Answer Table For The Multiple-Choice Questions: Quantum Mechanics Name

1
Quantum Mechanics NAME:

Homework 3: Formalism: Homeworks are not handed in or marked. But you get a mark for reporting
that you have done them. Once you’ve reported completion, you may look at the already posted supposedly
super-perfect solutions.
Answer Table for the Multiple-Choice Questions

a b c d e a b c d e
1. O O O O O 16. O O O O O
2. O O O O O 17. O O O O O
3. O O O O O 18. O O O O O
4. O O O O O 19. O O O O O
5. O O O O O 20. O O O O O
6. O O O O O 21. O O O O O
7. O O O O O 22. O O O O O
8. O O O O O 23. O O O O O
9. O O O O O 24. O O O O O
10. O O O O O 25. O O O O O
11. O O O O O 26. O O O O O
12. O O O O O 27. O O O O O
13. O O O O O 28. O O O O O
14. O O O O O 29. O O O O O
15. O O O O O 30. O O O O O
2
007 qmult 00100 1 1 5 easy memory: vector addition

1. The sum of two vectors belonging to a vector space is:
a) a scalar.
b) another vector, but in a different vector space.
c) a generalized cosine.
d) the Schwarz inequality.
e) another vector in the same vector space.
SUGGESTED ANSWER: (e)
Wrong Answers:
b) Exactly wrong.
Redaction: Jeffery, 2001jan01
007 qmult 00200 1 4 4 easy deducto-memory: Schwarz inequality

2. “Let’s play Jeopardy! For $100, the answer is: |hα|βi|2 ≤ hα|αihβ|βi.”
What is , Alex?
a) the triangle inequality b) the Heisenberg uncertainty principle c) Fermat’s last theorem
d) the Schwarz inequality e) Schubert’s unfinished last symphony
SUGGESTED ANSWER: (d)
Wrong answers:
a) Semi-plausible: that’s the other simple inequality of vector space linear algebra.
b) Where’s −h?
c) Andrew Wiles finally proved this theorem in 1994. See
http://www-groups.dcs.st-andrews.ac.uk/~history/HistTopics/.
e) There’s some famous unfinished symphony. It may be Schubert’s.
007 qmult 00300 1 4 5 easy deducto-memory: Gram-Schmidt procedure

3. Any set of linearly independent vectors can be orthonormalized by the:
a) Pound-Smith procedure. b) Li Po tao. c) Sobolev method. d) Sobolev-P method.
e) Gram-Schmidt procedure.
Wrong Answers:
a) A play on the real answer.
b) Li Po was 8th century Chinese poet: see
http://www.santafe.edu/~shalizi/Poetry/.
Tao is the way. Tao can an also be a personal name: e.g. Tao Pang my old friend at
University of Nevada, Las Vegas.
c) A radiative transfer method for moving atmospheres.
d) A variation on the Sobolev method for treating polarized transfer. It’s all my own
invention.
007 qmult 00400 1 4 4 moderate memory: definition unitary matrix

4. A unitary matrix is defined by the expression:
a) U = U T , where superscript T means transpose. b) U = U † . c) U = U ∗ .
d) U −1 = U † . e) U −1 = U ∗ .
SUGGESTED ANSWER: (d)
Wrong Answers:
a) Symmetric matrix.
3
b) Hermitian matrix.
c) Real matrix.
e) Some kind of matrix.
007 qmult 00500 2 3 4 moderate math: trivial eigenvalue problem

5. What are the eigenvalues of
1 −i
?
i 1
a) Both are 0. b) 0 and 1. c) 0 and −1. d) 0 and 2. e) −i and 1.

SUGGESTED ANSWER: (d) The eigenvalue equation is
0 = (1 − λ) − 1 = −2λ + λ2
which clearly has solutions 0 and 2.

Wrong Answers:
e) As the matrix is Hermitian, it’s eigenvalues must be pure real: −i cannot be an eigenvalue.
007 qfull 00090 1 5 0 easy thinking: ordinary vector space

Extra keywords: (Gr-77:3.1)
6. Consider ordinary 3-dimensional vectors with complex components specified by a 3-tuple: (x, y, z). They
constitute a 3-dimensional vector space. Are the following subsets of this vector space vector spaces? If
so, what is their dimension? HINT: See Gr-435 for all the properties a vector space must have.
a) The subset of all vectors (x, y, 0).
b) The subset of all vectors (x, y, 1).
c) The subset of all vectors of the form (a, a, a), where a is any complex number.
SUGGESTED ANSWER:
a) It would be pedantic to enumerate the properties. The subset of all vectors (x, y, 0) is clearly a
vector space of dimension 2. The dimensionality is clear since one only needs two basis vectors
to construct any member: say (1, 0, 0) and (0, 1, 0).
b) Nyet: the sum of two members of the subset, e.g.,
(x1 , y1 , 1) + (x3 , y2 , 1) = (x1 + x2 , y1 + y2 , 2)
is not in the subset.

c) Da—if that is Russian for yes. The subset is, in fact, isomorphic to the set of complex numbers.
The only basis vector of the set can be chosen conveniently to be (1, 1, 1). With only one basis
vector needed the set is 1-dimensional.
007 qfull 00100 2 5 0 moderate thinking: vector space, polynomial

7. A vector space is constituted by a set of vectors {|αi, |βi, |γi, . . .} and a set of scalars {a, b, c, . . .}
(ordinary complex numbers is all that quantum mechanics requires) subject to two operations vector
addition and scalar multiplication obeying the certain rules. Note it is the relations between vectors
that make them constitute a vector space. What they “are” we leave general. The rules are:
i) A sum of vectors is a vector:
|αi + |βi = |γi ,
4
where |αi and |βi are any vectors in the space and |γi also in the space. Note we have not defined
what vector addition consists of. That definition goes beyond the general requirements.
ii) Vector addition is commutative:
|αi + |βi = |βi + |αi .
iii) Vector addition is associative:
(|αi + |βi) + |γi = |αi + (|βi + |γi) .
iv) There is a zero or null vector |0i such that
|αi + |0i = |αi ,
v) For every vector |αi there is an inverse vector | − αi such that
|αi + | − αi = |0i .
vi) Scalar multiplication of a vector gives a vector:
a|αi = |βi .
vii) Scalar multiplication is distributive on vector addition:
a(|αi + |βi) = a|αi + a(|βi) .
viii) Scalar multiplication is distributive on scalar addition:
(a + b)|αi = a|αi + b|αi .
ix) Scalar multiplication is associative with respect to scalar multiplication:
(ab)|αi = a(b|αi) .
x) One has
0|αi = |0i .
xi) Finally, one has

1|αi = |αi .
NOTE: Note that
|0i = 0|αi = [1 + (−1)]|αi = |αi + (−1)|αi ,
and thus we find that
| − αi = −|αi .
So the subtraction of a vector is just the addition of its inverse. This is consistent with all ordinary
math.
If any vector in the space can be written as linear combination of a set of linearly independent
vectors, that set is called a basis and is said to span the set. The number of vectors in the basis is the
dimension of the space. In general there will be infinitely many bases for a space.
Finally the question. Consider the set of polynomials {P (x)} (with complex coefficients) and degree
less than n. For each of the subsets of this set (specified below) answer the following questions: 1) Is
the subset a vector space? Inspection usually suffices to answer this question. 2) If not, what property
does it lack? 3) If yes, what is the most obvious basis and what is the dimension of the space?
a) The subset that is the whole set.
5
b) The subset of even polynomials.

c) The subset where the highest term has coefficient a (i.e., the leading coefficient is a) and a is a
general complex number, except a 6= 0.
d) The subset where P (x = g) = 0 where g is a general real number. (To be really clear, I mean the
subset of polynomials that are equal to zero at the point x = g.)
e) The subset where P (x = g) = h where g is a general real number and h is a general complex
number, except h 6= 0.
SUGGESTED ANSWER:
a) By inspection the set has all the requisite properties, and so does constitute a vector space.
The set {xℓ } with ℓ an integer in the range [0, n − 1] is an obvious basis set for the vector
space since the elements are all linearly independent and any polynomial of degree less than
n can be constructed by linear combination from them. From the number of basis vectors we
conclude that the dimension of the space is n.
Note the choice of basis is obvious and convenient, but not unique. For example say we
define
a+ = x1 + x0
and
a− = x1 − x0 .
We can construct the two lowest degree basis vectors x0 and x1 from a+ and a− :
1
x1 = (a+ + a− )
2
and
1
x0 = (a+ − a− ) .
2
Therefore we can replace x0 and x1 in the basis by a+ and a− to create a new basis. Note also
that a+ and a− are in fact independent. Say we tried to set
a− = Ca+ + linear combination of other basis vectors ,
where C is some constant to be determined. The other basis vectors have no x0 or x1 in them,
and so can contribute nothing. Solving for C we find that C must equal both 1 and −1. This
is impossible: a+ and a− are thus proven to be linearly independent.
b) By inspection the set has all the requisite properties, and so does constitute a vector space.
The set {xℓ } with ℓ an even integer in the range [0, n − 2] if n is even and [0, n − 1] if n is odd
is an obvious basis set for the vector space since the elements are all linearly independent and
any even polynomial of degree less than n can be constructed by linear combination from them.
The dimension of the space is (n − 2)/2 + 1 = n/2 if n is even and (n − 1)/2 + 1 = (n + 1)/2 if
n is odd. The fact that there is a x0 vector in our convenient basis clues us in that there must
be those “+1”’s in the expressions for the dimensions—if it wasn’t obvious from the foregoing.
c) No. The polynomial that is the sum of two elements have leading coefficient 2a, and thus this
polynomial is not in the specified subset.
d) By inspection the set has all the requisite properties, and so does constitute a vector space.
The set {(x − g)ℓ } with ℓ an integer in the range [1, n − 1] is a natural basis set for the vector
space since the elements are all linearly independent and any polynomial in the subset of order
less than n can be constructed by linear combination from them. Maybe a demonstration is
needed. Say P (x) is a general element in the subset, then
n−1
X n−1
X
P (x) = aℓ xℓ = aℓ [(x − g) + g]ℓ
ℓ=0 ℓ=0
n−1
X n−1
X
= bℓ (x − g)ℓ = bℓ (x − g)ℓ ,
ℓ=0 ℓ=1
6
where we have done a rearrangement of terms using the binomial theorem implicitly and
where we have recognized that b0 = 0 since by hypothesis P (x) is in the subset. Since P (x)
is a general member of the subset, {(x − g)ℓ } is a basis for the subset. The dimension of the
space is obviously n − 1: there is no place for the non-existent (x − g)0 unit vector.
e) No. The polynomial that is the sum of two elements when evaluated at g equals 2h, and thus
polynomial is not in the specified subset.
007 qfull 00110 2 5 0 moderate thinking: unique expansion in basis

8. Prove that the expansion of a vector in terms of some basis is unique: i.e., the set of expansion coefficients
for the vector is unique.
SUGGESTED ANSWER:
Say |ψi is a general vector of the vector space for which {|φi i} is the basis. Assume that we
have two different expansions X X
|ψi = ci |φi i = di |φi i ,
i i
then X
0= (ci − di )|φi i .
i
But since the basis vectors all independent by the hypothesis that they are basis vectors, no term
i can be completely eliminated by any combination of the other terms. The only possibility is that
ci − di = 0 for all i or ci = di for all i. Thus, our two different expansions must, in fact, be the
same. An expansion is unique.
If one would like a more concrete demonstration, assume that cj − dj 6= 0 for some j. Then
we can divide the second sum above through by cj − dj , move |φj i to the other side, and multiply
through by −1 to get
X ci − di
|φj i = − |φi i .
cj − dj
i, but i6=j
But now we find |φj i is a linear combination of the other “basis” vectors. This cannot be if our
vectors are independent. We conclude again that ci = di for all i and the expansion is unique.
007 qfull 00200 3 5 0 tough thinking: Gram-Schmidt orthonormalization

9. Say {|αi i} is a basis (i.e., a set of linearly independent vectors that span a vector space), but it is
not orthonormal. The first step of the Gram-Schmidt orthogonalization procedure is to normalize the
(nominally) first vector to create a new first vector for a new orthonormal basis:
|α1 i
|α′1 i = ,
||α1 ||
where recall that the norm of a vector |αi is given by

p
||α|| = || |α1 i || = hα|αi .
The second step is create a new second vector that is orthogonal to the new first vector using the old
second vector and the new first vector:
|α2 i − |α′1 ihα′1 |α2 i

|α′2 i = .
|| |α2 i − |α′1 ihα′1 |α2 i ||
Note we have subtracted the projection of |α2 i on |α′1 i from |α2 i and normalized.
7
a) Write down the general step of the Gram-Schmidt procedure.

b) Why must an orthonormal set of non-null vectors be a linearly independent.
c) Is the result of a Gram-Schmidt procedure independent of the order the original vectors are used?
HINT: Say you first use vector |αa i of the old set in the procedure. The first new vector is just
|αa i normalized: i.e., |α′a i=|αa i/||αa ||. All the other new vectors will be orthogonal to |α′a i. But
what if you started with |αb i which in general is not orthogonal to |αa i?
d) How many orthonormalized bases can an n dimensional space have in general? (Ignore the strange
n = 1 case.) HINT: Can’t the Gram-Schmidt procedure be started with any vector at all in the
vector space?
e) What happens in the procedure if the original vector set {|αi i} does not, in fact, consist of all
linearly independent vectors? To understand this case analyze another apparently different case.
In this other case you start the Gram-Schmidt procedure with n original vectors. Along the way
the procedure yields null vectors for the new basis. Nothing can be done with the null vectors:
they can’t be part of a basis or normalized. So you just put those null vectors and the vectors
they were meant to replace aside and continue with the procedure. Say you got m null vectors in
the procedure and so ended up with n − m non-null orthonormalized vectors. Are these n − m
new vectors independent? How many of the old vectors were used in constructing the new n − m
non-null vectors and which old vectors were they? Can all the old vectors be recontructed from the
the new n − m non-null vectors? Now answer the original question.
f) If the original set did consist of n linearly independent vectors, why must the new orthonormal set
consist of n linearly independent vectors? HINT: Should be just a corollary of the part (e) answer.
g) Orthonormalize the 3-space basis consisting of
     
1+i i 0
|α1 i =  1  , |α2 i =  3  , and |α3 i =  32  .
i 1 0
Input the vectors into the procedure in the reverse of their nominal order: why might a marker
insist on this? Note setting kets equal to columns is a lousy notation, but you-all know what I
mean. The bras, of course, should be “equated” to the row vectors. HINT: Make sure you use the
normalized new vectors in the construction procedure.
SUGGESTED ANSWER:
a) Behold:
Pi−1
|αi i − j=1 |α′j ihα′j |αi i
|α′i i = Pi−1 ,
|| |αi i − j=1 |α′j ihα′j |αi i ||
where the summation upper limit −1 is interpreted as meaning the whole summation is just a
zero.
Just to demonstrate concretely that |α′i i is indeed orthogonal to all |α′k i with k < i (i.e.,
the all already constructed new basis vectors), we evaluate hα′k |α′i i:
Pi−1
hα′k |αi i − ′ ′ ′
j=1 hαk |αj ihαj |αi i
hα′k |α′i i = Pi−1
|| |αi i − j=1 |α′j ihα′j |αi i ||
hα′k |αi i − i−1 ′
P
j=1 δkj hαj |αi i
= Pi−1 ′
|| |αi i − j=1 |αj ihα′j |αi i ||
hα′k |αi i − hα′k |αi i
= Pi−1
|| |αi i − j=1 |α′j ihα′j |αi i ||
=0,
where we have used the fact that the new vectors for k < i are orthogonal. Pretty clearly
|α′i i is normalized: it’s explicitly so actually. So |α′i i is orthonormalized with respect to all the
previously constructed new basis vectors.
8
b) Quite obviously if the set is orthonormal, none of the set can be expanded in terms of a
combination of the others. Since they can’t, they are by definition all linearly independent.
To be absolutely concrete assume contrafactually that one of a set of orthonormal vectors
is expanded in the others: X
|αi i = |αj i .
j6=i
Then take the scalar product with bra hαi |:
hαi |αi i = 0
by orthogonality. But the vectors were non-null by hypothesis, and so our assumption of
dependence was wrong.
c) No. The first vector you input to the procedure yields a normalized version of itself. In general
all the other vectors arn’t orthogonal to that vector: therefore their counterparts in the new
set are not in general just normalized versions of themselves. But you can start the procedure
with any of the original vectors you like. Thus you can preserve any one original vector you
wish in normalized form, but not in general the others. Thus every starting input vector can
in general lead to a different orthonormal set. I think it’s intuitively suggested that if you have
n linearly independent original vectors that up to n! orthonormalized bases can be generated.
But the proof is beyond me at this time of night. Hm: it seems obviously true though.
d) There are infinitely many choices of vector direction in a vector space of dimension n ≥ 2.
Any such vector can be used as the starting vector in a Gram-Schmidt procedure. The other
n − 1 linear independent original vectors can be obtained from almost any old basis: the n − 1
vectors must be linearly independent of your starting vector of course.
e) Consider n vectors in an original set {|αi i} and carry out the Gram-Schmidt procedure to the
end. Along the way m of the new vectors were null; you just ignored these and the old vectors
they were meant to replace. Null vectors can’t be normalized or form part of a basis. At the
end of the procedure you have a new set of n − m non-null, orthonormal vectors {|α′i i}. Since
they are orthogonal, they are all linearly independent. Thus the set {|α′i i} is an orthonormal
basis for an n − m dimensional space.
In constructing the n − m new vectors you, in fact, used only n − m of the original vectors:
the vectors you used were the n − m ones that did not give null new vectors.
You can reconstruct the entire original set from the n − m new vectors. First, every |αi i
that gave a null vector in the procedure is clearly a linear combination of the newly constructed
basis:
i−1
X
|αi i = |α′j ihα′j |αi i .
j=1
Second, every |αi i that gave a non-null vector is also a linear combination of the newly
constructed basis:
 
i−1
X i−1
X
|αi i = || |αi i − |α′j ihα′j |αi i || |α′i i + |α′j ihα′j |αi i .
j=1 j=1
Thus, the original set {|αi i} is entirely contained in the n − m space spanned by the new basis.
Since the n − m new vectors were constructed from n − m old vectors and the n − m
new vectors can be used to construct any of the n vectors in the original set, the n − m old
vectors used to construct the new n − m vectors can be used to construct m vectors of the
original set that gave nulls in the procedure. Thus there were in the original set only n − m
independent vectors and the original set spanned only an n − m dimensional space. Of course,
one doesn’t mean that the n − m old vectors used in the Gram-Schmidt procedure were “the
independent” ones and the m neglected old vectors were “the non-independent” ones. One
means some set of m vectors of the n old vectors could be constructed from the others and
eliminated beforehand from the Gram-Schmidt procedure. Which set of m old vectors could
be chosen in various ways in general depending on the exact nature of the n old vectors. In the
9
Gram-Schmidt procedure which m old vectors get neglected depends in general on the order
one chooses to input vectors into the procedure.
Now to answer the original question. If the original set of vectors is not independent then
one will get nulls in the Gram-Schmidt procedure. If one doesn’t get null vectors, then one
constructs independent vectors out of non-independent vectors which is a logical inconsistency.
The number of non-null vectors the procedure yields is exactly the number of independent
vectors in the original set or (meaning the same thing) is the dimension of the space spanned
by the original set. The Gram-Schmidt procedure offers a systematic a way to test for the
linear independence of a set of vectors. Every null new vector means one more of the original
vectors was not linearly independent.
f) From the part (e) answer, it follows that if all the original vectors were linearly independent,
then no no null vectors are generated in the Gram-Schmidt procedure. Thus the procedure
generates n new vectors given n old linearly independent vectors. The n new vectors are clearly
linearly independent since they are orthogonal: see the part (b) answer. Thus if the original
set consisted of n linearly independent vectors, the new set must too.
g) I think inputting the vectors in the reverse of the nominal order gives the simplest result. I
insist the student’s do so, so that everyone should get the same orthonormalized set if all goes
well. Recall from the part (c) answer that the orthonormalized set does depend on the order
the original vectors are used in the procedure. Behold:
 
0
|α′3 i =  1 
0
     
i 0 i
 3  −  1  (0, 1, 0) ·  3 
1 0 1
|α′2 i = √
...
 
i
1
= √ 0
2 1
         
1+i 0 1+i i 1+i
1
 1  −  1  (0, 1, 0) ·  1  −  0  (−i, 0, 1) ·  1 
2
i 0 i 1 i
|α′3 i = √
...
     
1+i 0 i
 1  −  1  − 1  0  (−i + 1 + i)
2
i 0 1
= √
...
   
1+i i
 0  − 1 0
2
i 1
= √
...
i
 
r 1+
2 02 .

=
5  1
i−
2
007 qfull 00300 2 3 0 moderate math: prove the Schwarz inequality

10. As Andy Rooney says (or used to say if this problem has reached the stage where only old fogies
remember that king of the old fogies) don’t you just hate magic proofs where you start from some
10
unmotivated expression and do a number of unmotivated steps to arrive at a result that you could never
have been guessed from the way you were going about getting it. Well sans too many absurd steps, let
us see if we can prove the Schwarz inequality
|hα|βi|2 ≤ hα|αihβ|βi
for general vectors |αi and |βi. Note the equality only holds in two cases. First when |βi = a|αi, where
a is some complex constant. Second, when either or both of |αi and |βi are null vectors: in this case,
one has zero equals zero.
NOTE: A few facts to remember about general vectors and inner products. Say |αi and |βi are general
vectors. By the definition of the inner product, we have that hα|βi = hβ|αi∗ . This implies that hα|αi is
pure real. If c is a general complex number, then the inner product of |αi and c|βi is hα|c|βi = chα|βi.
Next we note that that another inner-product property
p is that hα|αi ≥ 0 and the equality only holds if
|αi is the null vector. The norm of |αi is ||α|| = hα|αi and |αi can be normalized if it is not null: i.e.,
for |αi not null, the normalized version is |α̂i = |αi/||α||.
a) In doing the proof of the Schwarz inequality, it is convenient to have the result that the bra
corresponding to c|βi (where |βi is a general vector and c is a general complex number) is hβ|c∗ .
Prove this correspondance. HINT: Consider general vector |αi and the inner product
hα|c|βi
and work your way by valid steps to

hβ|c∗ |αi∗
and that completes the proof since
hα|γi = hγ|αi∗
for general vectors |αi and |γi.
b) The next thing to do is to figure out what the Schwarz inequality is saying about vectors including
those 3-dimensional things we have always called vectors. Let us a restrict the generality of |αi by
demanding it not be a null vector for which the Schwarz inequality is already proven. Since |αi
is not null, it can be normalized. Let |α̂i = |αi/||α|| be the normalized version of |αi. Divide the
Schwarz inequality by ||α||2 . Now note that the component of |βi along the |α̂i direction is
|βk i = |α̂ihα̂|βi .
Evaluate hβk |βk i. Now what is the Schwarz inequality telling us.
c) The vector component of |βi that is orthogonal to |α̂i (and therefore |βk i) is
|β⊥ i = |βi − |βk i .
Prove this and then prove the Schwarz inquality itself (for |αi not null) by evaluating hβ|βi expanded
in components. What if |αi is a null vector?
SUGGESTED ANSWER:
a) Behold:
hα|c|βi = chα|βi = chβ|αi∗ = (c∗ hβ|αi)∗ = hβ|c∗ |αi∗ .
Since |αi, |βi, and c are general, we find that bra corresponding to general c|βi is hβ|c∗ .
b) Well doing the suggested division we get
|hα̂|βi|2 ≤ hβ|βi .
Now we find
hβk |βk i = hβ|α̂ihα̂|α̂ihα̂|βi = hβ|α̂ihα̂|βi = |hα̂|βi|2 ,
where we have used the part (a) result and the fact that hα̂|α̂i = 1. We see that the left-
hand side of the inequality is the magnitude squared of |βk i and the right-hand side is the
11
magnitude squared of |βi. Now it is clear. The Schwarz inequality just means that magnitude
of the component of a vector along some axis is less than or equal to the magnitude of the
vector itself.
The equality holds if |βi is aligned with the axis defined by |α̂i. In this case,
|βi = a|α̂i ,
where a is some constant, and
|βk i = a|α̂ihα̂|α̂i = a|α̂i = |βi .
So |βk i is |βi in this case. This is just what the question preamble preambled.
There are contexts in which the Schwarz inequality is used for some purpose not clearly
connected to the meaning found above. Most importantly in quantum mechanics, the Schwarz
inequality is used in its ordinary form in proving the generalized uncertainy principle (Gr-110).
c) Behold:
hα̂|β⊥ i = hα̂|βi − hα̂|βk i = hα̂|βi − hα̂|α̂ihα̂|βi = hα̂|βi − hα̂|βi = 0 ,
where we have used the normalization of |α̂i. Since |βk i = |α̂ihα̂|βi (i.e., |βk i is parallel to
|αi), we have that |β⊥ i is orthogonal to |βk i too.
Now
|βi = |βk i + |β⊥ i ,
and so
hβ|βi = hβk |βk i + hβk |β⊥ i + hβ⊥ |βk i + hβ⊥ |β⊥ i

= hβk |βk i + hβ⊥ |β⊥ i ,
where we have used orthogonality of |β⊥ i and |βk i. By an inner-product property, we know
that
hβk |βk i ≥ 0 and hβ⊥ |β⊥ i ≥ 0 .
Thus, we have
hβk |βk i = |hα̂|βi|2 ≤ hβ|βi .
The last expression is equivalent to the Schwarz inequality for |αi not null as we saw in the
part (a) answer. So we have proven the Schwarz inequality for the case that |αi is not a null
vector.
If |αi is a null vector, then both sides of the Schwarz inequality are zero and the inequality
holds in its equality case. This is just what the question preamble preambled.
I have to say that WA-513–514’s proof of the Schwarz inequality seems unnecessarily
obscure since it hides the meaning of Schwarz inequality. It also deters people from finding the
above proof which seems the straightforward one to me and is based on the meaning of the
Schwarz inequality.
007 qfull 00310 1 3 0 easy math: find a generalized angle

11. The general inner-product vector space definition of generalized angle according to Gr-440 is
|hα|βi|
cos θgen = p ,
hα|αihβ|βi
where |αi and |βi are general non-zero vectors.

a) Is this definition completely consistent with the ordinary definition of an angle from the ordinary
vector dot product? Why or why not?
12
b) Find the generalized angle between vectors

   
1+i 4−i
|αi =  1  and |βi =  0  .
i 2 − 2i
SUGGESTED ANSWER:
a) Well no the definition is not completely consistent. Say ~a and ~b are ordinary vectors with
magnitudes a and b, respectively. If you put them in the generalized angle formula you get
~a · ~b

cos θgen = .
ab
But the ordinary vector dot product formula is
~a · ~b
cos θ = .
ab
Thus
cos θ , for cos θ ≥ 0;
cos θgen = | cos θ| =
− cos θ = cos(π − θ) , for cos θ ≤ 0.
Thus for θ ∈ [0, π/2], we get θgen = θ and for θ ∈ [π/2, π], θgen = π − θ.
b) Well
hα|αi = 4 ,
hβ|βi = 25 ,
and  
4−i
hα|βi = ( 1 − i 1 −i )  0  = 3 − 5i + 0 − 2i − 2 = 1 − 7i .
2 − 2i
Thus the cosine of the generalized angle is

√
1 + 49 1
cos θgen = = √ .
2×5 5 2
The corresponding angle is θgen = π/4 = 45◦ .
007 qfull 00400 1 3 0 easy math: prove triangle inequality

12. Prove the triangle inequality:
||(|αi + |βi)|| ≤ ||α|| + ||β|| .
HINT: Start with ||(|αi + |βi)||2 , expand, and use reality and the Schwarz inequality
|hα|βi|2 ≤ hα|αihβ|βi = ||α||2 × ||β||2 .
SUGGESTED ANSWER:
Behold:
||(|αi + |βi)||2 = hα|αi + hβ|βi + hα|βi + hβ|αi

= hα|αi + hβ|βi + 2Re[hα|βi]
≤ hα|αi + hβ|βi + 2|hα|βi|
≤ hα|αi + hβ|βi + 2||α|| × ||β||
= ||α||2 + ||β||2 + 2||α|| × ||β||
2
= (||α|| + ||β||) ,
13
where we have used the Schwarz inequality to get the 4th line. The triangle inequality follows at
once:
||(|αi + |βi)|| ≤ ||α|| + ||β|| .
Interpreting the triangle in terms of ordinary vectors, it means that the sum of the magnitudes
of the two vectors is always greater than or equal to the magnitude of the vector sum. The equality
only holds if the vectors are parallel and not antiparallel. The Schwarz inequality interpretted in
terms of ordinary vectors is that the magnitude of a component of a vector along an axis is always
than or equal to the magnitude of the vector. The equality only holds when the vector is aligned
with the axis. One can see the situation by dividing the Schwarz inequality by one vector magnitude
in order to create a unit vector for an axis: e.g.,
|hα̂|βi|2 ≤ hβ|βi .
007 qfull 00500 3 3 0 tough math: simple matrix identities

13. Prove the following matrix identities:
a) (AB)T = B T AT , where superscript “T” means transpose.
b) (AB)† = B † A† , where superscript † means Hermitian conjugate.
c) (AB)−1 = B −1 A−1 .
d) (U V )−1 = (U V )† (i.e., U V is unitary) given that U and V are unitary. In other words, prove the
product of unitary matrices is unitiary.
e) (AB)† = AB (i.e., AB is Hermitian) given that A and B are commuting Hermitian matrices. Does
the converse hold: i.e., does (AB)† = AB imply A and B are commuting Hermitian matrices?
HINTS: Find a trivial counterexample. Try B = A−1 .
f) (A + B)† = A + B (i.e., A + B is Hermitian) given that A and B are Hermitian. Does the converse
hold? HINT: Find a trivial counterexample to the converse.
g) (U + V )† = (U + V )−1 (i.e., U + V is unitary) given that U and V are unitary—that is, prove
this relation if it’s indeed true—if it’s not true, prove that it’s not true. HINT: Find a simple
counterexample: e.g., two 2 × 2 unit matrices.
SUGGESTED ANSWER:
a) Behold:
T T
(AB)T ij = (AB)ji = Ajk Bki = Bki Ajk = Bik Akj = B T AT ij ,

where we have used the Einstein rule of summation on repeated indices. Since the matrix
element is general this completes the proof.
b) Recall that to complex conjugate a matrix is to change its elements to their complex conjugates.
Note that
[(AB)ij ]∗ = (Aik Bkj )∗ = A∗ik Bkj
∗
,
and so (AB)∗ = A∗ B ∗ . Thus, complex conjugating the part (a) identity gives
∗ ∗ ∗
(AB)T = B T AT
or
(AB)† = B † A† ,
which completes the proof. Note that the effect of the complex conjugation and transpose
operations is independent of order.
c) Note
(AB)−1 (AB) = 1
14
where 1 is the unit matix. Thus
(AB)−1 A = B −1 and (AB)−1 = B −1 A−1 ,
which completes the proof.

d) Behold:
(U V )−1 = V −1 U −1 = V † U † = (U V )† .
e) Behold:
AB = BA = B † A† = (AB)† ,
and so AB is Hermitian given that A and B are commuting Hermitian operators.
Is the converse of the above result true? Say we are given that AB is Hermitian. Then
we know
AB = (AB)† = B † A† ,
but I don’t think that we can make any further very general statements about A and B. The
converse probably doesn’t hold, but to be sure one would have to find a counterexample. Well
how about the trivial counterexample where B = A−1 . We know that AA−1 = 1, where I,
the unit matrix, is trivially Hermitian. But any nonsingular matrix A has an inverse, and they
arn’t all Hermitian. For example
0 2
A=
1 0
is not Hermitian and has an inverse:

0 2 0 1 1 0
1 = .
1 0 2 0 0 1
The converse is false.

f) Behold:
(A + B)† = A† + B † = A + B .
So the “forverse” is true. The converse is not true: i.e., (A + B)† = A + B does not imply A
and B are Hermitian. For example

0 1 0 0
A= and B=
0 0 1 0
are not Hermitian, but their sum clearly is.

g) The suggested counterexample is too trivial we all agree. Instead consider the following
matrices which are both Hermitian and their own inverses, and hence are unitary:

1 0 0 1
and .
0 1 1 0
Their sum
1 1
1 1
is its own Hermitian conjugate, but this Hermitian conjugate is not the inverse:

1 1 1 1 2 2
=
1 1 1 1 2 2
Thus the sum of unitary matrices is not in general unitary.

007 qfull 00510 2 5 0 moderate thinking: commuting operations

Extra keywords: (Gr-84)
15
14. There are 4 simple operations that can be done to a matrix: inversing, (−1), complex conjugating (∗),
transposing (T ), and Hermitian conjugating (†). Prove that all these operations mutually commute. Do
this systematically: there are
4 4!
= =6
2 2!(4 − 2)!
combinations of the 2 operations. We assume the matrices have inverses for the proofs involving them.
SUGGESTED ANSWER: We will do the proofs in nearly wordless sequences for brevity and
clarity.
i) For −1 and ∗:
1 = AA−1 ,
1 = (AA−1 )∗ ,
1 = A∗ (A−1 )∗ ,
(A∗ )−1 = (A−1 )∗ .
Note 1 is the unit matrix. We have used the facts that
(z1 + z2 )∗ = [(x1 + x2 ) + i(y1 + y2 )]∗

= (x1 + x2 ) − i(y1 + y2 )
= z1∗ + z2∗
and
(z1 z2 )∗ = [(x1 + iy1 )(x2 + iy2 )]∗

= [x1 x2 − y1 y2 + i(x1 y2 + x2 y1 )]∗
= x1 x2 − y1 y2 − i(x1 y2 + x2 y1 )
= (x1 − iy1 )((x2 − iy2 )
= z1∗ z2∗ .
ii) For −1 and T :
1 = AA−1 ,
1 = (AA−1 )T ,
1 = (A−1 )T AT ,
(AT )−1 = (A−1 )T .
Here we have used result

(AB)T = B T AT
which can be proved as follows:
T T
(AB)T ik = (AB)ki = Akj Bji = Bij Ajk = B T AT ik ,

where we have used Einstein summation.

iii) For −1 and †:
1 = AA−1 ,
1 = (AA−1 )† ,
1 = (A−1 )† A† ,
(A† )−1 = (A−1 )† .
Here we have used result

(AB)† = B † A†
16
which can be proved as follows:

† †
(AB)† ik = [(AB)∗ ]ki = [A∗ B ∗ ]ki = A∗kj Bji
∗
Ajk = B † A† ik ,

= Bij
where we have used Einstein summation. Note we havn’t used the commuting property of
transposing and complex conjugating in this proof, but we have supposed that Hermitian
conjugation applies transposition first and then complex conjugation. Since we prove
transposition and complex conjugation commute in in part (iv) just below there is no concern.
iv) For ∗ and T
∗
(A∗ )T = AT
by inspection. But if you really need a tedious proof, then behold:
h ∗ i h i∗ h
T
i
AT = AT ij
= (Aji )∗ = (A∗ )ji = (A∗ ) .
ij ij
Note ∗
(A∗ )T = AT = A† .
v) For ∗ and †:
∗ ∗
(A∗ )† = (A∗ )T = A† ,

where we have used the fact that transposing and complex conjugating commute, and thus
that Hermitian conjugation does not care which order they are applied in.
vi) For T and †:
† h ∗ iT T
AT = AT = A† ,
where we have used the fact that transposing and complex conjugating commute, and thus
that Hermitian conjugation does not care which order they are applied in.
007 qfull 00700 2 5 0 moderate thinking: square-integrable, inner product

Extra keywords: no analog Griffiths’s problem, but discussion Gr-95–6, Gr2005-94–95
15. If f (x) and g(x) are square-integrable complex functions, then the inner product
Z ∞
hf |gi = f ∗ g dx
−∞
exists: i.e., is convergent to a finite value. In other words, that f (x) are g(x) are square-integrable is
sufficient for the inner product’s existence.
a) Prove the statement for the case where f (x) and g(x) are real functions. HINT: In doing this it
helps to define a function

f (x) where |f (x)| ≥ |g(x)| (which we call the f region);
h(x) =
g(x) where |f (x)| < |g(x)| (which we call the g region),
and show that it must be square-integrable. Then “squeeze” hf |gi.

b) Now prove the statement for complex f (x) and g(x). HINTS: Rewrite the functions in terms of
their real and imaginary parts: i.e.,
f (x) = fRe (x) + ifIm (x)

and
g(x) = gRe (x) + igIm (x) .
Now expand Z ∞
hf |gi = f ∗ g dx
−∞
17
in the terms of the new real and imaginary parts and reduce the problem to the part (a) problem.
c) Now for the easy part. Prove the converse of the statement is false. HINT: Find some trivial
counterexample.
d) Now another easy part. Say you have a vector space of functions {fi } with inner product defined
by Z ∞
fj∗ fk dx .
−∞
Prove the following two statements are equivalent: 1) the inner product property holds; 2) the
functions are square-integrable.
SUGGESTED ANSWER:
a) Define
f (x) where |f (x)| ≥ |g(x)| (which we call the f region);
h(x) =
g(x) where |f (x)| < |g(x)| (which we call the g region).
For the first case, we that
h2 = f 2 = |f |2 ≥ |f ||g|
and for the second that
h2 = g 2 = |g|2 ≥ |f ||g|
likewise. Thus in all cases
h2 ≥ |f ||g| .
Clearly now,
Z ∞ Z Z Z ∞ Z ∞
2 2 2 2
h dx = f dx + g dx ≤ f dx + g 2 dx .
−∞ f region g region −∞ −∞
Thus, h(x) is square-integrable. Since for all x
h(x)2 ≥ f (x)g(x) ≥ −h(x)2 ,
it follows that Z ∞ Z ∞ Z ∞
h2 dx ≥ f g dx ≥ − h2 dx .
−∞ −∞ −∞
Thus the inner product exists: QED.

b) We first break the functions into real and imaginary parts
f (x) = fRe (x) + ifIm (x)

and
g(x) = gRe (x) + igIm (x) .
Now
Z ∞ Z ∞ Z ∞
f ∗ f dx = 2
fRe dx + 2
fIm dx
−∞ −∞ −∞
and Z ∞ Z ∞ Z ∞
∗ 2 2
g g dx = gRe dx + gIm dx ,
−∞ −∞ −∞
and so the part functions are also square-integrable. And now

Z ∞ Z ∞ Z ∞ Z ∞ Z ∞
∗ ∗ ∗ ∗ ∗
f g dx = fRe gRe dx + fIm gIm dx − i fIm gRe dx + i fRe gIm dx
−∞ −∞ −∞ −∞ −∞
Z ∞ Z ∞ Z ∞ Z ∞
= fRe gRe dx + fIm gIm dx − i fIm gRe dx + i fRe gIm dx .
−∞ −∞ −∞ −∞
18
All the inner product integrals on the right-hand side of this equation just involve pure real
functions that are all square-integrable, and so they all exist by the part (a) answer. Thus the
inner product Z ∞
f ∗ g dx
−∞
exists: QED.
c) Let f (x) = x2 and g(x) = 0. Clearly,
Z ∞
f ∗ g dx = 0
−∞
exists, but Z ∞ Z ∞
∗
f f dx = 2 x2 dx
−∞ 0
does not. This counterexample proves the converse is false.

I confess, I never saw the above case myself. It seems so trivial. I was thinking of cases
where neither function is just zero. If neither is just zero, then one might guess that they could
both be normalizable wave functions which must not be just zero everywhere. So a case where
the inner product exists for non-zero functions, but one of the functions is not normalizable
proves that the inner product existing for non-zero functions not guarantee normalizability.
For examample, let f (x) = |x| and g(x) = (1/2)e−|x|. Clearly,
Z ∞ Z ∞ Z ∞
∗ −x
f g dx = xe dx = −xe−x |∞
0 + e−x dx = 1 .
−∞ 0 0
f ∗ f dx = 2 x2 dx
−∞ 0
does not. This counterexample proves the converse is false.

2
Another example is when f (x) = x2 and g(x) = e−x . Clearly after using a table of
integrals (Hod-313), Z ∞ √
∗ π
f g dx =
−∞ 2
f ∗ f dx = x4 dx
−∞ −∞
does not. This counterexample proves the converse is false too or even also.
2
Still another goody is when f (x) = 1/x and g(x) = xe−x . Clearly after using a table of
integrals (Hod-313), Z ∞
√
f ∗ g dx = π
−∞
f ∗ f dx = x−2 dx
−∞ −∞
does not.
d) Say we have the vector space of functions {fi (x)} for which statement 1 holds: i.e., the inner
product property holds: i.e., Z ∞
fj∗ fk dx
−∞
exists for all functions in the vector space. The inner product also exists for j = k, and so the
functions are square-integrable: statement 2 follows from statement 1. If statement 2 holds
(i.e., the functions are all square-integrable), then it follows from the part (b) answer proof
that the inner product property holds: viz., statement 1 follows from statement 2. Thus the
two statements are equivalent.
19
You could put the statements in an if-and-only-if statement or necessary-and-sufficient

statement. But I find those kind of statements hard to disentangle. I prefer saying the
statments are equivalent: i.e., each one implies the other.
007 qfull 01000 2 3 0 moderate math: reduced SHO operator, Hermiticity

Extra keywords: (Gr-99:3.28), dimensionless simple harmonic oscillator Hamiltonian
16. Consider the operator
d2
Q = − 2 + x2 .
dx
2
a) Show that f (x) = e−x /2
is an eigenfunction of Q and determine its eigenvalue.
b) Under what conditions, if any, is Q a Hermitian operator? HINTS: Recall
hg|Q† |f i∗ = hf |Q|gi
is the defining relation for the Hermitian conjugate Q† of operator Q. You will have to write
the matrix element hf |Q|gi in the position representation and use integration by parts to find the
conditions.
SUGGESTED ANSWER:
a) Behold:
d2

d
2
Qf = − 2 + x f = − (−xf ) + x2 f = −x2 f + f + x2 f = f .
dx dx
Thus f (x) is an eigenfunction and the eigenvalue is 1. By the by, Q is a dimensionless version
of the simple harmonic oscillator Hamiltonian (see, e.g., Gr-37).
b) Behold:
hg|Q† |f i∗ = hf |Q|gi
Z b
= f ∗ Qg dx
a
Z b 2

∗d g ∗ 2
= −f + f x g dx
a dx2
b Z b 2 ∗ Z b
df ∗

dg d f
= −f ∗ gx2 f ∗ dx

+ g − g 2 dx +
dx dx a a dx a
b Z b !∗
∗

dg df
= −f ∗ g ∗ Qf dx

+ g +
dx dx a a
∗
b
dg df
= −f ∗ g + hg|Q|f i∗ ,

+
dx dx a
where we require f and g to be general square-integrable functions over the range [a, b], we have
used the general definition of the Hermitian conjugate, and we have used integration by parts
twice. Clearly Q is Hermitian if the boundary condition term vanishes since then Q† = Q.
For instance, if the boundaries were at minus and plus infinity, then the boundary condition
term would have to vanish since the functions are square-integrable and so must be zero at
minus and plus infinity. Or if periodic conditions were imposed on the functions, the boundary
condition term would have to vanish again.
007 qfull 01100 2 5 0 moderate thinking: Hilbert space problems

17. Do the following.
20
a) Show explicitly that any linear combination of two functions in the Hilbert space L2 (a, b) is also in
L2 (a, b). (By explicitly, I mean don’t just refer to the definition of a vector space which, of course
requires the sum of any two vectors to be a vector.)
b) For what values of real number s is f (x) = |x|s in L2 (−a, a)
c) Show that f (x) = e−|x| is in L2 = L2 (−∞, ∞). Find the wavenumber space representation of f (x):
recall the wavenumber “orthonormal” basis states in the position representation are
eikx
hx|ki = √ .
2π
SUGGESTED ANSWER:
a) Given f and g in L2 (a, b),
Z b
hf + g|f + gi = (f + g)∗ (f + g) dx
a
Z b Z b Z b Z b
= |f |2 dx + |g|2 dx + f ∗ g dx + g ∗ f dx
a a a a
= hf |f i + hg|gi + hf |gi + hg|f i .
The inner products hf |f i and hg|gi exist by hypothesis. And their existence by theorem (see
Gr-95–96) verifies that hf |gi and hg|f i exist. Thus hf + g|f + gi exists and f + g are in the
Hilbert space.
b) Behold:  2s+1 a
x
for s 6= −1/2;


Z a  2s + 1

hf |f i = 2 x2s dx = 2 × a 0
0 
 ln(x) for s = −1/2.


0
Clearly the integral only exists for s > −1/2. It’s value when it exists is 2a2s+1 /(2s + 1).
c) Behold: Z ∞
hf |f i = 2 e−2x dx = 1 .
0
Thus hf |f i exists and f is in L2 .

Behold:
Z ∞
|f i = dx |xihx|f i ,
−∞
Z ∞
hk|f i = dx hk|xihx|f i ,
−∞
Z ∞ Z ∞
e−ikx e−ikx
f (k) = dx √ f (x) = dx √ e−|x|
−∞ 2π −∞ 2π
r Z ∞ Z ∞
1 −(−ik+1)x −(ik+1)x
= dx e + dx e
2π 0 0
r ∞ ∞
1 e−(−ik+1)x e−(ik+1)x

= +
2π −(−ik + 1) 0 −(ik + 1) 0
r
2 1
= .
π 1 + k2
007 qfull 01200 2 5 0 moderate thinking: Hermitian conjugate of AB

21
18. Some general operator and vector identities should be proven. Recall the definition of the Hermitian
conjugate of general operator Q is giveny by
hα|Q|βi = hβ|Q† |αi∗ ,
where |αi and |βi are general vectors.

a) Prove that the bra corresponding to ket vector Q|βi is hβ|Q† for general Q and |βi. HINT:
Consider general vector |αi and the inner product
hα|Q|βi
and work your way by valid steps to

hβ|Q† |αi∗
and that completes the proof since
hα|γi = hγ|αi∗
for general vectors |αi and |γi.
b) Show that the Hermitian conjugate of a scalar c is just its complex conjugate.
c) Prove for operators, not matrices, that
(AB)† = B † A† .
The result is, of course, consistent with matrix representations of these operators. But there
are representations in which the operators are not matrices: e.g., the momentum operator in the
position representation is differentiating operator
−
h ∂
p= .
i ∂x
Our proof holds for such operators too since we’ve done the proof in the general operator-vector
formalism.
d) Generalize the proof in part (c) for an operator product of any number.
e) Prove that (A + B)† = A† + B † .
f) Prove that c[A, B] is a Hermitian operator for Hermitian A and B only when c is pure imaginary
constant.
SUGGESTED ANSWER:
a) Behold:
hα|Q|βi == hβ|Q† |αi∗
by the definition of the Hermitian conjugate. And that completes the proof.
b) Let |αi, |βi and and c be general. Just using the vector rules that we have
hα|c† |βi = hβ|c|αi∗ = c∗ hβ|αi∗ = c∗ hα|βi = hα|c∗ |βi .
Thus c† = c∗ .
c) For general |αi and |βi,
hβ|(AB)† |αi = hα|AB|βi∗ = [(hα|A)(B|βi)]∗ = hβ|B † A† |αi
where we have used the definition of the Hermitian conjugate operator and the fact that the
bra corresponding to Q|γi for any operator Q and ket |γi is hγ|Q† . Since |αi and |βi are
general, it follows that
(AB)† = B † A† .
22
d) Behold:
(ABCD . . .)† = (BCD . . .)† A† = (CD . . .)† B † A† = . . . D† C † B † A† .
e) Behold:
hα|(A + B)† |βi = hβ|(A + B)|αi∗ = hβ|A|αi∗ + hβ|B|αi∗ = hα|A† |βi + hα|B † |βi ,
and given that |αi and |βi are general, it follows that
(A + B)† = A† + B † .
Thus, the Hermitian conjugate operation is distributive over addition.

f) Behold:
(c[A, B])† = (cAB)† − (cBA)† = B † A† c† − A† B † c† = BAc∗ − ABc∗ = −c∗ [A, B] .
If c∗ = −c, then c is pure imaginary and
(c[A, B])† = c[A, B] .
Thus (c[A, B]) is Hermitian only for c pure imaginary.

007 qfull 01400 2 5 0 moderate thinking: bra ket projector completeness

Extra keywords: (Gr-118:3.57) See also CT-115, 138
19. For an inner product vector space there is some rule for calculating the inner product of two general
vectors: an inner product being a complex scalar. If |αi and |βi are general vectors, then their inner
product is denoted by
hα|βi ,
where in general the order is significant. Obviously different rules can be imagined for a vector space
which would lead to different values for the inner products. But the rule must have three basic properties:
(1) hβ|αi = hα|βi∗ ,

(2) hα|αi ≥ 0 , where hα|αi = 0 if and only if |αi = |0i,
and
(3) hα| b|βi + c|γi = bhα|βi + chα|γi ,
where |αi, |βi, and |γi are general vectors of the vector space and b and c are general complex scalars.
There are some immediate corollaries of the properties. First, if hα|βi is pure real, then
hβ|αi = hα|βi .
Second, if hα|βi is pure imaginary, then
hβ|αi = −hα|βi .
Third, if
|δi = b|βi + c|γi ,
then
hδ|αi∗ = hα|δi = bhα|βi + chα|γi
which implies
hδ|αi = b∗ hβ|αi + c∗ hγ|αi .
This last result makes
hβ|b + hγ|c |αi = b∗ hβ|αi + c∗ hγ|αi
∗ ∗
23
a meaningful expression. The 3rd rule for a vector product inner space and last corollary together mean
that the distribution of inner product multiplication over addition happens in the normal way one is
used to.
Dirac had the happy idea of defining dual space vectors with the notation hα| for the dual vector
of |αi: hα| being called the bra vector or bra corresponding to |αi, the ket vector or ket: “bra” and
“ket” coming from “bracket.” Mathematically, the bra hα| is a linear function of the vectors. It has the
property of acting on a general vector |βi and yielding a complex scalar: the scalar being exactly the
inner product hα|βi.
One immediate consequence of the bra definition can be drawn. Let |αi, |βi, and a be general and
let
|α′ i = a|αi .
Then
hα′ |βi = hβ|α′ i∗ = a∗ hβ|αi∗ = a∗ hα|βi
implies that the bra corresponding to |α′ i is given by
hα′ | = a∗ hα| = hα|a∗ .
The use of bra vectors is perhaps unnecessary, but they do allow some operations and properties
of inner product vector spaces to be written compactly and intelligibly. Let’s consider a few nice uses.
a) The projection operator or projector on to unit vector |ei is defined by
Pop = |eihe| .
This operator has the property of changing a vector into a new vector that is |ei times a scalar. It
is perfectly reasonable to call this new vector the component of the original vector in the direction
of |ei: this definition of component agrees with our 3-dimensional Euclidean definition of a vector
component, and so is a sensible generalization of that the 3-dimensional Euclidean definition. This
generalized component would also be the contribution of a basis of which |ei is a member to the
expansion of the original vector: again the usage of the word component is entirely reasonable. In
symbols
Pop |αi = |eihe|αi = a|ei ,
where a = he|αi.
2 n
Show that Pop = Pop , and then that Pop = Pop , where n is any integer greater than or equal
to 1. HINTS: Write out the operators explicitly and remember |ei is a unit vector.
b) Say we have
Pop |αi = a|αi ,
where Pop = |eihe| is the projection operator on unit vector |ei and |αi is unknown non-null vector.
Solve for the TWO solutions for a. Then solve for the |αi vectors corresponding to these solutions.
HINTS: Act on both sides of the equation with he| to find an equation for one a value. This
equation won’t yield the 2nd a value—and that’s the hint for finding the 2nd a value. Substitute
the a values back into the original equation to determine the corresponding |αi vectors. Note one
a value has a vast degeneracy in general: i.e., many vectors satisfy the original equation with that
a value.
c) The Hermitian conjugate of an operator Q is written Q† . The definition of Q† is given by the
expression
hβ|Q† |αi = hα|Q|βi∗ ,
where |αi and |βi are general vectors. Prove that the bra corresponding to ket Q|βi must hβ|Q† for
general |αi. HINTS: Let |β ′ i = Q|βi and substitute this for Q|βi in the defining equation of the
Hermitian conjugate operator. Note operators are not matrices (although they can be represented
as matrices in particular bases), and so you are not free to use purely matrix concepts: in particular
the concepts of tranpose and complex conjugation of operators are not generally meaningful.
24
d) Say we define a particular operator Q by
Q = |φihψ| ,
where |φi and |ψi are general vectors. Solve for Q† . Under what condition is
Q† = Q ?
When an operator equals its Hermitian conjugate, the operator is called Hermitian just as in the
case of matrices.
e) Say {|ei i} is an orthonormal basis. Show that
|ei ihei | = 1 ,
where we have used Einstein summation and 1 is the unit operator. HINT: Expand a general
vector |αi in the basis.
SUGGESTED ANSWER:
a) Behold:
2
Pop = |eihe|eihe| = |eihe| = Pop ,
where we have used the normalization of |ei. Now
n
Pop = |eihe|ein−1 he| = |eihe| = Pop ,
n
where we have used the normalization of |ei. I would actually accept that Pop = Pop is true
2
by inspection given that Pop = Pop is true.
b) Well
he|Pop |αi = ahe|αi
leads to
he|αi = ahe|αi .
If he|αi =
6 0, a = 1. If he|αi = 0, a = 0 in order for
Pop |αi = |eihe|αi = 0 = a|αi
to hold remembering that |αi is not null. If a = 1, then
|eihe|αi = |αi
or
|αi = |eihe|αi .
Thus |αi points in the |ei direction. If a = 0, then
he|αi = 0 .
Any vector |αi perpendicular to |ei is a solution.

We have, in fact, solved the eigenvalue problem for the Pop operator. The eigenvalues are
1 and 0 and the corresponding normalized eigenvectors are |ei and
|βi − |eihe|βi
,
||(|βi − |eihe|βi)||
where |βi is general non-null vector. This set of eigenvectors is actually complete. General
vector |βi can be expanded so
|βi = |eihe|βi + (|βi − |eihe|βi) .

25
Act on this vector with |ei and one gets he|βi. Act on it with any |e⊥ i orthogonal to |ei and
one gets he⊥ |βi
c) Well
hβ|Q† |αi = hα|β ′ i∗ = hβ ′ |ai .
Since a is general we must interpret
hβ ′ | = hβ|Q† .
d) Well
hβ|Q† |αi = hα|Q|βi∗ = hα|φi∗ hψ|βi∗ = hβ|ψihφ|αi ,
and so
Q† = |ψihφ| .
If |ψi = |φi, then Q† = |φihφ| = Q, and Q would be a Hermitian operator.
e) Since the basis is a basis
|αi = ai |ei i ,
where the coefficients are given by
ai = hei |αi .
Thus
|αi = |ei ihei |αi ,
and so we must identify
|ei ihei | = 1 .
This may seem trivial, but it is a great notational convenience. One can insert the unit operator
1 in anywhere one likes in expressions, and so one can insert |ei ihei | in those places too. This
makes expanding in various bases notationally very simple and obvious. The operator |ei ihei |
can be considered the projection operator onto a basis.
008 qmult 00090 1 4 5 easy deducto-memory: example Hilbert space

20. “Let’s play Jeopardy! For $100, the answer is: A space of all square-integrable functions on the x interval
(a, b).”
What is a , Alex?
a) non-inner product vector space b) non-vector space c) Dilbert space
d) Dogbert space e) Hilbert space
Wrong answers:
c) Now if this is late in the day: Dilbert was this . . .
008 qmult 00100 1 1 3 easy memory: complex conjugate of scalar product

21. The scalar product hf |gi∗ in general equals:
a) hf |gi. b) ihf |gi. c) hg|f i. d) hf |i|gi. e) hf |(−i)|gi.
SUGGESTED ANSWER: (c)
Wrong Answers:
a) This is true only if hf |gi∗ is pure real.
008 qmult 00200 1 4 3 easy deducto-memory: what operators do

22. “Let’s play Jeopardy! For $100, the answer is: It changes a vector into another vector.”
26
What is a/an , Alex?

a) wave function b) scalar product c) operator d) bra e) telephone operator
Wrong answers:
e) The adjective renders this incorrect.
008 qmult 00300 2 1 5 moderate memory: Hermitian conjugate of product

23. Given general operators A and B, (AB)† equals:
a) AB. b) A† B † . c) A. d) B. e) B † A† .
SUGGESTED ANSWER: (e) Using the operator definition of the Hermitian conjugate, the
proof is
hΨ|(AB)† |χi = hχ|AB|Ψi∗ = hΨ|B † A† |χi .
Recall the bra corresponding to Q|γi is hγ|Q† in general.
Wrong Answers:
a) Only if AB is Hermitian.
b) Only if A† and B † commute.
c) Only in some special case: most easily if A Hermitian and B † = 1.
d) Only in some special case: most easily if B Hermitian and A† = 1.
008 qmult 00400 2 5 5 moderate thinking: general Hermitian conjugation

24. The Hermitian conjugate of the operator λ|φihχ|ψihℓ|A (with λ a scalar and A an operator) is:
a) λ|φihχ|ψihℓ|A. b) λ|φihχ|ψihℓ|A† . c) A|ℓihψ|χihφ|λ∗ . d) A|ℓihψ|χihφ|λ.
e) A† |ℓihψ|χihφ|λ∗ .
SUGGESTED ANSWER: (e) Recall
(AB)† = B † A† .
This generalizes immediately to
(ABC . . .)† = (BC . . .)† A† = (C . . .)† A† B † = . . . C † B † A† .
Now the Hermitian conjugate of scalars λ and hχ|ψi are λ∗ and hψ|χi, respectively. For proof
consider scalar c and general vectors |αi and |βi:
hα|c† |βi = hβ|c|αi∗ = c∗ hβ|αi∗ = c∗ hα|βi = hα|c∗ |βi ,
and so c† = c∗ . The Hermitian conjugate of operator |φihℓ| is |ℓihψ|. For a proof consider the
general vectors |αi and |βi:
hα|φihℓ|βi = hφ|αi∗ hβ|ℓi∗ = hβ|ℓi∗ hφ|αi∗ = (hβ|ℓihφ|αi)∗ ,
and that completes the proof. Now put all the above results together and answer (e) follows.
Remeber scalars can be commuted freely.
Wrong Answers:
a) This is just the same thing all over again.
008 qmult 00500 1 1 5 easy memory: compatible observables

25. Compatible observables:
a) anticommute. b) are warm and cuddly with each other. c) have no hair. d) have no
complete simultaneous orthonormal basis. e) commute.
27

Wrong Answers:
d) Actually they are guaranteed to have a complete simultaneous basis which can (if it isn’t
already) be orthonormalized.
008 qmult 00600 1 1 3 easy memory: parity operator

26. The parity operator Π acting on f (x) gives:
df /dx. b) 1/f (x). c) f (−x). d) 0. e) a spherical harmonic.
Wrong Answers:
e) Whoa, fella.
008 qmult 00700 1 4 3 easy deducto-memory: braket expectation value

27. Given the position representation for an expectation value
Z ∞
hQi = Ψ(x)∗ QΨ(x) dx ,
−∞
what is the braket representation?

a) hQ|Ψ∗ |Qi. b) hΨ∗ |Q|Ψi. c) hΨ|Q|Ψi. d) hΨ|Q† |Ψi. e) hQ|Ψ|Qi.
Wrong Answers:
d) Only true if Q is Hermitian.
008 qmult 00800 1 4 3 easy deducto-memory: Hermitian eigenproblem

28. What are the three main properties of the solutions to a Hermitian operator eigenproblem?
a) (i) The eigenvalues are pure IMAGINARY. (ii) The eigenvectors are guaranteed orthogonal,
except for those governed by degenerate eigenvalues and these can always be orthogonalized.
(iii) The eigenvectors DO NOT span all space.
b) (i) The eigenvalues are pure REAL. (ii) The eigenvectors are guaranteed orthogonal, except
for those governed by degenerate eigenvalues and these can always be orthogonalized. (iii) The
eigenvectors span all space in ALL cases.
c) (i) The eigenvalues are pure REAL. (ii) The eigenvectors are guaranteed orthogonal, except
for those governed by degenerate eigenvalues and these can always be orthogonalized. (iii) The
eigenvectors span all space for ALL FINITE-DIMENSIONAL spaces. In infinite dimensional
cases they may or may not span all space. It is quantum mechanics postulate that the eigenvectors
of an observable (which is a Hermitian operator) span all space.
d) (i) The eigenvalues are pure IMAGINARY. (ii) The eigenvectors are guaranteed orthogonal,
except for those governed by degenerate eigenvalues and these can always be orthogonalized.
(iii) The eigenvectors span all space in ALL FINITE-DIMENSIONAL spaces. In infinite
dimensional cases they may or may not span all space.
e) (i) The eigenvalues are pure REAL. (ii) The eigenvectors are guaranteed orthogonal, except for
those governed by degenerate eigenvalues and these can always be orthogonalized.

Wrong Answers:
b) I once thought this was true, but Griffiths (Gr-93, Gr-99) set me straight.
e) Just checking to see if you can count.
28
008 qmult 00900 1 4 5 easy deducto-memory: definition observable

29. “Let’s play Jeopardy! For $100, the answer is: A physically significant Hermitian operator possessing a
complete set of eigenvectors.”
What is a/an , Alex?
a) conjugate b) bra c) ket d) inobservable e) observable
Wrong answers:
d) Exactly wrong.
008 qmult 01000 1 4 4 easy deducto-memory: time-energy inequality

30. In the precisely-formulated time-energy inequality, the ∆t is:
a) the standard deviation of time.
b) the standard deviation of energy.
c) a Hermitian operator.
d) the characteristic time for an observable’s value to change by one standard deviation.
e) the characteristic time for the system to do nothing.
SUGGESTED ANSWER: (d) Characteristic time is often defined precisely as the time for a
change if the rate of change is assumed to be constant. This is the usage in the precisely-formulated
time-energy inequality. Note that the longest-answer-is-right rule is obeyed here.
Wrong Answers: itemitema) Nah. That’s what it looks like since by convention it’s written with
the same symbol used for often standard deviation.
b) Nah.
c) It’s just a number.
e) Seems a bit vague.
008 qmult 02000 1 1 5 easy memory: common eigensets

Extra keywords: See CT-140ff
31. The statements “two observables commute” and “a common eigenset can be constructed for two
observables” are
in flat contradiction. b) unrelated. c) in non-intersecting Venn diagrams.
d) irrelevant in relation to each other. e) are equivalent in the sense that one implies the other.
SUGGESTED ANSWER: (e) Another triumph for the longest answer is right rule.
Wrong Answers:
a) Exactly wrong.
c) I’m not sure what this means.
008 qfull 00030 2 5 0 moderate thinking: Hermiticity and expectation values

32. Recall the definition of Hermitian conjugate for a general operator Q is
hα|Q|βi = hβ|Q† |αi∗ ,
where |αi and |βi are general vectors. If Q is Hermitian,
Q† = Q :
i.e., Q is its own Hermitian conjugate.

29
a) If Q is Hermitian, prove that the expectation value of a general vector |γi,
hγ|Q|γi ,
is pure real.
b) If the expectation value
hγ|Q|γi
is always pure real for general |γi, prove that Q is Hermitian. The statement to be proven is the converse
of the statement in part (a). HINT: First show that
hγ|Q|γi = hγ|Q† |γi .
Then let |αi and |βi be general vectors and construct a vector |ξi = |αi + c|βi, where c is a general
complex scalar. Note that the bra corresponding to c|βi is c∗ hβ|. Expand both sides of
hξ|Q|ξi = hξ|Q† |ξi ,
and then keep simplifying both sides making use of the first thing proven and the definition of a Hermitian
conjugate. It may be useful to note that
(A† )† = A and (A + B)† = A† + B † ,
where A and B are general operators and You should be able to construct an expression where choosing
c = 1 and then c = i requires Q = Q† .
c) What simple statement follows from the proofs in parts (a) and (b)?
SUGGESTED ANSWER:
a) Well
hγ|Q|γi = hγ|Q† |γi∗
by the general definition of Hermitian conjugation. But since Q† = Q, we have
hγ|Q|γi = hγ|Q|γi∗ .
Since the expection value is equal to its own complex conjugate, it is pure real: i.e., hγ|Q|γi is
pure real.
b) First note for general |γi that
hγ|Q|γi = hγ|Q† |γi∗ = hγ|Q† |γi ,
where we have used the definition of the Hermitian conjugate and then the fact that by
assumption hγ|Q|γi is pure real which imples that hγ|Q† |γi∗ is pure real too.
For general vectors |αi and |βi, we define
|ξi = |αi + c|βi ,
where c is a general complex number The point of this definition is to exploit our statement
above for the general vector |γi (which will hold for |αi, |βi, and |ξi) and then the generality
of |αi and |βi to get a statement like
hα|Q|βi = hα|Q† |βi
which verifies Q† = Q—but we havn’t got this statement yet. Why do we need the general
complex constant c? Well if you went through the steps without it, you would reach a point
where you needed and then you’d go back and put it in.
Starting from
hξ|Q|ξi = hξ|Q† |ξi ,
30
we proceed as follows
hξ|Q|ξi = hξ|Q† |ξi

hα|Q|αi + |c| hβ|Q|βi + chα|Q|βi + c hβ|Q|αi = hα|Q† |αi + |c|2 hβ|Q† |βi + chα|Q† |βi + c∗ hβ|Q† |αi
2 ∗
chα|Q|βi + c∗ hβ|Q|αi = chα|Q† |βi + c∗ hβ|Q† |αi

0 = chα|(Q − Q† )|βi + c∗ hβ|(Q − Q† )|αi
0 = chα|(Q − Q† )|βi + c∗ hα|(Q − Q† )† |βi∗
0 = chα|(Q − Q† )|βi − c∗ hα|(Q − Q† )|βi∗ . (1)
Equation (1) must be true for all choices of c. If c = 1, then the real parts of equation (1)
cancel identically, but imaginary parts give
Im[hα|(Q − Q† )|βi] = 0 .
If c = i, then the imaginary parts of equation (1) cancel identically since
i2 − (−i)(−i) = i2 − i2 = 0
and we find that the real parts give
Re[hα|(Q − Q† )|βi] = 0 .
Thus, we obtain
hα|(Q − Q† )|βi = 0
or
hα|Q|βi = hα|Q† |βi .
Remember that |αi and |βi are general, and thus
Q† = Q .
So Q must be Hermitian if all expection values of Q are pure real.

We could actually do one more step to reconfirm the Hermiticity of Q. Using the definition
of a Hermitian conjugate, we find
hα|Q|βi = hα|Q† |βi = hβ|(Q† )† |αi∗ = hβ|Q|αi∗
or
hα|Q|βi = hβ|Q|αi∗
which implies
Q† = Q .
I think that the above proof may be the simplest proof available.
Just to satisfy paranoia, we might prove some of the results we used in the above proof.
First, that c∗ hβ is the bra corresponding to the ket c|βi where c is a general complex number
and c|βi is a general vector. Consider general vector |αi and note that.
hα|c|βi = chα|βi = chβ|αi∗ = (c∗ hβ|αi)∗ = hβ|c∗ |αi∗
Clearly, c∗ hβ is the bra corresponding to the ket c|βi.

Second, that (Q† )† = Q. Consider general vectors |αi and |βi. Now using the definition
of the a Hermitian conjugate twice in a row
hα|Q|βi = hβ|Q† |αi∗ = hα|(Q† )† |βi ,
and so
(Q† )† = Q
QED.
31
Third, that (A + B)† = A† + B † . Consider general vectors |αi and |βi. The proof is
hα|(A + B)† |βi∗ = hβ|(A + B)|αi = hβ|A|αi + hβ|B|αi

= hα|A† |βi∗ + hα|B † |βi∗ = (hα|A† |βi + hα|B † |βi)∗
= hα|(A† + B † )|βi∗ ,
and thus
hα|(A + B)† |βi = hα|(A† + B † )|βi ,
and thus again
(A + B)† = A† + B †
QED.
c) The above two proofs show that the statements of parts (a) and (b) are equivalent: each implies
the other. We could phrase the situation with the sentence “if and only if Q is Hermitian is
hγ|Q|γi pure real for general |γi.” Or we could say “it is necessary and sufficient that Q be
Hermitian for hγ|Q|γi to be pure real for general |γi.” These ways of saying things always
seem asymmetric and obscure to me. I prefer to say that the statements of parts (a) and (b)
are equivalent.
008 qfull 00040 2 3 0 moderate math: solving an eigenproblem

Extra keywords: (Gr-94:3.22) also diagonalizing a matrix.
33. Consider
1 1−i
Q= .
1+i 0
In this problem, we will diagonalize this matrix: i.e., solve for its eigenvalues and eigenvectors. We also
actually explictly find the diagonal form—which is not usually necessary.
a) Is Q Hermitian?
b) Solve for the eigenvalues. Are they real?
c) Determine the normalized eigenvectors â. Since eigenvectors are not unique to within a phase factor,
the marker insists that you arrange your eigenvectors so that the first component of each is 1. Are
the eigenvectors orthogonal? HINT: The matrix equation for the eigenvectors is a homogeneous
matrix equation with non-trivial solutions (i.e., solutions that are not just zeros) for the eigenvalues
since the determinant of Q − λI vanishes for those eigenvalues. However, 1 equation obtained from
a N × N homogeneous matrix problem is always not independent: there are only N − 1 independent
equations and one can only solve for N − 1 components of the eigenvectors. So if you set the first
component of the solution vector to be 1, the N − 1 equations allow you to solve for the other
components. This solution vector is a valid solution vector, but its overall scale is arbitrary. There
is no determined scale for the eigenvectors of a homogeneous matrix problem: e.g., k times solution
vector ~a is also a solution. But, in quantum mechanics, physical vectors should be normalized and
the normalization constraint provides an N th independent equation, and thus allows a complete
solution of the eigenvectors to within a global phase factor. Normalization doesn’t set that global
phase factor since it cancels out in the normalization equation. The global phase factor can be
chosen arbitrarily for convenience. The global phase factor of a state no effect on the physics of the
state.
d) Obtaining the eigenvalues and eigenvectors is usually all that is meant by diagonalization, but one
can actually transform the eigenvalue matrix equation into a matrix equation where the matrix is
diagonal and the eigenvectors can be solved for by inspection. One component of an eigenvector
is 1 and the other components are zero. How does one transform to diagonal form? Consider our
matrix equation
Qâ = λâ .
Multipy both sides by the transformation matrix U to obtain
U Qâ = λU â
32
which is obviously the same as

U QU −1 U â = λU â .
If we define
â′ = U â and Q′ = U QU −1 ,
then the transformed matrix equation is just
Q′ â′ = λâ′ .
Prove that the transformation matrix U that gives the diagonalized matrix Q′ just consists of rows
that are the Hermitian conjugates of the eigenvectors. Then find the diagonalized matrix itself and
its eigenvalue.
e) Compare the determinant det|Q|, trace Tr(Q), and eigenvalues of Q to those of Q′ .
f) The matrix U that we considered in part (d) is actually unitary. This means that
U † = U −1 .
Satisfy yourself that this is true. Unitary transformations have the useful property that inner
products are invariant under them. If the inner product has a physical meaning and in particular
the magnitude of vector has a physical meaning, unitary transformations can be physically relevant.
In quantum mechanics, the inner product of a normalized state vector with itself 1 and this should
be maintained by all physical transformations, and so such transformations must be unitary. Prove
that
ha′ |b′ i = ha|bi
where
|a′ i = U |ai|b′ i = U |bi
and U is unitary.
SUGGESTED ANSWER:
a) Yes by inspection. But to be explicit
T
† ∗ T 1 1+i 1 1−i
Q = (Q ) = =Q,
1−i 0 1+i 0
where T is the transpose function. Thus, Q = Q† , and so Q is Hermitian.

b) The eigen equation is
(1 − λ)(−λ) − 2 = λ2 − λ − 2 = (λ − 2)(λ + 1) = 0 .
The eigenvalues are clearly

λ = −1 and λ=2.
The eigenvalues are both real as they should be for a Hermitian matrix.
c) The equation for the eigenvector components is
a1 + (1 − i)a2 = λa1
from which we obtain the unnormalized expression
1 1
! !
~a(λ) = λ − 1 = (λ − 1)(1 + i) .
1−i 2
The normalized eigenvectors are
1
r !
1

2 1
â(λ = 2) = 1+i and â(λ = −1) = √ .
3 3 −(1 + i)
2
33
The eigenvectors are orthogonal by inspection.

Note normalized eigenvectors that are physically the same can look quite different because
of global phase factors. For example,
1−i
 
1
r ! r r
2 2  2  (1 + i) 1 1 − i
â(λ = 2) = 1 + i = (1 + i) = √
3 3 1 3 1
 
2
2
2
iπ/4 1 1−i
=e √ .
3 1
The global phase factor eiπ/4 can be dropped because it is physically irrelevant. Physical
eigenvectors are unique only to within a global phase factor. Caveat markor.
d) Following the question lead-in, we define
r 
2 1−i
 3 √
6  .

U =

1 1−i 
√ − √
3 3
Because the eigenvectors are orthonomal, the inverse matrix U −1 constructed using the
eigenvectors for columns. Thus, we obtain.
 r 
2 1
√
U −1 = 
 3 3  .

1+i 1+i 
√ − √
6 3
Obviously,
U U −1 = I ,
where I is the unit matrix. If we multiply both sides from the right by U , we get
U U −1 U = U ,
and so U −1 U = 1 too as it should.

We now find the elements of Q′ . Since
Q′ = U QU −1 ,
we have
−1 −1
Q′ij = Uik Qkℓ Uℓj = Uik λj Ukj = λj δij = λi δij ,
where we have used the eigenproblem property itself, the fact that U U −1 = 1, and Einstein
summation (i.e., there is an implicit sum on repeated indexes. That
−1 −1
Qkℓ Uℓj = λj Ukj
takes a little cogitating on. But note that

−1
Qkℓ Uℓj = Vkj
where the columns of V are the columns of U 1 times the eigenvalue of the eigenvector that
makes up that row.
The proof is complete since the U matrix constructed as required yields a diagonal matrix
whose elements are the eigenvalues.
So Q′ is the required diagonalized matrix. In our case,

′ 2 0
Q = .
0 −1
34
By inspection, the transformed eigenvectors are

1 0
â(λ = 2) = and â(λ = −1) = .
0 1
e) Well
det|Q| = −2 = det|Q′ | , Tr(Q) = 1 = Tr(Q′ ) ,
and the eigenvalues of Q (i.e., 2 and −1) are clearly the eigenvalues of Q′ .
The above results are special canses of more general results. We can prove this.
Say S is a general transformation acting on general matrix A. Then
det|SAS −1 | = det|S| × det|A| × det|S −1 |

= det|A| × det|S| × det|S −1 |
= det|A| × det|SS −1 |
= det|A| ,
where we have used well known determinant properties: see, e.g., Mih-657. We see that the
determinant is generally invariant under a similarity transformation.
Now note that
Tr(ABC . . . Z) = Aij Bjk Cℓm . . . Zni = Bjk Cℓm . . . Zni Aij = Tr(BC . . . ZA) .
Thus trace is invariant under a cycling of the matrix product. Therefore
Tr(SAS −1 ) = Tr(AS −1 S) = Tr(A) .
Thus trace is invariant under a similarity transformation.

The invariance of the eigenvalues holds for Hermitian matrices under unitary similarity
transformations constructed from their eigenvectors. Consider Hermitian matrix H and unitary
matrix U constructed from the eigenvectors of H. The iℓth element of H ′ = U HU −1 is given
by X X X
−1 −1 −1
H′ = Uij Hjk Ukℓ = Uij λℓ Ujℓ = λℓ Uij Ujℓ = λℓ δiℓ ,
jk j j
where we have NOT used Einstein summation and where λℓ is the eigenvalue corresponding
†
to the eigenvector in the ℓth column of Ukℓ . Thus we find that H d is diagonal with diagonal
elements being the eigenvalues of H: these elements are the eigenvalues of H d by inspection.
Are eigenvalues invariant under general similarity transformations or general unitary similarity
transformations. I’d guess not, but we’ve done enough proofs for now.
f) By inspection, the matrix U we found in part (d) is unitary.
Well for general vectors |ai and |bi
ha|U |bi = hb|U † |ai∗
by the definition of the Hermitian conjugate. There in general the bra corresponding to ket
U |bi is hb|U † .
Now for general vectors |ai and |bi, we can write
ha′ |b′ i = ha|U † U |bi = ha|U −1 U |bi = ha|bi ,
where we have used the unitary nature of U (i.e., U † = U −1 ).

008 qfull 00100 2 5 0 moderate thinking: expectation values two ways

35
34. Consider the observable Q and the general NORMALIZED vector |Ψi. By quantum mechanics
postulate, the expectation of Qn , where n ≥ 0 is some integer, for |Ψi is
hQn i = hΨ|Qn |Ψi .
a) Assume Q has a discrete spectrum of eigenvalues qi and orthonormal eigenvectors |qi i. It follows
from the general probabilistic interpretation postulate of quantum mechanics, that expectation
value of Qn for |Ψi is given by X
hQn i = qin |hqi |Ψi|2 .
i
n
|hqi |Ψi|2
P
Show that this expression for hQ i also follows from the one in the preamble. What is i
equal to?
b) Assume Q has a continuous spectrum of eigenvalues q and Dirac-orthonormal eigenvectors |qi.
(Dirac-orthonormal means that hq ′ |qi = δ(q ′ − q), where δ(q ′ − q) is the Dirac delta function. The
term Dirac-orthonormal is all my own invention: it needed to be.) It follows from the general
probabilistic interpretation postulate of quantum mechanics, that expectation value of Qn for |Ψi
is given by Z
hQ i = dq q n |hq|Ψi|2 .
n
Show that this expression for hQn i also follows from the one in the preamble. What is dq |hq|Ψi|2
R
equal to?
SUGGESTED ANSWER:
a) Since the eigenvectors |qi i constitute a complete orthonormal set,
X
|Ψi = |qi ihqi |Ψi .
i
Thus
hQn i = hΨ|Qn |Ψi

X
= hΨ|Qn |qi ihqi |Ψi
i
X
= hΨ|qin |qi ihqi |Ψi
i
X
= qin |hqi |Ψi|2 ,
i
and so everything is consistent. Since |Ψi is normalized, we find for n = 0 that

X
hΨ|Ψi = 1 = |hqi |Ψi|2 .
i
b) Since the eigenvectors |qi constitute a complete Dirac-orthonormal set,

Z
|Ψi = dq |qihq|Ψi .
Thus
hQn i = hΨ|Qn |Ψi

Z
= dq hΨ|Qn |qihq|Ψi
Z
= dq hΨ|q n |qihq|Ψi
Z
= dq q n |hq|Ψi|2 ,
36
and so everything is consistent. Since |Ψi is normalized, we find for n = 0 that

Z
hΨ|Ψi = 1 = dq |hq|Ψi|2 .
008 qfull 00200 2 5 0 moderate thinking: simple commutator identities

35. Prove the following commutator identities.
a) [A, B] = −[B, A].
 
X X X
b)  ai Ai , bj Bj  = ai bj [Ai , Bj ], where the ai ’s and bj ’s are just complex numbers.
i j ij
c) [A, BC] = [A, B]C + B[A, C].

d) [A, [B, C]] + [B, [C, A]] + [C, [A, B]] = 0. This has always seemed to me to be perfectly useless
however true.
e) (c[A, B])† = c∗ [B † , A† ], where c is a complex number.
f) The special case of the part (e) identity when A and B are Hermitian and c is pure imaginary. Is
the operator in this special case Hermitian or anti-Hermitian?
SUGGESTED ANSWER:
a) [A, B] = AB − BA = −(BA − AB) = −[B, A].
b) Behold:
  !    !
X X X X X X
 ai Ai , bj Bj  = ai Ai  bj Bj  −  bj Bj  ai Ai
i j i j j i
X X X
= ai bj Ai Bj − ai bj Bj Aj = ai bj (Ai Bj − Bj Aj )
ij ij ij
X
= ai bj [Ai , Bj ] .
ij
c) [A, BC] = ABC − BCA = ABC − BAC + BAC − BCA = [A, B]C + B[A, C].
d) Behold:
[A, [B, C]] + [B, [C, A]] + [C, [A, B]] = (ABC − ACB − BCA + CBA)
+ (BCA − BAC − CAB + ACB)
+ (CAB − CBA − ABC + BAC)
=0.
e) Behold:
(c[A, B])† = c∗ (B † A† − A† B † ) = c∗ [B † , A† ] .
f) Using the parts (e) and (a) answers for A and B Hermitian and c pure imaginary,
(c[A, B])† = c∗ [B † , A† ] = −c[B, A] = c[A, B] .
In this special case, c[A, B] is itself Hermitian. If c is pure real then c[A, B] is anti-
Hermitian (see Gr-83).
008 qfull 00300 3 5 0 tough thinking: nontrivial commutator identities

37
Extra keywords: (Gr-111:3.41) but considerably extended.

36. Prove the following somewhat more difficult commutator identities.
a) Given
[B, [A, B]] = 0 , prove [A, F (B)] = [A, B]F ′ (B) ,
where A and B are general operators aside from the given condition and F (B) is a general operator
function of B. HINTS: Proof by induction is probably best. Recall that any function of an operator
is (or is that should be) expandable in a power series of the operator: i.e.,
∞
X
F (B) = fn B n ,
n=0
where fn are constants.

b) [x, p] = i−
h.
c) [x, p ] = i−
n
h npn−1 . HINT: Recall the part (a) answer.
d) [p, xn ] = −i−
h nxn−1 . HINT: Recall the part (a) answer.
SUGGESTED ANSWER:
a) Any operator function can be expanded into an operator series. Thus
∞
X
F (B) = fn B n , (1)
n=0
where fn are constants. Also

 
X X X
 gi Gi , h j Hj  = gi hj [Gi , Hj ] , (2)
i j i,j
where Gi and Hj are arbitrary operators and gi and hj arbitrary constants. Therefore,
∞
X ∞
X
[A, F (B)] = fn [A, B n ] = fn [A, B]nB n−1 = [A, B]F ′ (B) (3)
n=0 n=0
follows if
[A, B n ] = [A, B]nB n−1 . (4)
Thus we only need to prove the last equation which we will do by induction.
Step 1: For n = 0, equation (4) is true by inspection. For n = 1, it is again true by inspection.
For n = 2 (which is the first non-trivial case),
[A, B 2 ] = [A, B]B + B[A, B] = [A, B](2B) , (5)
where we have used the condition [B, [A, B]] = 0. Equation (5) satisfies equation (4) as
required.
Step 2: Assume equation (4) for n − 1.
Step 3: For n, we find
[A, B n ] = [A, B n−1 B] = [A, B n−1 ]B + B n−1 [A, B]

= [A, B](n − 1)B n−2 B + [A, B]B n−1 = [A, B]nB n−1 , (6)
where we have used Step 2 and the condition [B, [A, B]] = 0.
We have now completed the prove of equation (4), and thus the proof of equation (3) which is
the given identity.
38
b) Behold:
−
h
[x, p] = xp − px = xp − − xp = i−
h .
i
c) From the part (b) answer [p, [x, p]] = [p, −h /i] = 0. Thus from the part (a) answer it follows
that
[x, pn ] = [x, p]npn−1 = i−
h npn−1 .
d) From the part (b) answer [x, [p, x]] = [x, −− h /i] = 0. Thus from the part (a) answer it follows
that
[p, xn ] = [p, x]nxn−1 = −i− h nxn−1 .
008 qfull 01600 3 5 0 tough thinking: neutrino oscillation

37. There are systems that exist apart from 3-dimensional Euclidean space: they are internal degrees of
freedom such intrinsic spin of an electron or the proton-neutron identity of a nucleon (isospin: see, e.g.,
En-162 or Ga-429). Consider such an internal system for which we can only detect two states:

1 0
|+i = and |−i = .
0 1
This internal system is 2-dimensional in the abstract vector sense of dimensional: i.e., it can be described
completely by an orthonormal basis of consisting of the 2 vectors we have just given. When we measure
this system we force it into one or other of these states: i.e., we make the fundamental perturbation.
But the system can exist in a general state of course:

c+ (t)
|Ψ(t)i = c+ (t)|+i + c− (t)|−i = .
c− (t)
a) Given that |Ψ(t)i is NORMALIZED, what equation must the coefficients c+ (t) and c− (t) satisfy.
b) For reasons only known to Mother Nature, the states we can measure (eigenvectors of whatever
operator they may be) |+i and |−i are NOT eigenstates of the Hamiltonian that governs the time
evolution of internal system. Let the Hamiltonian’s eigenstates (i.e., the stationary states) be |+′ i
and |−′ i: i.e.,
H|+′ i = E+ |+′ i and H|−′ i = E− |−′ i ,
where E+ and E− are the eigen-energies. Verify that the general state |Ψ(t)i expanded in these
energy eigenstates,
− −
|Ψ(t)i = c+ e−iE+ t/ h |+′ i + c− e−iE− t/ h |−′ i
satisfies the general vector form of the Schrödinger equation:
∂
i−
h |Ψ(t)i = H|Ψ(t)i .
∂t
HINT: This requires a one-line answer.

c) The Hamiltonian for this internal system has no differential operator form since there is no wave
function. The matrix form in the |+i and |−i representation is

f g
H= .
g f
Given that H is Hermitian, prove that f and g must be real.

d) Solve for the eigenvalues (i.e., eigen-energies) of Hamiltonian H and for its normalized eigenvectors
|+′ i and |−′ i in column vector form.
39
e) Given at t = 0 that
a
|Ψ(0)i =
b
show that
1 − 1 −
|Ψ(t)i = √ (a + b)e−i(f +g)t/ h |+′ i + √ (a − b)e−i(f −g)t/ h |−′ i
2 2
and then show that
cos(gt/− −i sin(gt/−

−if t/−
h h) h)
|Ψ(t)i = e a +b .
−i sin(gt/−
h) cos(gt/−h)
HINT: Recall the time-zero coefficients of expansion in basis {|φi i} are given by hφi |Ψ(0)i.
f) For the state found given the part (e) question, what is the probability at any time t of measuring
(i.e., forcing by the fundamental perturbation) the system into state

1
|+i = ?
0
HINT: Note a and b are in general complex.

g) Set a = 1 and b = 0 in the probability expression found in the part (f) answer. What is the
probability of measuring the system in state |+i? in state |−i? What is the system doing between
the two states?
NOTE: The weird kind of oscillation between detectable states we have discussed is a simple model
of neutrino oscillation. Just as an example, the detectable states could be the electron neutrino and
muon neutrino and the particle oscillates between them. Really there are three flavors of neutrinos and
a three-way oscillation may occur. There is growing evidence that neutrino oscillation does happen.
(This note may be somewhat outdated due to that growth of evidence.)
SUGGESTED ANSWER:
a) Behold:
∗ ∗ c+ (t)
1 = hΨ(t)|Ψ(t)i = (c+ (t) , c− (t) ) = |c+ (t)|2 + |c− (t)|2 .
c− (t)
Thus c+ (t) and c− (t) satisfy
1 = |c+ (t)|2 + |c− (t)|2 .
b) Behold:
∂ − −
i−
h |Ψ(t)i = E+ c+ e−iE+ t/ h |+′ i + E− c− e−iE− t/ h |−′ i
∂t
− −
= c+ e−iE+ t/ h H|+′ i + c− e−iE− t/ h H|−′ i
= H|Ψ(t)i .
c) Since H is Hermitian, H † = H. Therefore

∗
g∗

f † f g
= H = H .
g∗ f ∗ g f
Thus f ∗ = f and g ∗ = g. Therefore f and g are real.

d) The eigenvalue equation is
(f − E)2 − g 2 = 0
which by inspection has solutions
E± = f ± g .
The equation for the column vector elements is
f u± + gv± = (f ± g)u±
40
leading to
v± = ±u± .
Thus the eigenvectors are given by

1 1
|±′ i = √ .
2 ±1
e) The time zero coefficients are given by

1 a 1
c± = √ (1, ±1) · = √ (a ± b) .
2 b 2
It immediately follows that
1 − 1 −
|Ψ(t)i = √ (a + b)e−i(f +g)t/ h |+′ i + √ (a − b)e−i(f −g)t/ h |−′ i
2 2

1 −
−i(f +g)t/ h 1 1 1 −i(f −g)t/−h 1 1
= √ (a + b)e √ + √ (a − b)e √
2 2 1 2 2 −1
− − − −
1 − e−igt/ h + eigt/ h 1 −if t/− e−igt/ h − eigt/ h
= ae−if t/ h − − + be h
− −
2 e−igt/ h − eigt/ h 2 e−igt/ h + eigt/ h

− h) h)
= e−if t/ h a + b .
−i sin(gt/− h) cos(gt/−h)
f) The amplitude for measuring the system in state |+i is

−if t/−
h h) h)
h+|Ψ(t)i = (1, 0) · e a +b
−i sin(gt/− h) cos(gt/−h)
−
= e−if t/ h a cos(gt/−
h ) − ib sin(gt/−

h) .
Thus the probability is
|h+|Ψ(t)i|2 = |a|2 cos2 (gt/−

h ) + |b|2 sin2 (gt/−
h ) + 2Re[a(−ib)∗ ] cos(gt/− h ) sin(gt/−h)
= |a| cos (gt/−
2 2
h ) + |b| sin (gt/−
2 2
h ) + 2Re[iab ] cos(gt/−
∗
h ) sin(gt/− h)
= |a|2 cos2 (gt/−
h ) + |b|2 sin2 (gt/−
h ) − 2Im[ab∗ ] cos(gt/−
h ) sin(gt/− h) ,
where we have used the fact that
Re[iz] = Re[ix − y] = −y = −Im[z] .
If ab∗ were pure real (which would be most easily arranged if a and b separately were pure
real), the cross term would vanish.
g) If a = 1 and b = 0, the probability of measuring the system in state |+i is
|h+|Ψ(t)i|2 = cos2 (gt/−

h) .
The probability of measuring the system in state |−i is
|h−|Ψ(t)i|2 = 1 − cos2 (gt/−

h ) = sin2 (gt/−
h) .
The system is oscillating between the two states.

Appendix 2 Quantum Mechanics Equation Sheet
Note: This equation sheet is intended for students writing tests or reviewing material. Therefore it neither
intended to be complete nor completely explicit. There are fewer symbols than variables, and so some
symbols must be used for different things.
1 Constants not to High Accuracy

Constant Name Symbol Derived from CODATA 1998
λCompton
Bohr radius aBohr = = 0.529 Å
2πα
Boltzmann’s constant k = 0.8617 × 10−6 eV K−1
= 1.381 × 10−16 erg K−1
h
Compton wavelength λCompton = = 0.0246 Å
me c
Electron rest energy m e c2 = 5.11 × 105 eV
Elementary charge squared e2 = 14.40 eV Å
e2
Fine Structure constant α= − = 1/137.036
hc
− 2
h
Kinetic energy coefficient = 3.81 eV Å2
2me
− 2
h
= 7.62 eV Å2
me
Planck’s constant h = 4.15 × 10−15 eV
Planck’s h-bar −
h = 6.58 × 10−16 eV
hc = 12398.42 eV Å
−
hc = 1973.27 eV Å
1
Rydberg Energy ERyd = me c2 α2 = 13.606 eV
2
2 Some Useful Formulae
n
dn (f g) X n dk f dn−k g
Leibniz’s formula =
dxn k dxk dxn−k
k=0
(x − hxi)2

1
Normalized Gaussian P = √ exp −
σ 2π 2σ 2
3 Schrödinger’s Equation
p2

∂Ψ(x, t)
HΨ(x, t) = + V (x) Ψ(x, t) = i−
h
2m ∂t
p2

Hψ(x) = + V (x) ψ(x) = Eψ(x)
2m
p2

∂Ψ(~r , t) ∂
HΨ(~r , t) = + V (~r ) Ψ(~r , t) = i−
h H|Ψi = i−
h |Ψi
2m ∂t ∂t
p2

Hψ(~r ) = + V (~r ) ψ(~r ) = Eψ(~r ) H|ψi = E|ψi
2m 41
42
4 Some Operators
−
h ∂ 2 ∂
2
p= p2 = −−
h
i ∂x ∂x2
− 2
p2 h ∂2
H= + V (x) = − + V (x)
2m 2m ∂x2
−
h 2
p= ∇ p2 = −−
h ∇2
i
− 2
p2 h
H= + V (~r ) = − ∇2 + V (~r )
2m 2m
∂ ∂ ∂ ∂ 1 ∂ 1 ∂
∇ = x̂ + ŷ + ẑ = r̂ + θ̂ + θ̂
∂x ∂y ∂z ∂r r ∂θ r sin θ ∂θ
∂2 ∂2 ∂2 ∂2

1 ∂ ∂ 1 ∂ ∂ 1
∇2 = 2
+ 2+ 2 = 2 r2 + 2 sin θ + 2 2
∂x ∂y ∂z r ∂r ∂r r sin θ ∂θ ∂θ r sin θ ∂φ2
5 Kronecker Delta and Levi-Civita Symbol
1, ijk cyclic;
(
1, i = j;
δij = εijk = −1, ijk anticyclic;
0, otherwise
0, if two indices the same.
εijk εiℓm = δjℓ δkm − δjm δkℓ (Einstein summation on i)
6 Time Evolution Formulae

dhAi ∂A 1
General = + − hi[H(t), A]i
dt ∂t h
dh~r i 1 dh~
pi
Ehrenfest’s Theorem = h~
pi and = −h∇V (~r )i
dt m dt
−
cj (0)e−iEj t/ h |φj i
X
|Ψ(t)i =
j
7 Simple Harmonic Oscillator (SHO) Formulae
2
!
1 −
h ∂2 1
V (x) = mω 2 x2 − + mω 2 x2 ψ = Eψ
2 2m ∂x2 2
β 1/2 1
r
mω 2 2 1 −
β= − ψn (x) = 1/4 √ Hn (βx)e−β x /2 En = n + hω
h π n
2 n! 2
H0 (βx) = H0 (ξ) = 1 H1 (βx) = H1 (ξ) = 2ξ

43
H2 (βx) = H2 (ξ) = 4ξ 2 − 2 H3 (βx) = H3 (ξ) = 8ξ 3 − 12ξ
8 Position, Momentum, and Wavenumber Representations
− 2
h k2
p=−
hk Ekinetic = ET =
2m
Ψ(k, t)
|Ψ(p, t)|2 dp = |Ψ(k, t)|2 dk Ψ(p, t) = √
−
h
−
h ∂
−
h ∂

xop = x pop = Q x, ,t position representation
i ∂x i ∂x
−
h ∂
−
h ∂

xop =− pop = p Q − , p, t momentum representation
i ∂p i ∂p
∞ − ∞
eipx/ h eikx
Z Z
δ(x) = dp δ(x) = dk
−∞ 2π−h −∞ 2π
∞ − ∞
eipx/ h eikx
Z Z
Ψ(x, t) = Ψ(p, t) dp Ψ(x, t) = Ψ(k, t) dk
−∞ (2π−h )1/2 −∞ (2π)1/2
∞ − ∞
e−ipx/ h e−ikx
Z Z
Ψ(p, t) = Ψ(x, t) dx Ψ(k, t) = Ψ(x, t) dx
−∞ (2π−h )1/2 −∞ (2π)1/2
− ~
ei~p·~r/ h 3 eik·~r
Z Z
Ψ(~r , t) = Ψ(~
p , t) d p Ψ(~r , t) = Ψ(~k , t) d3 k
all space (2π− h )3/2 all space (2π)3/2
− ~
e−i~p·~r/ h 3 e−ik·~r 3
Z Z
Ψ(~
p , t) = Ψ(~r , t) d r Ψ(~k , t) = Ψ(~r , t) d r
all space (2π− h )3/2 all space (2π)3/2
9 Commutator Formulae
 
X X X
[A, BC] = [A, B]C + B[A, C]  ai Ai , bj Bj  = ai bj [Ai , bj ]
i j i,j
if [B, [A, B]] = 0 then [A, F (B)] = [A, B]F ′ (B)
[x, p] = i−
h [x, f (p)] = i−
h f ′ (p) [p, g(x)] = −i−
h g ′ (x)
[a, a† ] = 1 [N, a] = −a [N, a† ] = a†
10 Uncertainty Relations and Inequalities

44
−
h 1
σx σp = ∆x∆p ≥ σQ σQ = ∆Q∆R ≥ |hi[Q, R]i|
2 2
−
h
σH ∆tscale time = ∆E∆tscale time ≥
2
11 Probability Amplitudes and Probabilities
Ψ(x, t) = hx|Ψ(t)i P (dx) = |Ψ(x, t)|2 dx ci (t) = hφi |Ψ(t)i P (i) = |ci (t)|2
12 Spherical Harmonics
1/2 1/2
1 3 3
Y0,0 =√ Y1,0 = cos(θ) Y1,±1 = ∓ sin(θ)e±iφ
4π 4π 8π
2
L2 Yℓm = ℓ(ℓ + 1)−
h Yℓm Lz Yℓm = m−
h Yℓm |m| ≤ ℓ m = −ℓ, −ℓ + 1, . . . , ℓ − 1, ℓ
0 1 2 3 4 5 6 ...
s p d f g h i ...
13 Hydrogenic Atom
ψnℓm = Rnℓ (r)Yℓm (θ, φ) ℓ≤n−1 ℓ = 0, 1, 2, . . . , n − 1
a

me
−
h λC e2
az = a0 = = α= −
Z mreduced me cα 2πα hc

−3/2 −r/aZ 1 −3/2 1 r
R10 = 2aZ e R20 = √ aZ 1− e−r/(2aZ )
2 2 aZ
1 −3/2 r −r/(2aZ )
R21 = √ aZ e
24 aZ
( 3 )1/2
2 (n − ℓ − 1)! 2r
Rnℓ = − e−ρ/2 ρℓ L2ℓ+1
n+ℓ (ρ) ρ=
naZ 2n[(n + ℓ)!]3 nrZ
q
x d
e−x xq

Lq (x) = e Rodrigues’s formula for the Laguerre polynomials
dx
j
d
Ljq (x) = Lq (x) Associated Laguerre polynomials
dx
aZ 2
hrinℓm = 3n − ℓ(ℓ + 1)
2
Nodes = (n − 1) − ℓ not counting zero or infinity

45
1 Z 2 mreduced Z 2 mreduced Z 2 mreduced

En = − me c2 α2 2 = −ERyd 2 = −13.606 2 eV
2 n me n me n me
14 General Angular Momentum Formulae
[Ji , Jj ] = i−
h εijk Jk (Einstein summation on k) [J 2 , J~ ] = 0
2
J 2 |jmi = j(j + 1)−
h |jmi Jz |jmi = m−
h |jmi
J± |jmi = −
p
J± = Jx ± iJy h j(j + 1) − m(m ± 1)|jm ± 1i
( )
1
J{ x } = 2
1 (J+ ± J− ) †
J± J± = J∓ J± = J 2 − Jz (Jz ± −
h)
y
2i
[Jf i , Jgj ] = δf g i−
h εijk Jk J~ = J~1 + J~2 J 2 = J12 + J22 + J1+ J2− + J1− J2+ + 2J1z J2z
X
J± = J1± + J2± |j1 j2 jmi = |j1 j2 m1 m2 ihj1 j2 m1 m2 |j1 j2 jmij1 j2 jmi
m1 m2 ,m=m1 +m2
jX
1 +j2
|j1 − j2 | ≤ j ≤ j1 + j2 (2j + 1) = (2j1 + 1)(2j2 + 1)

|j1 −j2 |
15 Spin 1/2 Formulae
−
h 0 1
−
h 0 −i
−
h 1 0

Sx = Sy = Sz =
2 1 0 2 i 0 2 0 −1
1 1
|±ix = √ (|+i ± |−i) |±iy = √ (|+i ± i|−i) |±iz = |±i
2 2
1
| + +i = |1, +i|2, +i | + −i = √ (|1, +i|2, −i ± |1, −i|2, +i) | − −i = |1, −i|2, −i
2

0 1 0 −i 1 0
σx = σy = σz =
1 0 i 0 0 −1
σi σj = δij + iεijk σk [σi , σj ] = 2iεijk σk {σi , σj } = 2δij
~ · ~σ )(B
(A ~ · ~σ ) = A
~·B
~ + i(A
~ × B)
~ · ~σ
~ · n̂)
d(S i ~ ~ α~ ~ ~
= − − [S ~ · n̂]
· α̂, S ~ · n̂ = e−iS·~
S S · n̂0 eiS·~α |n̂± i = e−iS·~α |ẑ± i
dα h
46
eixA = 1 cos(x) + iA sin(x) if A2 = 1 e−i~σ ·~α/2 = 1 cos(x) − i~σ · α̂ sin(x)
σi f (σj ) = f (σj )σi δij + f (−σj )σi (1 − δij )
e−
h
µBohr = = 0.927400915(23) × 10−24 J/T = 5.7883817555(79) × 10−5 eV/T
2m
α
g = 2 1+ + . . . = 2.0023193043622(15)
2π
~
L ~
S ~ + g S)
(L ~
~µorbital = −µBohr − µ
~ spin = −gµBohr − ~µtotal = ~µorbital + ~µspin = −µBohr −
h h h
~ (Lz + gSz )
Hµ = −~
µ·B Hµ = µBohr Bz −
h
16 Time-Independent Approximation Methods
∞
X
H = H (0) + λH (1) |ψi = N (λ) λk |ψn(k) i
k=0
m
X ∞
X
H (1) |ψn(m−1) i(1 − δm,0 ) + H (0) |ψn(m) i = E (m−ℓ) |ψn(ℓ) i |ψn(ℓ>0) i = anm |ψn(0) i
ℓ=0 m=0, m6=n
D E
(0) (0)
ψk |H (1) |ψn
(0)
X
|ψn1st i = |ψn(0) i + λ (0) (0)
|ψk i
all k, k6=n En − Ek
D E
En1st = En(0) + λ ψn(0) |H (1) |ψn(0)
(0) (1) (0) 2
D E
D E X ψk |H |ψn
En2nd = En(0) + λ ψn(0) |H (1) |ψn(0) + λ2 (0) (0)
all k, k6=n En − Ek
hφ|H|φi
E(φ) = δE(φ) = 0
hφ|φi
Hkj = hφk |H|φj i H~c = E~c
17 Time-Dependent Perturbation Theory
∞
sin2 (x)
Z
π= dx
−∞ x2
47
2π
Γ0→n = − |hn|Hperturbation|0i|2 δ(En − E0 )
h
18 Interaction of Radiation and Matter
~
~ op = − 1 ∂ Aop
E ~ op = ∇ × A
B ~ op
c ∂t
19 Box Quantization
2πn 2π 3 (2π)3
kL = 2πn, n = 0, ±1, ±2, . . . k= ∆kcell = ∆kcell =
L L V
k 2 dk dΩ
dNstates = g
(2π)3 /V
20 Identical Particles
1
|a, bi = √ (|1, a; 2, bi ± |1, b; 2, ai)
2
1
ψ(~r1 , ~r 2 ) = √ (ψa (~r 1 )ψb (~r 2 ) ± ψb (~r 1 )ψa (~r 2 ))
2
21 Second Quantization
(a† )Nn (a† )N1

[ai , a†j ] = δij [ai , aj ] = 0 [a†i , a†j ] = 0 |N1 , . . . , Nn i = √n . . . √1 |0i
Nn ! N1 !
{ai , a†j } = δij {ai , aj } = 0 {a†i , a†j } = 0 |N1 , . . . , Nn i = (a†n )Nn . . . (a†1 )N1 |0i
X e−i~p·~r † X ei~p·~r
Ψs (~r )† = √ ap~s Ψs (~r ) = √ ap~s
p
~
V p
~
V
[Ψs (~r ), Ψs′ (~r ′ )]∓ = 0 [Ψs (~r )† , Ψs′ (~r ′ )† ]∓ = 0 [Ψs (~r ), Ψs′ (~r ′ )† ]∓ = δ(~r − ~r ′ )δss′
1
|~r1 s1 , . . . , ~rn sn i = √ Ψsn (~r n )† . . . Ψsn (~r n )† |0i
n!
√
Ψs (~r )† |~r1 s1 , . . . , ~rn sn i n + 1|~r1 s1 , . . . , ~rn sn , ~rsi
Z
|Φi = d~r1 . . . d~rn Φ(~r1 , . . . , ~rn )|~r1 s1 , . . . , ~rn sn i
X Z ∞
X
1n = d~r1 . . . d~rn |~r1 s1 , . . . , ~rn sn ih~r1 s1 , . . . , ~rn sn | 1 = |0ih0| + 1n
s1 ...sn n=1
48
X p2 †
ap†~s ap~s
X
N= T = a ap~s
2m p~s
p
~s p
~s
XZ 1 X
Z
ρs (~r ) = Ψs (~r )† Ψs (~r ) N= d~r ρs (~r ) T = d~r ∇Ψs (~r )† · ∇Ψs (~r )
s
2m s
~js (~r ) = 1
Ψs (~r )† ∇Ψs (~r ) − Ψs (~r )∇Ψs (~r )†

2im
3n sin(x) − x cos(x) Gs (~r − ~r ′ )2

Gs (~r − ~r ′ ) = gss′ (~r − ~r ′ ) = 1 − δss′
2 x3 (n/2)2
1X
Z
v2nd = d~rd~r ′ v(~r − ~r ′ )Ψs (~r )† Ψs′ (~r ′ )† Ψs′ (~r ′ )Ψs (~r )
2 ′
ss
1 X X
Z
vp~−~p ′ δp~+~q,~p′ +~q′ ap†~s aq†~s′ aq~ ′ s′ ap~ ′ s
′
v2nd = vp~−~p ′ = d~r e−i(~p−~p )·~
r
v(~r )
2V ′ ′ ′
pp qq ss
22 Klein-Gordon Equation
2 "
− 2 #
p 1 − ∂ h 2 2
E = p 2 c2 + m 2 c4 ih Ψ(~r, t) = ∇ + m c Ψ(~r, t)
c2 ∂t i
" 2 #
1 ∂2

mc
− ∇2 + − Ψ(~r, t) = 0
c2 ∂t2 h
i− ∂Ψ∗ −

h ∗ ∂Ψ ~j = h (Ψ∗ ∇Ψ − Ψ∇Ψ∗ )
ρ= Ψ − Ψ
2mc2 ∂t ∂t 2im
2 "
− 2 #
1 − ∂ h e~ 2 2
i h − eΦ Ψ(~r, t) = ∇ − A + m c Ψ(~r, t)
c2 ∂t i c
− −
p, E) = ei(~p·~r−Et)/ h
Ψ+ (~ p, E) = e−i(~p·~r−Et)/ h
Ψ− (~

Answer Table For The Multiple-Choice Questions: Quantum Mechanics Name

Uploaded by

Copyright:

Available Formats

Answer Table For The Multiple-Choice Questions: Quantum Mechanics Name

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Answer Table For The Multiple-Choice Questions: Quantum Mechanics Name

Uploaded by

Copyright:

Available Formats

1

Quantum Mechanics NAME:

Answer Table for the Multiple-Choice Questions

007 qmult 00100 1 1 5 easy memory: vector addition

007 qmult 00200 1 4 4 easy deducto-memory: Schwarz inequality

007 qmult 00300 1 4 5 easy deducto-memory: Gram-Schmidt procedure

007 qmult 00400 1 4 4 moderate memory: definition unitary matrix

007 qmult 00500 2 3 4 moderate math: trivial eigenvalue problem

a) Both are 0. b) 0 and 1. c) 0 and −1. d) 0 and 2. e) −i and 1.

which clearly has solutions 0 and 2.

007 qfull 00090 1 5 0 easy thinking: ordinary vector space

(x1 , y1 , 1) + (x3 , y2 , 1) = (x1 + x2 , y1 + y2 , 2)

is not in the subset.

Redaction: Jeffery, 2001jan01

007 qfull 00100 2 5 0 moderate thinking: vector space, polynomial

iii) Vector addition is associative:

(|αi + |βi) + |γi = |αi + (|βi + |γi) .

iv) There is a zero or null vector |0i such that

|αi + |0i = |αi ,

v) For every vector |αi there is an inverse vector | − αi such that

vi) Scalar multiplication of a vector gives a vector:

vii) Scalar multiplication is distributive on vector addition:

a(|αi + |βi) = a|αi + a(|βi) .

viii) Scalar multiplication is distributive on scalar addition:

(a + b)|αi = a|αi + b|αi .

ix) Scalar multiplication is associative with respect to scalar multiplication:

xi) Finally, one has

b) The subset of even polynomials.

007 qfull 00110 2 5 0 moderate thinking: unique expansion in basis

007 qfull 00200 3 5 0 tough thinking: Gram-Schmidt orthonormalization

where recall that the norm of a vector |αi is given by

|α2 i − |α′1 ihα′1 |α2 i

a) Write down the general step of the Gram-Schmidt procedure.

Then take the scalar product with bra hαi |:

Redaction: Jeffery, 2001jan01

007 qfull 00300 2 3 0 moderate math: prove the Schwarz inequality

and work your way by valid steps to

|β⊥ i = |βi − |βk i .

where a is some constant, and

|βk i = a|α̂ihα̂|α̂i = a|α̂i = |βi .

hβ|βi = hβk |βk i + hβk |β⊥ i + hβ⊥ |βk i + hβ⊥ |β⊥ i

007 qfull 00310 1 3 0 easy math: find a generalized angle

where |αi and |βi are general non-zero vectors.

b) Find the generalized angle between vectors

Thus the cosine of the generalized angle is

007 qfull 00400 1 3 0 easy math: prove triangle inequality

|hα|βi|2 ≤ hα|αihβ|βi = ||α||2 × ||β||2 .

||(|αi + |βi)||2 = hα|αi + hβ|βi + hα|βi + hβ|αi

Redaction: Jeffery, 2001jan01

007 qfull 00500 3 3 0 tough math: simple matrix identities

where 1 is the unit matix. Thus

(AB)−1 A = B −1 and (AB)−1 = B −1 A−1 ,

which completes the proof.

The converse is false.

are not Hermitian, but their sum clearly is.

Thus the sum of unitary matrices is not in general unitary.

007 qfull 00510 2 5 0 moderate thinking: commuting operations

Note 1 is the unit matrix. We have used the facts that

(z1 + z2 )∗ = [(x1 + x2 ) + i(y1 + y2 )]∗