0 then µ(E) = 0.
Problem 5. Show that compact subsets are measurable for any Borel
measure. (This just means that compact sets are Borel sets if you
follow through the tortuous terminology.)
Problem 6. Show that the smallest σ­algebra containing the sets
(a, ∞] ⊂ [−∞, ∞]
for all a ∈ R, generates what is called above the ‘Borel’ σ­algebra on
[−∞, ∞].
Problem 7. Write down a careful proof of Proposition 1.1.
Problem 8. Write down a careful proof of Proposition 1.2.
Problem 9. Let X be the metric space
X = {0} ∪ {1/n; n ∈ N = {1, 2, . . .}} ⊂ R
with the induced metric (i.e. the same distance as on R). Recall why
X is compact. Show that the space C0 (X) and its dual are infinite
dimensional. Try to describe the dual space in terms of sequences; at
least guess the answer.
Problem 10. For the space Y = N = {1, 2, . . .} ⊂ R, describe C0 (Y )
and guess a description of its dual in terms of sequences.
Problem 11. Let (X, M, µ) be any measure space (so µ is a measure
on the σ­algebra M of subsets of X). Show that the set of equivalence
classes of µ­integrable functions on X, with the equivalence relation
given by (4.8), is a normed linear space with the usual linear structure
and the norm given by
�
�f � = |f |dµ.
X

Problem 12. Let (X, M) be a set with a σ­algebra. Let µ : M → R be

a finite measure in the sense that µ(φ) = 0 and for any {Ei }∞
i=1 ⊂ M
with Ei ∩ Ej = φ for i =
� j,
�∞ � ∞

�
(17.1) µ Ei = µ(Ei )
i=1 i=1
 LECTURE NOTES FOR 18.155, FALL 2004 105

with the series on the right always absolutely convergenct (i.e., this is
part of the requirement on µ). Define
∞
�
(17.2) |µ| (E) = sup |µ(Ei )|
i=1

for E ∈ M, with the supremum over all measurable decompositions

E = ∞ i=1 Ei with the Ei disjoint. Show that |µ| is a finite, positive
measure.
Hint 1. You must show that |µ| (E) = ∞
� 
i=1|µ| (Ai ) if i Ai = E,
Ai ∈ M being disjoint. Observe that if Aj = l Ajl is a measurable
decomposition of Aj then together the Ajl give a decomposition of E.
Similarly, if E = j Ej is any such decomposition of E then Ajl =
Aj ∩ El gives such a decomposition of Aj .
Hint 2. See [5] p. 117!
Problem 13. (Hahn Decomposition)
With assumptions as in Problem 12:
(1) Show that µ+ = 21 (|µ| + µ) and µ− = 12 (|µ| − µ) are positive
measures, µ = µ+ − µ− . Conclude that the definition of a
measure based on (4.16) is the same as that in Problem 12.
(2) Show that µ± so constructed are orthogonal in the sense that
there is a set E ∈ M such that µ− (E) = 0, µ+ (X \ E) = 0.
Hint. Use the definition of |µ| to show that for any F ∈ M
and any � > 0 there is a subset F � ∈ M, F � ⊂ F such that
µ+ (F � ) ≥ µ+ (F ) − � and µ− (F � ) ≤ �. Given δ > 0 apply
this result repeatedly (say with � = 2−n δ) to find a decreas­
ing sequence of sets F1 = X, Fn ∈ M, Fn+1 ⊂ Fn such that
µ+ (Fn�) ≥ µ+ (Fn−1 ) − 2−n δ and µ− (Fn ) ≤ 2−n δ. Conclude that
G = n Fn has µ+ (G) ≥ µ+ (X) − δ and µ− (G) = 0. Now  let
Gm be chosen this way with δ = 1/m. Show that E = m Gm
is as required.
Problem 14. Now suppose that µ is a finite, positive Radon measure
on a locally compact metric space X (meaning a finite positive Borel
measure outer regular on Borel sets and inner regular on open sets).
Show that µ is inner regular on all Borel sets and hence, given � > 0
and E ∈ B(X) there exist sets K ⊂ E ⊂ U with K compact and U
open such that µ(K) ≥ µ(E) − �, µ(E) ≥ µ(U ) − �.
Hint. First take U open, then use its inner regularity to find K
with K � � U and µ(K � ) ≥ µ(U ) − �/2. How big is µ(E\K � )? Find
V ⊃ K � \E with V open and look at K = K � \V .
106 RICHARD B. MELROSE

Problem 15. Using Problem 14 show that if µ is a finite Borel measure

on a locally compact metric space X then the following three conditions
are equivalent
(1) µ = µ1 −µ2 with µ1 and µ2 both positive finite Radon measures.
(2) |µ| is a finite positive Radon measure.
(3) µ+ and µ− are finite positive Radon measures.

Problem 16. Let � � be a norm on a vector space V . Show that �u� =

(u, u)1/2 for an inner product satisfying (5.1) ­ (5.4) if and only if the
parallelogram law holds for every pair u, v ∈ V .
Hint (From Dimitri Kountourogiannis)
If � · � comes from an inner product, then it must satisfy the polari­
sation identity:

(x, y) = 1/4(�x + y�2 − �x − y�2 − i�x + iy�2 − i�x − iy�2 )

i.e, the inner product is recoverable from the norm, so use the RHS
(right hand side) to define an inner product on the vector space. You
will need the paralellogram law to verify the additivity of the RHS.
Note the polarization identity is a bit more transparent for real vector
spaces. There we have

(x, y) = 1/2(�x + y�2 − �x − y�2 )

both are easy to prove using �a�2 = (a, a).

Problem 17. Show (Rudin does it) that if u : Rn → C has continuous

partial derivatives then it is differentiable at each point in the sense of
(6.5).

Problem 18. Consider the function f (x) = �x�−1 = (1+ |x|2 )−1/2 . Show
that
∂f
= lj (x) · �x�−3
∂xj
with lj (x) a linear function. Conclude by induction that �x�−1 ∈
C0k (Rn ) for all k.

Problem 19. Show that exp(− |x|2 ) ∈ S(Rn ).

Problem 20. Prove (7.7), probably by induction over k.

Problem 21. Prove Lemma 7.4.

 LECTURE NOTES FOR 18.155, FALL 2004 107

Hint. Show that a set U � 0 in S(Rn ) is a neighbourhood of 0 if and

only if for some k and � > 0 it contains a set of the form
⎧ ⎫
⎪
⎪ ⎪
⎪
⎨ � � � ⎬
n α β
ϕ ∈ S(R ) ; sup �x D ϕ� < � .
⎪ ⎪
⎪
⎩ |α|≤k, ⎪
⎭
|β|≤k

Problem 22. Prove (8.7), by estimating the integrals.

Problem 23. Prove (8.9) where
� �
� ∂ψ
ψj (z; x ) = (z + tx� ) dt .
0 ∂zj
Problem 24. Prove (8.20). You will probably have to go back to first
principles to do this. Show that it is enough to assume u ≥ 0 has
compact support. Then show it is enough to assume that u is a simple,
and integrable, function. Finally look at the definition of Lebesgue
measure and show that if E ⊂ Rn is Borel and has finite Lebesgue
measure then
lim µ(E\(E + t)) = 0
|t|→∞

where µ = Lebesgue measure and

E + t = {p ∈ Rn ; p� + t , p� ∈ E} .
Problem 25. Prove Leibniz’ formula
� �α �
α
D x (ϕψ) = Dα x ϕ · dα−β
x ψ
β≤α
β

for any C ∞ functions and ϕ and ψ. Here α and β are multiindices,

β ≤ α means βj ≤ αj for each j? and
� � �� �
α αj
= .
β j
βj

I suggest induction!
Problem 26. Prove the generalization of Proposition 8.10 that u ∈
S � (Rn ), supp(w) ⊂ {0} implies there are constants cα , |α| ≤ m, for
some m, such that
�
u= cα D α δ .
|α|≤m
108 RICHARD B. MELROSE

Hint This is not so easy! I would be happy if you can show that
u ∈ M (Rn ), supp u ⊂ {0} implies u = cδ. To see this, you can show
that
ϕ ∈ S(Rn ), ϕ(0) = 0
⇒ ∃ϕj ∈ S(Rn ) , ϕj (x) = 0 in |x| ≤ �j > 0(↓ 0) ,
sup |ϕj − ϕ| → 0 as j → ∞ .
To prove the general case you need something similar — that given m,
if ϕ ∈ S(Rn ) and Dα x ϕ(0) = 0 for |α| ≤ m then ∃ ϕj ∈ S(Rn ), ϕj = 0
in |x| ≤ �j , �j ↓ 0 such that ϕj → ϕ in the C m norm.
Problem 27. If m ∈ N, m� > 0 show that u ∈ H m (Rn ) and Dα u ∈
� �
H m (Rn ) for all |α| ≤ m implies u ∈ H m+m (Rn ). Is the converse true?
Problem 28. Show that every element u ∈ L2 (Rn ) can be written as a
sum n
�
u = u0 + Dj uj , uj ∈ H 1 (Rn ) , j = 0, . . . , n .
j=1

Problem 29. Consider for n = 1, the locally integrable function (the

Heaviside function),
�
0 x≤0
H(x) =
1 x > 1.
Show that Dx H(x) = cδ; what is the constant c?
Problem 30. For what range of orders m is it true that δ ∈ H m (Rn ) , δ(ϕ) =
ϕ(0)?
Problem 31. Try to write the Dirac measure explicitly (as possible) in
the form (10.8). How many derivatives do you think are necessary?
Problem 32. Go through the computation of ∂E again, but cutting out
a disk {x2 + y 2 ≤ �2 } instead.
Problem 33. Consider the Laplacian, (11.4), for n = 3. Show that
E = c(x2 + y 2 )−1/2 is a fundamental solution for some value of c.
Problem 34. Recall that a topology on a set X is a collection F of
subsets (called the open sets) with the properties, φ ∈ F, X ∈ F and
F is closed under finite intersections and arbitrary unions. Show that
the following definition of an open set U ⊂ S � (Rn ) defines a topology:
∀ u ∈ U and all ϕ ∈ S(Rn ) ∃� > 0 st.
|(u� − u)(ϕ)| < � ⇒ u� ∈ U .
 LECTURE NOTES FOR 18.155, FALL 2004 109

This is called the weak topology (because there are very few open
sets). Show that uj → u weakly in S � (Rn ) means that for every open
set U � u ∃N st. uj ∈ U ∀ j ≥ N .
Problem 35. Prove (11.18) where u ∈ S � (Rn ) and ϕ, ψ ∈ S(Rn ).
Problem 36. Show that for fixed v ∈ S � (Rn ) with compact support
S(Rn ) � ϕ �→ v ∗ ϕ ∈ S(Rn )
is a continuous linear map.
Problem 37. Prove the ?? to properties in Theorem 11.6 for u ∗ v where
u ∈ S � (Rn ) and v ∈ S � (Rn ) with at least one of them having compact
support.
Problem 38. Use Theorem 11.9 to show that if P (D) is hypoelliptic
then every parametrix F ∈ S(Rn ) has sing supp(F ) = {0}.
Problem 39. Show that if P (D) is an ellipitic differential operator of
order m, u ∈ L2 (Rn ) and P (D)u ∈ L2 (Rn ) then u ∈ H m (Rn ).
Problem 40 (Taylor’s theorem). . Let u : Rn −→ R be a real­valued
function which is k times continuously differentiable. Prove that there
is a polynomial p and a continuous function v such that
|v(x)|
u(x) = p(x) + v(x) where lim = 0.
|x|↓0 |x|k

Problem 41. Let C(Bn ) be the space of continuous functions on the

(closed) unit ball, Bn = {x ∈ Rn ; |x| ≤ 1}. Let C0 (Bn ) ⊂ C(Bn ) be
the subspace of functions which vanish at each point of the boundary
and let C(Sn−1 ) be the space of continuous functions on the unit sphere.
Show that inclusion and restriction to the boundary gives a short exact
sequence
C0 (Bn ) �→ C(Bn ) −→ C(Sn−1 )
(meaning the first map is injective, the second is surjective and the
image of the first is the null space of the second.)
Problem 42 (Measures). A measure on the ball is a continuous linear
functional µ : C(Bn ) −→ R where continuity is with respect to the
supremum norm, i.e. there must be a constant C such that
|µ(f )| ≤ C sup |f (x)| ∀ f ∈ C(Bn ).
x∈Rn

Let M (B ) be the linear space of such measures. The space M (Sn−1 )

n

of measures on the sphere is defined similarly. Describe an injective

map
M (Sn−1 ) −→ M (Bn ).
110 RICHARD B. MELROSE

Can you define another space so that this can be extended to a short
exact sequence?
Problem 43. Show that the Riemann integral defines a measure
�
n
(17.3) C(B ) � f �−→ f (x)dx.
Bn

Problem 44. If g ∈ C(Bn ) and µ ∈ M (Bn ) show that gµ ∈ M (Bn )

where (gµ)(f ) = µ(f g) for all f ∈ C(Bn ). Describe all the measures
with the property that
xj µ = 0 in M (Bn ) for j = 1, . . . , n.
Problem 45 (Hörmander, Theorem 3.1.4). Let I ⊂ R be an open, non­
empty interval.
i) Show (you may use results from class) that there exists ψ ∈
Cc∞ (I) with R ψ(x)ds = 1.
�
ii) Show that any φ ∈ Cc∞ (I) may be written in the form
�
˜ ˜ ∞
φ = φ + cψ, c ∈ C, φ ∈ Cc (I) with φ̃ = 0.
R

Show that if φ̃ ∈ Cc∞ (I) and R φ̃ = 0 then

�
iii) there exists µ ∈
Cc∞ (I) such that dµ
dx
= φ̃ in I.
−∞
iv) Suppose u ∈ C (I) satisfies du dx
= 0, i.e.
dφ
u(−) = 0 ∀ φ ∈ Cc∞ (I),
dx
show that u = c for some constant c.
v) Suppose that u ∈ C −∞ (I) satisfies du
dx
= c, for some constant c,
show that u = cx + d for some d ∈ C.
Problem 46. [Hörmander Theorem 3.1.16]
i) Use Taylor’s formula to show that there is a fixed ψ ∈ Cc∞ (Rn )
such that any φ ∈ Cc∞ (Rn ) can be written in the form
�n
φ = cψ + xj ψj
j=1

where c ∈ C and the ψj ∈ Cc∞ (Rn ) depend on φ.

ii) Recall that δ0 is the distribution defined by
δ0 (φ) = φ(0) ∀ φ ∈ Cc∞ (Rn );
explain why δ0 ∈ C −∞ (Rn ).
iii) Show that if u ∈ C −∞ (Rn ) and u(xj φ) = 0 for all φ ∈ Cc∞ (Rn )
and j = 1, . . . , n then u = cδ0 for some c ∈ C.
 LECTURE NOTES FOR 18.155, FALL 2004 111

iv) Define the ‘Heaviside function’

� ∞
H(φ) = φ(x)dx ∀ φ ∈ Cc∞ (R);
0
−∞
show that H ∈ C (R).
v) Compute dxd
H ∈ C −∞ (R).
Problem 47. Using Problems 45 and 46, find all u ∈ C −∞ (R) satisfying
the differential equation
du
x = 0 in R.
dx
These three problems are all about homogeneous distributions on
the line, extending various things using the fact that
�
exp(z log x) x > 0
xz+ =
0 x≤0
is a continuous function on R if Re z > 0 and is differentiable if Re z > 1
and then satisfies
d z
x+ = zxz−1
+ .
dx
We used this to define
1 1 1 dk z+k
(17.4) xz+ = ··· x if z ∈ C \ −N.
z+kz+k−1 z + 1 dxk +
Problem 48. [Hadamard regularization]
i) Show that (17.4) just means that for each φ ∈ Cc∞ (R)
� ∞ k
z (−1)k d φ
x+ (φ) = (x)xz+k dx, Re z > −k, z ∈
/ −N.
(z + k) · · · (z + 1) 0 dxk
ii) Use integration by parts to show that
(17.5) �� �
∞ �k
xz+ (φ) = lim φ(x)xz dx − Cj (φ)�z+j , Re z > −k, z ∈
/ −N
�↓0 � j=1

for certain constants Cj (φ) which you should give explicitly.

[This is called Hadamard regularization after Jacques Hadamard,
feel free to look at his classic book [3].]
iii) Assuming that −k + 1 ≥ Re z > −k, z �= −k + 1, show that
there can only be one set of the constants with j < k (for each
choice of φ ∈ Cc∞ (R)) such that the limit in (17.5) exists.
112 RICHARD B. MELROSE

iiv) Use ii), and maybe iii), to show that

d z z−1
x+ = zx+ in C −∞ (R) ∀ z ∈
/ −N0 = {0, 1, . . . }.
dx
z+1
v) Similarly show that xxz+ = x+ for all z ∈
/ −N.
z
vi) Show that x+ = 0 in x < 0 for all z ∈
/ −N. (Duh.)
d
Problem 49. [Null space of x dx − z]
i) Show that if u ∈ C −∞ (R) then ũ(φ) = u(φ̃), where φ̃(x) =
φ(−x) ∀ φ ∈ Cc∞ (R), defines an element of C −∞ (R). What is ũ
if u ∈ C 0 (R)? Compute δ�0 .
ii) Show that d ũ = − �
dx
d
dx
u.
iii) Define xz− = x� z
+ for z ∈/ −N and show that dx x− = −zxz−1
d z
− and
z z+1
xx− = −x− .
iv) Suppose that u ∈ C −∞ (R) satisfies the distributional equation
d
(x dx − z)u = 0 (meaning of course, x du dx
= zu where z is a
constant). Show that
� � � �
u�x>0 = c+ xz− �x>0 and u�x<0 = c− xz− �x<0
z
for some constants c± . Deduce that v = u− c+ x+ − c− xz− satisfies
d
(17.6) (x − z)v = 0 and supp(v) ⊂ {0}.
dx
d d k
v) Show that for each k ∈ N, (x dx + k + 1) dx k δ0 = 0.
−∞
vi) Using the fact that any v ∈ C (R) with supp(v) ⊂ {0} is
dk
a finite sum of constant multiples of the dx k δ0 , show that, for

z∈/ −N, the only solution of (17.6) is v = 0.

vii) Conclude that for z ∈
/ −N
� �
−∞ d
(17.7) u ∈ C (R); (x − z)u = 0
dx
is a two­dimensional vector space.
Problem 50. [Negative integral order] To do the same thing for negative
integral order we need to work a little differently. Fix k ∈ N.
i) We define weak convergence of distributions by saying un → u in
Cc∞ (X), where un , u ∈ C −∞ (X), X ⊂ Rn being open, if un (φ) →
u(φ) for each φ ∈ Cc∞ (X). Show that un → u implies that
∂un
∂xj
∂u
→ ∂x j
for each j = 1, . . . , n and f un → f u if f ∈ C ∞ (X).
 LECTURE NOTES FOR 18.155, FALL 2004 113

ii) Show that (z + k)xz+ is weakly continuous as z → −k in the

sense that for any sequence zn → −k, zn ∈ / −N, (zn +k)xz+n → vk
where
1 1 dk+1
vk = ··· x+ , x+ = x1+ .
−1 −k + 1 dxk+1
iii) Compute vk , including the constant factor.
iv) Do the same thing for (z + k)xz− as z → −k.
v) Show that there is a linear combination (k + z)(xz+ + c(k)xz− )
such that as z → −k the limit is zero.
vi) If you get this far, show that in fact xz+ + c(k)x− z
also has a
weak limit, uk , as z → −k. [This may be the hardest part.]
d
vii) Show that this limit distribution satisfies (x dx + k)uk = 0.
viii) Conclude that (17.7) does in fact hold for z ∈ −N as well.
[There are still some things to prove to get this.]
Problem 51. Show that for any set G ⊂ Rn
∞
�
∗
v (G) = inf v(Ai )
i=1

where the infimum is taken over coverings of G by rectangular sets

(products of intervals).
Problem 52. Show that a σ­algebra is closed under countable intersec­
tions.
Problem 53. Show that compact sets are Lebesgue measurable and
have finite volume and also show the inner regularity of the Lebesgue
measure on open sets, that is if E is open then
(17.8) v(E) = sup{v(K); K ⊂ E, K compact}.
Problem 54. Show that a set B ⊂ Rn is Lebesgue measurable if and
only if
v ∗ (E) = v ∗ (E ∩ B) + v ∗ (E ∩ B � ) ∀ open E ⊂ Rn .
[The definition is this for all E ⊂ Rn .]
Problem 55. Show that a real­valued continuous function f : U −→ R
on an open set, is Lebesgue measurable, in the sense that f −1 (I) ⊂
U ⊂ Rn is measurable for each interval I.
Problem 56. Hilbert space and the Riesz representation theorem. If
you need help with this, it can be found in lots of places – for instance
[6] has a nice treatment.
114 RICHARD B. MELROSE

i) A pre­Hilbert space is a vector space V (over C) with a ‘positive

definite sesquilinear inner product’ i.e. a function
V × V � (v, w) �→ �v, w� ∈ C
satisfying
• �w, v� = �v, w�
• �a1 v1 + a2 v2 , w� = a1 �v1 , w� + a2 �v2 , w�
• �v, v� ≥ 0
• �v, v� = 0 ⇒ v = 0.
Prove Schwarz’ inequality, that
1 1
|�u, v�| ≤ �u� 2 �v� 2 ∀ u, v ∈ V.
Hint: Reduce to the case �v, v� = 1 and then expand
�u − �u, v�v, u − �u, v�v� ≥ 0.
ii) Show that �v� = �v, v�1/2 is a norm and that it satisfies the
parallelogram law:
(17.9) �v1 + v2 �2 + �v1 − v2 �2 = 2�v1 �2 + 2�v2 �2 ∀ v1 , v2 ∈ V.
iii) Conversely, suppose that V is a linear space over C with a norm
which satisfies (17.9). Show that
4�v, w� = �v + w�2 − �v − w�2 + i�v + iw�2 − i�v − iw�2
defines a pre­Hilbert inner product which gives the original
norm.
iv) Let V be a Hilbert space, so as in (i) but complete as well.
Let C ⊂ V be a closed non­empty convex subset, meaning
v, w ∈ C ⇒ (v + w)/2 ∈ C. Show that there exists a unique
v ∈ C minimizing the norm, i.e. such that
�v� = inf �w�.
w∈C

Hint: Use the parallelogram law to show that a norm mini­

mizing sequence is Cauchy.
v) Let u : H → C be a continuous linear functional on a Hilbert
space, so |u(ϕ)| ≤ C�ϕ� ∀ ϕ ∈ H. Show that N = {ϕ ∈
H; u(ϕ) = 0} is closed and that if v0 ∈ H has u(v0 ) �= 0 then
each v ∈ H can be written uniquely in the form
v = cv0 + w, c ∈ C, w ∈ N.
vi) With u as in v), not the zero functional, show that there exists
a unique f ∈ H with u(f ) = 1 and �w, f � = 0 for all w ∈ N .
Hint: Apply iv) to C = {g ∈ V ; u(g) = 1}.
 LECTURE NOTES FOR 18.155, FALL 2004 115

vii) Prove the Riesz Representation theorem, that every continuous

linear functional on a Hilbert space is of the form
uf : H � ϕ �→ �ϕ, f � for a unique f ∈ H.
Problem 57. Density of Cc∞ (Rn ) in Lp (Rn ).
i) Recall in a few words why simple integrable functions
� are dense
in L1 (Rn ) with respect to the norm �f �L1 = Rn |f (x)|dx.
ii) Show that simple functions N
�
j=1 cj χ(Uj ) where the Uj are open
and bounded are also dense in L1 (Rn ).
iii) Show that if U is open and bounded then F (y) = v(U ∩ Uy ),
where Uy = {z ∈ Rn ; z = y + y � , y � ∈ U } is continuous in y ∈ Rn
and that
v(U ∩ Uy� ) + v(U � ∩ Uy ) → 0 as y → 0.
iv) If U is open and bounded and ϕ ∈ Cc∞ (Rn ) show that
�
f (x) = ϕ(x − y)dy ∈ Cc∞ (Rn ).
U

v) Show that if U is open and bounded then

�
sup |χU (x) − χU (x − y)|dx → 0 as δ ↓ 0.
|y|≤δ

vi) If U is open and bounded and ϕ ∈ Cc∞ (Rn ), ϕ ≥ 0,

�
ϕ = 1
then
fδ → χU in L1 (Rn ) as δ ↓ 0
where � �y �
−n
fδ (x) = δ χU (x − y)dy.
ϕ
δ
Hint: Write χU (x) = δ −n ϕ yδ χU (x) and use v).
� � �

vii) Conclude that Cc∞ (Rn ) is dense in L1 (Rn ).

viii) Show that Cc∞ (Rn ) is dense in Lp (Rn ) for any 1 ≤ p < ∞.
Problem 58. Schwartz representation theorem. Here we (well you) come
to grips with the general structure of a tempered distribution.
i) Recall briefly the proof of the Sobolev embedding theorem and
the corresponding estimate
n
sup |φ(x)| ≤ C�φ�H m , < m ∈ R.
x∈Rn 2
ii) For m = n + 1 write down a(n equivalent) norm on the right in
a form that does not involve the Fourier transform.
116 RICHARD B. MELROSE

iii) Show that for any α ∈ N0

�
|Dα (1 + |x|2 )N φ | ≤ Cα,N (1 + |x|2 )N |Dβ φ|.
� �
β≤α

iv) Deduce the general estimates

sup (1 + |x|2 )N |Dα φ(x)| ≤ CN �(1 + |x|2 )N φ�H N +n+1 .
|α|≤N
x∈Rn

v) Conclude that for each tempered distribution u ∈ S � (Rn ) there

is an integer N and a constant C such that
|u(φ)| ≤ C�(1 + |x|2 )N φ�H 2N ∀ φ ∈ S(Rn ).
vi) Show that v = (1 + |x|2 )−N u ∈ S � (Rn ) satisfies
|v(φ)| ≤ C�(1 + |D|2 )N φ�L2 ∀ φ ∈ S(Rn ).
vi) Recall (from class or just show it) that if v is a tempered dis­
tribution then there is a unique w ∈ S � (Rn ) such that (1 +
|D|2 )N w = v.
vii) Use the Riesz Representation Theorem to conclude that for each
tempered distribution u there exists N and w ∈ L2 (Rn ) such
that
(17.10) u = (1 + |D|2 )N (1 + |x|2 )N w.
viii) Use the Fourier transform on S � (Rn ) (and the fact that it is an
isomorphism on L2 (Rn )) to show that any tempered distribu­
tion can be written in the form
u = (1 + |x|2 )N (1 + |D|2 )N w for some N and some w ∈ L2 (Rn ).
ix) Show that any tempered distribution can be written in the form
u = (1+|x|2 )N (1+|D|2 )N +n+1 w˜ for some N and some w˜ ∈ H 2(n+1) (Rn ).
x) Conclude that any tempered distribution can be written in the
form

u = (1 + |x|2 )N (1 + |D|2 )M U for some N, M

and a bounded continuous function U
Problem 59. Distributions of compact support.
i) Recall the definition of the support of a distribution, defined in
terms of its complement
�
Rn \supp(u) = p ∈ Rn ; ∃ U ⊂ Rn , open, with p ∈ U such that u�U = 0
� �
 LECTURE NOTES FOR 18.155, FALL 2004 117

ii) Show that if u ∈ C −∞ (Rn ) and φ ∈ Cc∞ (Rn ) satisfy

supp(u) ∩ supp(φ) = ∅

then u(φ) = 0.
iii) Consider the space C ∞ (Rn ) of all smooth functions on Rn , with­
out restriction on supports. Show that for each N

�f �(N ) = sup |Dα f (x)|

|α|≤N, |x|≤N

is a seminorn on C ∞ (Rn ) (meaning it satisfies �f � ≥ 0, �cf � =

|c|�f � for c ∈ C and the triangle inequality but that �f � = 0
does not necessarily imply that f = 0.)
iv) Show that Cc∞ (Rn ) ⊂ C ∞ (Rn ) is dense in the sense that for
each f ∈ C ∞ (Rn ) there is a sequence fn in Cc∞ (Rn ) such that
�f − fn �(N ) → 0 for each N.
v) Let E � (Rn ) temporarily (or permanantly if you prefer) denote
the dual space of C ∞ (Rn ) (which is also written E(Rn )), that
is, v ∈ E � (Rn ) is a linear map v : C ∞ (Rn ) −→ C which is
continuous in the sense that for some N

(17.11) |v(f )| ≤ C�f �(N ) ∀ f ∈ C ∞ (Rn ).

Show that such a v ‘is’ a distribution and that the map E � (Rn ) −→
C −∞ (Rn ) is injective.
vi) Show that if v ∈ E � (Rn ) satisfies (17.11) and f ∈ C ∞ (Rn ) has
f = 0 in |x| < N + � for some � > 0 then v(f ) = 0.
vii) Conclude that each element of E � (Rn ) has compact support
when considered as an element of C −∞ (Rn ).
viii) Show the converse, that each element of C −∞ (Rn ) with compact
support is an element of E � (Rn ) ⊂ C −∞ (Rn ) and hence conclude
that E � (Rn ) ‘is’ the space of distributions of compact support.
I will denote the space of distributions of compact support by Cc−∞ (R).

Problem 60. Hypoellipticity of the heat operator H = iDt + Δ =

n
Dx2j on Rn+1 .
�
iDt +
j=1

(1) Using τ to denote the ‘dual variable’ to t and ξ ∈ Rn to denote

the dual variables to x ∈ Rn observe that H = p(Dt , Dx ) where
p = iτ + |ξ |2 .
(2) Show that |p(τ, ξ)| > 21 (|τ | + |ξ |2 ) .
118 RICHARD B. MELROSE

� 0 where it
(3) Use an inductive argument to show that, in (τ, ξ) =
makes sense,
|α|
1 � qk,α,j (ξ)
(17.12) Dτk Dξα =
p(τ, ξ) j=1
p(τ, ξ)k+j+1

where qk,α,j (ξ) is a polynomial of degree (at most) 2j − |α|.

(4) Conclude that if φ ∈ Cc∞ (Rn+1 ) is identically equal to 1 in a
neighbourhood of 0 then the function
1 − φ(τ, ξ)
g(τ, ξ) =
iτ + |ξ|2
is the Fourier transform of a distribution F ∈ S � (Rn ) with
sing supp(F ) ⊂ {0}. [Remember that sing supp(F ) is the com­
plement of the largest open subset of Rn the restriction of F to
which is smooth].
(5) Show that F is a parametrix for the heat operator.
(6) Deduce that iDt + Δ is hypoelliptic – that is, if U ⊂ Rn is an
open set and u ∈ C −∞ (U ) satisfies (iDt + Δ)u ∈ C ∞ (U ) then
u ∈ C ∞ (U ).
(7) Show that iDt − Δ is also hypoelliptic.

Problem 61. Wavefront set computations and more – all pretty easy,
especially if you use results from class.
i) Compute WF(δ) where δ ∈ S � (Rn ) is the Dirac delta function
at the origin.
ii) Compute WF(H(x)) where H(x) ∈ S � (R) is the Heaviside func­
tion
�
1 x>0
H(x) = .
0 x≤0
Hint: Dx is elliptic in one dimension, hit H with it.
iii) Compute WF(E), E = iH(x1 )δ(x� ) which is the Heaviside in
the first variable on Rn , n > 1, and delta in the others.
iv) Show that Dx1 E = δ, so E is a fundamental solution of Dx1 .
v) If f ∈ Cc−∞ (Rn ) show that u = E � f solves Dx1 u = f.
vi) What does our estimate on WF(E � f ) tell us about WF(u) in
terms of WF(f )?

Problem 62. The wave equation in two variables (or one spatial vari­
able).
 LECTURE NOTES FOR 18.155, FALL 2004 119

i) Recall that the Riemann function

�
− 41 if t > x and t > −x
E(t, x) =
0 otherwise
is a fundamental solution of Dt2 − Dx2 (check my constant).
ii) Find the singular support of E.
iii) Write the Fourier transform (dual) variables as τ, ξ and show
that
WF(E) ⊂ {0} × S1 ∪ {(t, x, τ, ξ); x = t > 0 and ξ + τ = 0}
∪ {(t, x, τ, ξ); −x = t > 0 and ξ = τ } .
iv) Show that if f ∈ Cc−∞ (R2 ) then u = E �f satisfies (Dt2 −Dx2 )u =
f.
v) With u defined as in iv) show that
supp(u) ⊂ {(t, x); ∃
(t� , x� ) ∈ supp(f ) with t� + x� ≤ t + x and t� − x� ≤ t − x}.
vi) Sketch an illustrative example of v).
vii) Show that, still with u given by iv),
sing supp(u) ⊂ {(t, x); ∃ (t� , x� ) ∈ sing supp(f ) with
t ≥ t� and t + x = t� + x� or t − x = t� − x� }.
viii) Bound WF(u) in terms of WF(f ).
Problem 63. A little uniqueness theorems. Suppose u ∈ Cc−∞ (Rn ) recall
that the Fourier transform û ∈ C ∞ (Rn ). Now, suppose u ∈ Cc−∞ (Rn )
satisfies P (D)u = 0 for some non­trivial polynomial P, show that u = 0.
Problem 64. Work out the elementary behavior of the heat equation.
i) Show that the function on R × Rn , for n ≥ 1,
� n � 2
�
t− 2 exp − |x4|t t>0
F (t, x) =
0 t≤0
is measurable, bounded on the any set {|(t, x)| ≥ R} and is
integrable on {|(t, x)| ≤ R} for any R > 0.
ii) Conclude that F defines a tempered distibution on Rn+1 .
iii) Show that F is C ∞ outside the origin.
iv) Show that F satisfies the heat equation
�n
(∂t − ∂x2j )F (t, x) = 0 in (t, x) �= 0.
j=1
120 RICHARD B. MELROSE

v) Show that F satisfies

(17.13) F (s2 t, sx) = s−n F (t, x) in S � (Rn+1 )
where the left hand side is defined by duality “F (s2 t, sx) = Fs ”
where
t x
Fs (φ) = s−n−2 F (φ1/s ), φ1/s (t, x) = φ( 2 , ).
s s
vi) Conclude that
n
�
(∂t − ∂x2j )F (t, x) = G(t, x)
j=1

where G(t, x) satisfies

(17.14) G(s2 t, sx) = s−n−2 G(t, x) in S � (Rn+1 )
in the same sense as above and has support at most {0}.
vii) Hence deduce that
n
�
(17.15) (∂t − ∂x2j )F (t, x) = cδ(t)δ(x)
j=1

for some real constant c.

Hint: Check which distributions with support at (0, 0) satisfy
(17.14).
viii) If ψ ∈ Cc∞ (Rn+1 ) show that u = F � ψ satisfies

(17.16) u ∈ C ∞ (Rn+1 ) and

sup (1 + |x|)N |Dα u(t, x)| < ∞ ∀ S > 0, α ∈ Nn+1 , N.
x∈Rn , t∈[−S,S]

ix) Supposing that u satisfies (17.16) and is a real­valued solution

of
�n
(∂t − ∂x2j )u(t, x) = 0
j=1

in Rn+1 , show that

�
v(t) = u2 (t, x)
Rn

is a non­increasing function of t.
Hint: Multiply the equation by u and integrate over a slab
[t1 , t2 ] × Rn .
 LECTURE NOTES FOR 18.155, FALL 2004 121

x) Show that c in (17.15) is non­zero by arriving at a contradiction

from the assumption that it is zero. Namely, show that if c = 0
then u in viii) satisfies the conditions of ix) and also vanishes
in t < T for some T (depending on ψ). Conclude that u = 0
for all ψ. Using properties of convolution show that this in turn
implies that F = 0 which is a contradiction.
xi) So, finally, we know that E = 1c F is a fundamental solution of
the heat operator which vanishes in t < 0. Explain why this
allows us to show that for any ψ ∈ Cc∞ (R × Rn ) there is a
solution of
n
�
(17.17) (∂t − ∂x2j )u = ψ, u = 0 in t < T for some T.
j=1

What is the largest value of T for which this holds?

xii) Can you give a heuristic, or indeed a rigorous, explanation of
why
|x|2
�
c= exp(− )dx?
Rn 4
xiii) Explain why the argument we used for the wave equation to
show that there is only one solution, u ∈ C ∞ (Rn+1 ), of (17.17)
does not apply here. (Indeed such uniqueness does not hold
without some growth assumption on u.)
Problem 65. (Poisson summation formula) As in class, let L ⊂ Rn be
an integral lattice of the form
� n
�
�
L= v= kj v j , kj ∈ Z
j=1

where the vj form a basis of Rn and using the dual basis wj (so wj · vi =
δij is 0 or 1 as i �= j or i = j) set
� n
�
�
L◦ = w = 2π kj wj , kj ∈ Z .
j=1

Recall that we defined

(17.18) C ∞ (TL ) = {u ∈ C ∞ (Rn ); u(z + v) = u(z) ∀ z ∈ Rn , v ∈ L}.
i) Show that summation over shifts by lattice points:
�
(17.19) AL : S(Rn ) � f �−→ AL f (z) = f (z − v ) ∈ C ∞ (TL ).
v∈L

defines a map into smooth periodic functions.

122 RICHARD B. MELROSE

ii) Show that there exists f ∈ Cc∞ (Rn ) such that AL f ≡ 1 is the
costant function on Rn .
iii) Show that the map (17.19) is surjective. Hint: Well obviously
enough use the f in part ii) and show that if u is periodic then
AL (uf ) = u.
iv) Show that the infinite sum
�
(17.20) F = δ(· − v) ∈ S � (Rn )
v∈L

does indeed define a tempered distribution and that F is L­

periodic and satisfies exp(iw · z)F (z) = F (z) for each w ∈ L◦
with equality in S � (Rn ).
v) Deduce that F̂ , the Fourier transform of F, is L◦ periodic, con­
clude that it is of the form
�
(17.21) F̂ (ξ) = c δ(ξ − w)
w∈L◦

vi) Compute the constant c.

vii) Show that AL (f ) = F � f.
viii) Using this, or otherwise, show that AL (f ) = 0 in C ∞ (TL ) if and
only if fˆ = 0 on L◦ .

Problem 66. For a measurable set Ω ⊂ Rn , with non­zero measure,

set H = L2 (Ω) and let B = B(H) be the algebra of bounded linear
operators on the Hilbert space H with the norm on B being

(17.22) �B�B = sup{�Bf �H ; f ∈ H, �f �H = 1}.

i) Show that B is complete with respect to this norm. Hint (prob­

ably not necessary!) For a Cauchy sequence {Bn } observe that
Bn f is Cauchy for each f ∈ H.
ii) If V ⊂ H is a finite­dimensional subspace and W ⊂ H is a
closed subspace with a finite­dimensional complement (that is
W + U = H for some finite­dimensional subspace U ) show that
there is a closed subspace Y ⊂ W with finite­dimensional com­
plement (in H) such that V ⊥ Y, that is �v, y� = 0 for all v ∈ V
and y ∈ Y.
iii) If A ∈ B has finite rank (meaning AH is a finite­dimensional
vector space) show that there is a finite­dimensional space V ⊂
H such that AV ⊂ V and AV ⊥ = {0} where

V ⊥ = {f ∈ H; �f, v� = 0 ∀ v ∈ V }.
 LECTURE NOTES FOR 18.155, FALL 2004 123

Hint: Set R = AH, a finite dimensional subspace by hypothesis.

Let N be the null space of A, show that N ⊥ is finite dimensional.
Try V = R + N ⊥ .
iv) If A ∈ B has finite rank, show that (Id −zA)−1 exists for all but
a finite set of λ ∈ C (just quote some matrix theory). What
might it mean to say in this case that (Id −zA)−1 is meromor­
phic in z? (No marks for this second part).
v) Recall that K ⊂ B is the algebra of compact operators, defined
as the closure of the space of finite rank operators. Show that
K is an ideal in B.
vi) If A ∈ K show that

Id +A = (Id +B)(Id +A� )

where B ∈ K, (Id +B)−1 exists and A� has finite rank. Hint:

Use the invertibility of Id +B when �B�B < 1 proved in class.
vii) Conclude that if A ∈ K then
� �⊥
{f ∈ H; (Id +A)f = 0} and (Id +A)H are finite dimensional.

Problem 67. [Separable Hilbert spaces]

i) (Gramm­Schmidt Lemma). Let {vi }i∈N be a sequence in a
Hilbert space H. Let Vj ⊂ H be the span of the first j ele­
ments and set Nj = dim Vj . Show that there is an orthonormal
sequence e1 , . . . , ej (finite if Nj is bounded above) such that Vj
is the span of the first Nj elements. Hint: Proceed by induction
over N such that the result is true for all j with Nj < N. So,
consider what happens for a value of j with Nj = Nj−1 + 1 and
add element eNj ∈ Vj which is orthogonal to all the previous
ek ’s.
ii) A Hilbert space is separable if it has a countable dense subset
(sometimes people say Hilbert space when they mean separable
Hilbert space). Show that every separable Hilbert space has a
complete orthonormal sequence, that is a sequence {ej } such
that �u, ej � = 0 for all j implies u = 0.
iii) Let {ej } an orthonormal sequence in a Hilbert space, show that
for any aj ∈ C,

N
� N
�
2
� aj ej � = |aj |2 .
j=1 j=1
124 RICHARD B. MELROSE

iv) (Bessel’s inequality) Show that if ej is an orthormal sequence

in a Hilbert space and u ∈ H then
N
�
� �u, ej �ej �2 ≤ �u�2
j=1

and conclude (assuming the sequence of ej ’s to be infinite) that

the series
∞
�
�u, ej �ej
j=1
converges in H.
v) Show that if ej is a complete orthonormal basis in a separable
Hilbert space then, for each u ∈ H,
∞
�
u= �u, ej �ej .
j=1

Problem 68. [Compactness] Let’s agree that a compact set in a metric

space is one for which every open cover has a finite subcover. You may
use the compactness of closed bounded sets in a finite dimensional
vector space.
i) Show that a compact subset of a Hilbert space is closed and
bounded.
ii) If ej is a complete orthonormal subspace of a separable Hilbert
space and K is compact show that given � > 0 there exists N
such that
�
(17.23) |�u, ej �|2 ≤ � ∀ u ∈ K.
j≥N

iii) Conversely show that any closed bounded set in a separable

Hilbert space for which (17.23) holds for some orthonormal basis
is indeed compact.
iv) Show directly that any sequence in a compact set in a Hilbert
space has a convergent subsequence.
v) Show that a subspace of H which has a precompact unit ball
must be finite dimensional.
vi) Use the existence of a complete orthonormal basis to show that
any bounded sequence {uj }, �uj � ≤ C, has a weakly conver­
gent subsequence, meaning that �v, uj � converges in C along
the subsequence for each v ∈ H. Show that the subsequnce can
be chosen so that �ek , uj � converges for each k, where ek is the
complete orthonormal sequence.
 LECTURE NOTES FOR 18.155, FALL 2004 125

Problem 69. [Spectral theorem, compact case] Recall that a bounded

operator A on a Hilbert space H is compact if A{�u� ≤ 1} is pre­
compact (has compact closure). Throughout this problem A will be a
compact operator on a separable Hilbert space, H.
i) Show that if 0 �= λ ∈ C then
Eλ = {u ∈ H; Au = λu}.
is finite dimensional.
ii) � {0})
If A is self­adjoint show that all eigenvalues (meaning Eλ =
are real and that different eigenspaces are orthogonal.
iii) Show that αA = sup{|�Au, u�|2 }; �u� = 1} is attained. Hint:
Choose a sequence such that |�Auj , uj �|2 tends to the supre­
mum, pass to a weakly convergent sequence as discussed above
and then using the compactness to a furhter subsequence such
that Auj converges.
iv) If v is such a maximum point and f ⊥ v show that �Av, f � +
�Af, v� = 0.
v) If A is also self­adjoint and u is a maximum point as in iii)
deduce that Au = λu for some λ ∈ R and that λ = ±α.
vi) Still assuming A to be self­adjoint, deduce that there is a finite­
dimensional subspace M ⊂ H, the sum of eigenspaces with
eigenvalues ±α, containing all the maximum points.
vii) Continuing vi) show that A restricts to a self­adjoint bounded
operator on the Hilbert space M ⊥ and that the supremum in
iii) for this new operator is smaller.
viii) Deduce that for any compact self­adjoint operator on a sep­
arable Hilbert space there is a complete orthonormal basis of
eigenvectors. Hint: Be careful about the null space – it could
be big.
Problem 70. Show that a (complex­valued) square­integrable function
u ∈ L2 (Rn ) is continuous in the mean, in the sense that
�
(17.24) lim sup |u(x + y) − u(x)|2 dx = 0.
�↓0 |y|<�

Hint: Show that it is enough to prove this for non­negative functions

and then that it suffices to prove it for non­negative simple functions
and finally that it is enough to check it for the characteristic function
of an open set of finite measure. Then use Problem 57 to show that it
is true in this case.
Problem 71. [Ascoli­Arzela] Recall the proof of the theorem of Ascoli
and Arzela, that a subset of C00 (Rn ) is precompact (with respect to the
126 RICHARD B. MELROSE

supremum norm) if and only if it is equicontinuous and equi­small at

infinity, i.e. given � > 0 there exists δ > 0 such that for all elements
u∈B
(17.25)
|y | < δ =⇒ sup |u(x + y) = u(x)| < � and |x| > 1/δ =⇒ |u(x)| < �.
x∈Rn

Problem 72. [Compactness of sets in L2 (Rn ).] Show that a subset B ⊂

L2 (Rn ) is precompact in L2 (Rn ) if and only if it satisfies the following
two conditions:
i) (Equi­continuity in the mean) For each � > 0 there exists δ > 0
such that
�
(17.26) |u(x + y) − u(x)|2 dx < � ∀ |y | < δ, u ∈ B.
Rn

ii) (Equi­smallness at infinity) For each � > 0 there exists R such

that
�
(17.27) |u|2 dx < � ∀ u ∈ B.
|x|>R|

Hint: Problem 70 shows that (17.26) holds for each u ∈ L2 (Rn ); check
that (17.27) also holds for each function. Then use a covering argument
to prove that both these conditions must hold for a compact subset of
L2 (R) and hence for a precompact set. One method to prove the con­
verse is to show that if (17.26) and (17.27) hold then B is bounded
and to use this to extract a weakly convergent sequence from any given
sequence in B. Next show that (17.26) is equivalent to (17.27) for the
set F (B), the image of B under the Fourier transform. Show, possi­
bly using Problem 71, that if χR is cut­off to a ball of radius R then
χR G(χR uˆn ) converges strongly if un converges weakly. Deduce from
this that the weakly convergent subsequence in fact converges strongly
so B¯ is sequently compact, and hence is compact.

Problem 73. Consider the space Cc (Rn ) of all continuous functions on

Rn with compact support. Thus each element vanishes in |x| > R for
some R, depending on the function. We want to give this a toplogy in
terms of which is complete. We will use the inductive limit topology.
Thus the whole space can be written as a countable union
(17.28)

Cc (Rn ) = {u : Rn ; u is continuous and u(x) = 0 for |x| > R}.

n

Each of the space on the right is a Banach space for the supremum
norm.
 LECTURE NOTES FOR 18.155, FALL 2004 127

(1) Show that the supreumum norm is not complete on the whole
of this space.
(2) Define a subset U ⊂ Cc (Rn ) to be open if its intersection with
each of the subspaces on the right in (17.28) is open w.r.t. the
supremum norm.
(3) Show that this definition does yield a topology.
(4) Show that any sequence {fn } which is ‘Cauchy’ in the sense that
for any open neighbourhood U of 0 there exists N such that fn −
fm ∈ U for all n, m ≥ N, is convergent (in the corresponding
sense that there exists f in the space such that f − fn ∈ U
eventually).
(5) If you are determined, discuss the corresponding issue for nets.
Problem 74. Show that the continuity of a linear functional u : Cc∞ (Rn ) −→
C with respect to the inductive limit topology defined in (6.16) means
precisely that for each n ∈ N there exists k = k(n) and C = Cn such
that
(17.29) |u(ϕ)| ≤ C�ϕ�C k , ∀ ϕ ∈ C˙∞ (B(n)).
The point of course is that the ‘order’ k and the constnat C can both
increase as n, measuring the size of the support, increases.
Problem 75. [Restriction from Sobolev spaces] The Sobolev embedding
theorem shows that a function in H m (Rn ), for m > n/2 is continuous
– and hence can be restricted to a subspace of Rn . In fact this works
more generally. Show that there is a well defined restriction map
1 1
(17.30) H m (Rn ) −→ H m− 2 (Rn ) if m >
2
with the following properties:
(1) On S (Rn ) it is given by u �−→ u(0, x� ), x� ∈ Rn−1 .
(2) It is continuous and linear.
Hint: Use the usual method of finding a weak version of the map on
smooth Schwartz functions; namely show that in terms of the Fourier
transforms on Rn and Rn−1
�
� −1
(17.31) �
u(0, ·)(ξ ) = (2π) ˆ 1 , ξ � )dξ1 , ∀ ξ � ∈ Rn−1 .
u(ξ
R
Use Cauchy’s inequality to show that this is continuous as a map on
Sobolev spaces as indicated and then the density of S (Rn ) in H m (Rn )
to conclude that the map is well­defined and unique.
Problem 76. [Restriction by WF] From class we know that the product
of two distributions, one with compact support, is defined provided
128 RICHARD B. MELROSE

they have no ‘opposite’ directions in their wavefront set:

/ WF(v) then uv ∈ Cc−∞ (Rn ).

(17.32) (x, ω) ∈ WF(u) =⇒ (x, −ω) ∈

Show that this product has the property that f (uv) = (f u)v = u(f v)
if f ∈ C ∞ (Rn ). Use this to define a restriction map to x1 = 0 for
distributions of compact support satisfying ((0, x� ), (ω1 , 0)) ∈
/ WF(u)
as the product

(17.33) u0 = uδ(x1 ).

[Show that u0 (f ), f ∈ C ∞ (Rn ) only depends on f (0, ·) ∈ C ∞ (Rn−1 ).

Problem 77. [Stone’s theorem] For a bounded self­adjoint operator A

show that the spectral measure can be obtained from the resolvent in
the sense that for φ, ψ ∈ H
1
(17.34) lim �[(A − t − i�)−1 − (A + t + i�)−1 ]φ, ψ� −→ µφ,ψ
�↓0 2πi
in the sense of distributions – or measures if you are prepared to work
harder!

Problem 78. If u ∈ S(Rn ) and ψ � = ψR + µ is, as in the proof of

Lemma 12.5, such that

supp(ψ � ) ∩ Css(u) = ∅

show that
S(Rn ) � φ �−→ φψ � u ∈ S(Rn )
is continuous and hence (or otherwise) show that the functional u1 u2
defined by (12.20) is an element of S � (Rn ).

Problem 79. Under the conditions of Lemma 12.10 show that

(17.35)
sx + ty
Css(u∗v)∩Sn−1 ⊂ { , |x| = |y | = 1, x ∈ Css(u), y ∈ Css(v), 0 ≤ s, t ≤ 1}.
|sx + ty |
Notice that this make sense exactly because sx + ty = 0 implies that
t/s = 1 but x + y =� 0 under these conditions by the assumption of
Lemma 12.10.

Problem 80. Show that the pairing u(v) of two distributions u, v ∈

b �
S (Rn ) may be defined under the hypothesis (12.50).
 LECTURE NOTES FOR 18.155, FALL 2004 129

Problem 81. Show that under the hypothesis (12.51)

(17.36)
WFsc (u∗v) ⊂ {(x+y, p); (x, p) ∈ WFsc (u)∩(Rn ×Sn−1 ), (y, p) ∈ WFsc (v)∩(Rn ×Sn−1 )}
s� θ� + s�� θ��
∪ {(θ, q) ∈ Sn−1 × Bn ; θ = � � , 0 ≤ s� , s�� ≤ 1,
|s θ + s�� θ�� |
(θ� , q) ∈ WFsc (u) ∩ (Sn−1 × Bn ), (θ�� , q) ∈ WFsc (v) ∩ (Sn−1 × Bn )}.
Problem 82. Formulate and prove a bound similar to (17.36) for WFsc (uv)
when u, v ∈ S � (Rn ) satisfy (12.50).
Problem 83. Show that for convolution u ∗ v defined under condition
(12.51) it is still true that
(17.37) P (D)(u ∗ v) = (P (D)u) ∗ v = u ∗ (P (D)v).
Problem 84. Using Problem 80 (or otherwise) show that integration is
defined as a functional
(17.38) {u ∈ S � (Rn ); (Sn−1 × {0}) ∩ WFsc (u) = ∅} −→ C.
� �
If u satisfies this condition, show that P (D)u = c u where c is the
constant term in P (D), i.e. P (D)1 = c.
Problem 85. Compute WFsc (E) where E = C/|x − y | is the standard
fundamental solution for the Laplacian on R3 . Using Problem 83 give
a condition on WFsc (f ) under which u = E�∗ f is defined and satisfies
Δu = f. Show that under this condition f is defined using Prob­
�lem 84. What can you say about WFsc (u)? Why is it not the case that
Δu = 0, even though this is true if u has compact support?
130 RICHARD B. MELROSE

18. Solutions to (some of) the problems

Solution 18.1 (To Problem 10). (by Matjaž Konvalinka).
Since the topology on N, inherited from R, is discrete, a set is com­
pact if and only if it is finite. If a sequence {xn } (i.e. a function N → C)
is in C0 (N) if and only if for any � > 0 there exists a compact (hence
finite) set F� so that |xn | < � for any n not in F� . We can assume that
F� = {1, . . . , n� }, which gives us the condition that {xn } is in C0 (N)
if and only if it converges to 0. We denote this space by c0 , and the
supremum norm by � · �0 . A sequence {xn } will be abbreviated to x.
Let l1 denote the space of (real or complex) sequences x with a finite
1­norm
�∞
�x�1 = |xn |.
n=1
We can define pointwise summation and multiplication with scalars,
and (l1 , � · �1 ) is a normed (in fact Banach) space. Because the func­
tional
�∞
y �→ xn yn
n=1
�∞ �∞
is linear and bounded (| n=1 xn yn | ≤ n=1 |xn ||yn | ≤ �x�0 �y�1 ) by
�x�0 , the mapping
Φ : l1 �−→ c∗0
defined by � ∞
�
�
x �→ y �→ xn yn
n=1
is a (linear) well­defined mapping with norm at most 1. In fact, Φ is
an isometry because if |xj | = �x�0 then |Φ(x)(ej )| = 1 where ej is
the j­th unit vector. We claim that Φ is also surjective (and hence an
isometric isomorphism). � If ϕ is a functional
�on c0 let us denote ϕ(ej )
∞ ∞
by xj . Then Φ(x)(y) = =1 ϕ(en )yn =
n� n=1 ϕ(yn en ) = ϕ(y) (the
last equality holds because ∞ y
n=1 n ne converges to y in c0 and ϕ is
continuous with respect to the topology in c0 ), so Φ(x) = ϕ.
Solution 18.2 (To Problem 29). (Matjaž Konvalinka) Since
� ∞
Dx H(ϕ) = H(−Dx ϕ) = i H(x)ϕ� (x) dx =
−∞
� ∞
i ϕ� (x) dx = i(0 − ϕ(0)) = −iδ(ϕ),
0
we get Dx H = Cδ for C = −i.
 LECTURE NOTES FOR 18.155, FALL 2004 131

Solution 18.3 (To Problem 40). (Matjaž Konvalinka) Let us prove this
� 0)
in the case where n = 1. Define (for b =
(b − x)k−1 (k−1)
U (x) = u(b) − u(x) − (b − x)u� (x) − . . . − u (x);
(k − 1)!
then
(b − x)k−1 (k)
U � (x) = − u (x).
(k − 1)!
For the continuously differentiable function V (x) = U (x)−(1−x/b)k U (0)
we have V (0) = V (b) = 0, so by Rolle’s theorem there exists ζ between
0 and b with
k(b − ζ)k−1
V � (ζ) = U � (ζ) + U (0) = 0
bk
Then
bk
U (0) = − U � (ζ),
k(b − ζ)k−1
u(k−1) (0) k−1 u(k) (ζ) k
u(b) = u(0) + u� (0)b + . . . + b + b .
(k − 1)! k!
The required decomposition is u(x) = p(x) + v(x) for
u�� (0) 2 u(k−1) (0) k−1 u(k) (0) k
p(x) = u(0) + u� (0)x + x + ... + x + x ,
2 (k − 1)! k!

u(k) (ζ) − u(k) (0) k

v(x) = u(x) − p(x) = x
k!
for ζ between 0 and x, and since u(k) is continuous, (u(x) − p(x))/xk
tends to 0 as x tends to 0.
The proof for general n is not much more difficult. Define the func­
tion wx : I → R by wx (t) = u(tx). Then wx is k­times continuously
differentiable,
n
�
� ∂u
wx (t) = (tx)xi ,
i=1
∂xi
n
� ∂2u
wx�� (t) = (tx)xi xj ,
i,j=1
∂xi ∂xj

� l! ∂lu
wx(l) (t) = l1 l2 li
(tx)xl11 xl22 · · · xlii
l1 !l2 ! · · · li ! ∂x1 ∂x2 · · · ∂xi
l1 +l2 +...+l =l
i
132 RICHARD B. MELROSE

so by above u(x) = wx (1) is the sum of some polynomial p (od degree

k), and we have
(k) (k)
u(x) − p(x) vx (1) wx (ζx ) − wx (0)
k
= k
= ,
|x| |x| k!|x|k
so it is bounded by a positive combination of terms of the form
∂lu ∂lu
� �
� �
� ∂xl1 ∂xl2 · · · ∂xli (ζx x) − ∂xl1 ∂xl2 · · · ∂xli (0)�
� �
1 2 i 1 2 i
with l1 + . . . + li = k and 0 < ζx < 1. This tends to zero as x → 0
because the derivative is continuous.
Solution 18.4 (Solution to Problem 41). (Matjž Konvalinka) Obviously
the map C0 (Bn ) → C(Bn ) is injective (since it is just the inclusion map),
and f ∈ C(Bn ) is in C0 (Bn ) if and only if it is zero on ∂Bn , ie. if and
only if f |Sn−1 = 0. It remains to prove that any map g on Sn−1 is the
restriction of a continuous function on Bn . This is clear since
�
|x|g(x/|x|) x = � 0
f (x) =
0 x=0
is well­defined, coincides with f on Sn−1 , and is continuous: if M is
the maximum of |g| on Sn−1 , and � > 0 is given, then |f (x)| < � for
|x| < �/M.
Solution 18.5. (partly Matjaž Konvalinka)
For any ϕ ∈ S(R) we have
� ∞ � ∞ � ∞
| ϕ(x)dx| ≤ 2
|ϕ(x)|dx ≤ sup((1+x| )|ϕ(x)|) (1+|x|2 )−1 dx
−∞ −∞ −∞
≤ C sup((1 + x|2 )|ϕ(x)|).
�
Thus S(R) � ϕ �−→ R ϕdx is continous.
Now, choose φ ∈ Cc∞ (R) with R φ(x)dx = 1. Then, for ψ ∈ S(R),
�
set
� x � ∞
(18.1) Aψ(x) = (ψ(t) − c(ψ)φ(t)) dt, c(ψ) = ψ(s) ds.
−∞ −∞
Note that the assumption on φ means that
� ∞
(18.2) Aψ(x) = − (ψ(t) − c(ψ)φ(t)) dt
x
Clearly Aψ is smooth, and in fact it is a Schwartz function since
d
(18.3) (Aψ(x)) = ψ(x) − cφ(x) ∈ S(R)
dx
 LECTURE NOTES FOR 18.155, FALL 2004 133

so it suffices to show that xk Aψ is bounded for any k as |x| → ±∞.

Since ψ(t) − cφ(t) ≤ Ck t−k−1 in t ≥ 1 it follows from (18.2) that
� ∞
k
|x Aψ(x)| ≤ Cx k
t−k−1 dt ≤ C � , k > 1, in x > 1.
x

A similar estimate as x → −∞ follows from (18.1). Now, A is clearly

linear, and it follows from the estimates above, including that on the
integral, that for any k there exists C and j such that
� �
�
sup |xα Dβ Aψ| ≤ C sup |xα Dβ ψ|.
α,β≤k x∈R
α� ,β � ≤j

Finally then, given u ∈ S � (R) define v(ψ) = −u(Aψ). From the

continuity of A, v ∈ S(R) and from the definition of A, A(ψ � ) = ψ.
Thus
dv
dv/dx(ψ) = v(−ψ � ) = u(Aψ � ) = u(ψ) =⇒ = u.
dx
�
Solution 18.6. We have to prove that �ξ�m+m u � ∈ L2 (Rn ), in other
words, that �
�
�ξ�2(m+m ) |u
�|2 dξ < ∞.
Rn
But that is true since
� �
2(m+m� ) 2 �
�ξ� |�|
u dξ = �ξ�2m (1 + ξ12 + . . . + ξn2 )m |u|
� 2 dξ =
Rn Rn
⎛ ⎞
� �� �
2m� 2m� 2α
� �
2α ⎠ 2 2
= �ξ� ⎝ Cα ξ |�|
u dξ = Cα �ξ� ξ |�|
u dξ
Rn |α|≤m |α|≤m Rn

m� m�
and since �ξ� ξ α u
� = �ξ� D � α u is in L2 (Rn ) (note that u ∈ H m (Rn )
�
follows from Dα u ∈ H m (Rn ), |α| ≤ m). The converse is also true since
Cα in the formula above are strictly positive.
Solution 18.7. Take v ∈ L2 (Rn ), and define subsets of Rn by
E0 = {x : |x| ≤ 1},
Ei = {x : |x| ≥ 1, |xi | = max |xj |}.
j
�n �n
Then obviously we have 1 = i=0 χEj a.e., and v = j=0 vj for vj =
√
χEj v. Then �x� is bounded by 2 on E0 , and �x�v0 ∈ L (Rn ); and on
2

Ej , 1 ≤ j ≤ n, we have
�x� (1 + n|xj |2 )1/2 � �1/2
≤ = n + 1/|xj |2 ≤ (2n)1/2 ,
|xj | |xj |
134 RICHARD B. MELROSE

2 n
�n�x�vj = xj wj for w2j ∈n L (R ). But that means that �x�v = w0 +
so
j=1 xj wj for wj ∈ L (R ).
If u is in L2 (Rn ) then u
� ∈ L2 (Rn ), and so there exist w0 , . . . , wn ∈
L2 (Rn ) so that
� n
�ξ�u
� = w0 + ξj wj ,
j=1
in other words
n
�
u
�=u
�0 + ξj u
�j
j=1

�j ∈ L2 (Rn ). Hence
where �ξ�u
n
�
u = u0 + Dj uj
j=1

where uj ∈ H 1 (Rn ).
Solution 18.8. Since
� ∞ � ∞
�
Dx H(ϕ) = H(−Dx ϕ) = i H(x)ϕ (x) dx = i ϕ� (x) dx = i(0−ϕ(0)) = −iδ(ϕ),
−∞ 0

we get Dx H = Cδ for C = −i.

Solution 18.9. It is equivalent to ask when �ξ�m δ�0 is in L2 (Rn ). Since
�
δ0 (ψ) = δ0 (ψ ) = ψ (0) =
� � � ψ(x) dx = 1(ψ),
Rn

this is equivalent to finding m such that �ξ�2m has a finite integral over
Rn . One option is to write �ξ� = (1 + r2 )1/2 in spherical coordinates,
and to recall that the Jacobian of spherical coordinates in n dimensions
has the form rn−1 Ψ(ϕ1 , . . . , ϕn−1 ), and so �ξ�2m is integrable if and only
if � ∞
rn−1
dr
0 (1 + r2 )m
converges. It is obvious that this is true if and only if n − 1 − 2m < −1,
ie. if and only if m > n/2.
Solution 18.10 (Solution to Problem31). We know that δ ∈ H m (Rn )
for any m < −n/1. Thus is just because �ξ�p ∈ L2 (Rn ) when p < −n/2.
Now, divide Rn into n + 1 regions, as above, being A0 = {ξ; |ξ | ≤ 1 and
Ai = {ξ; |ξi | = supj |ξj |, |ξ | ≥ 1}. Let v0 have Fourier transform χA0
and for i = 1, . . . , n, vi ∈ S; (Rn ) have Fourier transforms ξi−n−1 χAi .
Since |ξi | > c�ξ� on the support of v�i for each i = 1, . . . , n, each term
 LECTURE NOTES FOR 18.155, FALL 2004 135

is in H m for any m < 1 + n/2 so, by the Sobolev embedding theorem,

each vi ∈ C00 (Rn ) and
n
� �
(18.4) 1 = v̂0 ξin+1 v�i =⇒ δ = v0 + Din+1 vi .
i=1 i
How to see that this cannot be done with n or less derivatives? For
the moment I do not have a proof of this, although I believe it is true.
Notice that we are actually proving that δ can be written
�
(18.5) δ= Dα uα , uα ∈ H n/2 (Rn ).
|α|≤n+1

This cannot be improved to n from n + 1 since this would mean that

δ ∈ H −n/2 (Rn ), which it isn’t. However, what I am asking is a little
more subtle than this.
136 RICHARD B. MELROSE

References
[1] G.B. Folland, Real analysis, Wiley, 1984.
[2] F. G. Friedlander, Introduction to the theory of distributions, second ed., Cam­
bridge University Press, Cambridge, 1998, With additional material by M. Joshi.
MR 2000g:46002
[3] J. Hadamard, Le problème de Cauchy et les `equatons aux d´eriv´ees partielles
linéaires hyperboliques, Hermann, Paris, 1932.
[4] L. Hörmander, The analysis of linear partial differential operators, vol. 3,
Springer­Verlag, Berlin, Heidelberg, New York, Tokyo, 1985.
[5] W. Rudin, Real and complex analysis, third edition ed., McGraw­Hill, 1987.
[6] George F. Simmons, Introduction to topology and modern analysis, Robert E.
Krieger Publishing Co. Inc., Melbourne, Fla., 1983, Reprint of the 1963 original.
MR 84b:54002