Some Aspects of Localization Frames: Bhabhuti Chauhan

Some Aspects of Localization Frames
Bhabhuti Chauhan
Research Scholar
Department of Mathematics
University of Delhi
August 30, 2022

Table of Contents
1 Chapter 1: Introduction and Preliminaries

2 Chapter 2: Localization of Frames
O. Christensen, An introduction to frames and Riesz bases, Ap-
plied and Numerical Harmonic Analysis, Birkaüser, Inc., Boston,
MA, 2003
3 Chapter 3: Banach Frames, Decay in the Con-
text of Localization
A. Aldroubi and K. Grochenig, Nonuniform sampling and recon-
struction in shift-invariant space, SIAM Rev., 43 (4), 585 – 620,
2011
Bhabhuti Chauhan (Mathematics, DU) Some Aspects of Localization Frames August 30, 2022 2 / 49
Table of Contents (Contd.)
1 Chapter 4: Localization of Frames, Banach Frames

and the Invertibility of the Frame Operator
C. Chui, W. Czaja, M. Maggioni and G. Weiss, Characteriza-
tion of general tight frames with matrix dilations and tightedness
preserving oversampling, Journal of Fourier Analysis and Appli-
cations, 8 (2) (2002), 173 – 200
2 Chapter 5: Density, Overcompleness, and Local-
ization of Frames
R. Balan, P.G. Casazza, C. Heil and Z. Landau, Deficits and
excesses of frames, Adv. Comput Math. 18, 93 – 116
R. Balan, P.G. Casazza, C. Heil and Z. Landau, Excesses of Gabor
frames, Appl. Comput. Harmon. Anal. 14, 87 – 106
Introduction
Introduction
Definitions
Some Related Results
Introduction
The science of localization started in 1940s with Claude

Shannon who found fundamental limits of signal processing.
He developed a precise framework, the so-called ‘Information
Theory’. In 1949, Claude Shannon provided some results in
his paper entitled ‘Communication in the Presence of Noise’
in localization theory.
Introduction
The science of localization started in 1940s with Claude

Shannon who found fundamental limits of signal processing.
He developed a precise framework, the so-called ‘Information
Theory’. In 1949, Claude Shannon provided some results in
his paper entitled ‘Communication in the Presence of Noise’
in localization theory.
A typical problem in sampling theory is to reconstruct a func-
tion (signal) f in a separable Hilbert space from its samples,
for which a natural approach would be used to sample the
function f at as many accessible positions as possible.
The papers (published in 2009 and 2011) of Vetterli et
al. can be seen as the first work on Dynamical localiza-
tion. They consider the localization of signals under dif-
fusion evolution. They investigate the spatial-temporal
localization of a linear diffusion field, and show that it is
possible to compensate for insufficient spatial localiza-
tion densities by oversampling in time.
Frame
Let V ̸= {0} be a finite-dimension vector space. As standing
assumption we will assume that V is equipped with an inner
product ⟨·, ·⟩ which we choose to be linear in the first entry.
If the following condition satisfied.
(i)
V = span{ek }m
k=1
(ii)
{ek }m
k=1 is linearly independent i.e.
m
if ck ek = 0 for some scalar coefficient {ck }m
k=1 ,
P
k=1
then ck = 0 for all k = 1, 2, . . . , m.
Localization Frames
Let f = (fx )x ∈ X and E = (ey ), y ∈ Y are two frame of H = L2 (R).
Whose indexing sets X and Y are countable subset of R.
We can think of each fx ∈ F as being nbd of x and similarly for
each ey ∈ E then F is localized with respect to E if each fx can be
well-approximated by a finite linear combination of ey i.e.
fx = c1 e1 + c2 e2 + · · · + cy ey
X
fx = cy ey (1)
y ∈Y
both side take inner product with eJ , J = 1, 2, . . . , y then

X X
⟨fx , eJ ⟩ = ⟨cy ey , ej ⟩ = cy ⟨ey , ej ⟩
y ∈Y
Localization Frames (Contd.)

X 1 if J = y
= cy ⟨ey , ej ⟩ = cy
0 ̸ y
if J =
⟨fx , ej ⟩ = cJ
⟨fx , ey ht⟩ = cy in J = y (1),
X
fx = ⟨fx , ey ⟩ey
We say that F is localized with respect E if the magnitudes of the

coefficient |⟨fx , ey ⟩| is decay if distance between x and y increase.
F is localized with respect to E if cross-Gramian matrix
|⟨fx , ey ⟩|x ∈X , y ∈Y
has a decay of the diagonal.
Weight Functions
A weight function is a mathematical device used when performing
a sum, integral, or average to give some element more “weight”
or influence on the result than other elements in the same set.
Weight function m : X → R + is a positive function defined on a
separated set X with is typically finite or countable.
The weight function m(a) = 1, corresponds to the unweight
situation in which all elements have equal weight.
If function f : X → R is a real valued function then the un-
weighted sum of f on x is defined as f (a) but given a
P
a∈X
+
weight function m : X → R , the weighted sum is defined as
f (a) m(a).
P
a∈X
Weight Functions (Contd.)
N and X are countable index sets in some R d and we my
assume that both X , N are separated, this means that
inf |x − y | ≥ α > 0
x ,y ∈X ,x ̸=y
and like wise for N , whenX is sued to index a family of func-

tion an R d , the index x in fx (t) indicates that the essential
support of fx is centered at x .
Definition
A set X ⊆ Rd is separated if there exists a positive constant
c such that for every x , y ∈ X such that x ̸= y , 0 < c ≤
|x − y |.
Definition
Let S 1 := {F = (fx )x ∈X |F is a symmetrically ℓ1 -self-
localized frame of H}. For F, E ∈ S 1 , we define relation
F ∼ E if F is symmetrically ℓ1 -localized with respect to E .
Localization of the Dual
Definition
Let (B, ∥ · ∥Bd ) be a Banach space of sequences indexed by X . A
(countable) subset (fx : x ∈ X ) of B 1 , the dual of B, is called a
Banach frame for B if the following properties hold:
(a) The coefficient operator CE : B → Bd (X ) defined by CE f =
(fx (f ))x ∈X is bounded.
(b) We have the norm equivalence ∥f ∥B ≍ ∥fx (f )∥Bd .
(c) There exist a bounded operator R : Bd (X ) → B, called the
reconstruction operator, such that R((fx (f ))x ∈X ) = f .
Definition
For 1 ≤ p < ∞, the weighted ℓp -space ℓpm (Y ) on the index
set Y ⊂ Rd is defined by the norm
 1/p
|p m(y )p 
X
∥c∥ℓpm =  |cy
y ∈Y
with the usual modification for p = ∞. The weight m is a

non-negative function on Rd , which we may assume without
loss of generality is continuous.
Some Related Results
(1) (F, E ) is symmetrically ℓp -localized if there exist maps aX : X →
G, ay : Y → G such that max{supj∈G |aX 1 (j)|, supj∈G |aY−1 (j)|} ≤
K < ∞, and r ∈ ℓp (G) such the for all x ∈ X , y ∈ Y ,
⟨fx , ey ⟩| ≤ raX (x )−aY (y ) .

(2) F is symmetrically ℓp -self-localized if it is symmetrically ℓp -localized
with respect to itself.
(3) (F, E) is symmetrically ℓp -localized if there exist maps aX : X →
G, aY : Y → G such that max{supj∈G |aX−1 (j)|, supj∈G |aY−1 (j)|} ≤
K < ∞, and r ∈ ℓp (G) such that for all x ∈ X , y ∈ Y
|⟨fx , cy ⟩| ≤ raX (x ) − aY (y ).
(4) F is symmetrically ℓp -self-localized if it is symanetricully ℓp -localized
with respect to itself.
Chapter 2: Localization of Frames
O. Christensen, An introduction to frames and Riesz

bases, Applied and Numerical Harmonic Analysis, Birkaüser
Inc., Boston, MA, 2003
Chapter 2
Introduction
Localization of Frames
Some Theorem and Results
Introduction
In this chapter, we study localized frames and its analysis an

associated Banach space.
Throughout this chapter E = {ex : x ∈ X } denotes for
frame for a given Hilbert space H; this means that the frame
operator Sf = SE f = ⟨f , eX ⟩ex is boundedly invertible on
P
x ∈X
H. The set {gn : n ∈ N } is always a Riesz basis for H with
dual basis {g̃n : n ∈ N }. The set N is assumed to be
separated in Rd , i.e., inf |x − y | ≥ δ > 0, and X is
x ,y ∈X :x ̸=y
relatively separated, i.e., X is a finite union of separated sets.
We should think of the index as a kind of localization. For
instance, if H = L2 (Rd ), then the index x in ex (t) indicates
that the essential support of ex is centered at x .
Polynomially localized
A frame E = {ex : x ∈ X } for a Hilbert space H is said
to be polynomially localized with respect to the Riesz basis
{gn } iwht decay s > 0, if
|⟨ex · gn | ≤ c(1 + |x − n|)−s
and
|⟨ex · g̃n | ≤ c(1 + |x − n|)−s
Exponentially localized
A frame E = {ex : x ∈ X } for a H is called exponentially
localized with respect to the Riesz basis {gn }, if for some
α>0
max{|⟨ex , gn ⟩|, |⟨ex , g̃n ⟩|} ≤ ce −α|x −n| ∀ x ∈ X, n ∈ N
Some Theorems and Results
Theorem
If E is localized with respect to the Riesz basis {gn } for
some d > 0, then Ẽ is also localized.
If E is exponentially localized, then Ẽ is also exponen-
tially localized.
Theorem
F ∈ S ′ . Then for F, F ∈ S ′ and F ∼ F. Let F = (fx )x ∈X and
E = (ey )y ∈Y in a Hilbert space H, X and Y arbitrary index sets.
F is a symmetrically ℓP -localized with respect to E if there exist
maps aX : X → G, aY : Y → G such that
{sup |aX−1 (J)|, sup |aY−1 (g)|} ≤ K < ∞

J∈G
and r ∈ ℓP (G) such that for all x ∈ X , y ∈ Y
|⟨fx , ey ⟩| ≤ rax (x )−ay (y ) (2)
F ∈ S ′ and F is a symmetrically ℓ′ -self-localized frame of H, then

there exist maps a :→ G be an associated map.
Theorem (Contd.)
We say that (F, a) is ℓp -self localized if there exists r ∈ ℓP (G) such
that
∀ i, j ∈ I |⟨fi , fj ⟩| ≤ ra(i) − a(j)
and (F, a) is ℓp -localized with respect to its canonical dual frame
f = {f y }i∈I . If there exist r ∈ ℓP (G) such that
∀ i, j ∈ I |⟨fi , f j ⟩| ≤ ra(i) − a(j) (3)
comparing eq. (2) and (3) we get
ey = f˜j ∀ y ∈ y, j ∈ I
⇒ F =E
F is symmetrically ℓ′ -localized with respect to E = F
⇒ F, F ∈ S ′ and F ∼ F.
Chapter 3: Banach framed, Decay in the Context
of Localization
This chapter is based on the paper

A. Aldroubi and K. Grochenig, Nonuniform sampling and
reconstruction in shift-invariant space, SIAM Rev., 43 (4),
585 – 620, 2011
Chapter 3
Introduction
Banach Framed, Decay in the context of Localization
Introduction
In this chapter we study the Banach frames, decay in the

context of localization.
Let H = L2 (R) and consider two frames F = (fx )x ∈X and E = (ey )y ∈Y
of H whose indexing sets X and Y are countable subsets of R. We
can think of each fx ∈ F as being “concentrated” near x , and similarly
for each ey ∈ E. Roughly, F is localized with respect to E if each
fx can be well-approximated by a finite linear combination of ey ’s. In
other words, since fx can be written as
X
fx = ⟨fx , ey ⟩ẽy
y ∈Y
we say that F is localized with respect to E if the magnitude of the

coefficients |⟨fx , ey ⟩| exhibit a certain decay as the distance between x
and y increases. Equivalently, F is localized with respect to E if the
cross-Gramian matrix |⟨fx , ey ⟩|x ∈X ,y ∈Y has a decay of the diagonal.
Banach Framed, Decay in the context of
Localization
The frame F = (fx )x ∈X in L2 (Rd ) is polynomially localized with respect
to the Riesz basis E = (ey )y ∈Y is L2 (Rd ), visth decay s > 0 (or s-
localized), where X is a finite union of separated sets of Rt and Y is
a separated set of Rd , if for all x ∈ X , y ∈ Y , and C > 0,
|⟨fx , ey ⟩| ≤ C (1 + |x − y |)−5 and |⟨fx , ẽy ⟩| ≤ C (1 + |x − y |)−s .
Likewise, F is called exponentially localized with exponent α > 0 if

for sume α > 0 and C > 0,
|⟨fx , ey ⟩| ≤ Ce−α|x −y | and |⟨fx , ẽy ⟩| ≤ Ce −α|x −y | .
To prove that every polynomially localized frame is symmetrically lo-

calized, let F = (fx )x ∈X be s-localized with respect to a Ries basis
E = ⟨ey t⟩y ∈Y as above.
Theorem
→
− →− →
−
Let F ∈ S 1 . Then for F , F ∈ S 1 and F ∼ F This theorem applies
directly as our definition coincides with that of BCHL in the case of
self-localization.
Proposition
Let F = (fx )x ∈X and E = (ey )y ∈Y be frame sequences for Hilbert
space H, X , and Y arbitrary index sets. Let aX : X → G, aY : Y → G
be associated maps. Suppose the following conditions are satisfied:
(1) F is symmetrically ℓ1 -localszed with respeet to E, i.e., thene exsts
r ∈ ℓ1 (G) sweh that |⟨fx , ey ⟩| ≤ raX (x ) − aY (y )
(2) F is symmetrically ℓ1 -lox alized with respect to Ẽ, i.e., there exists
s ∈ ℓ1 (G) sweh that |⟨fx , ẽy ⟩| ≤ saX (x ) − aY (y )
Then F ∈ S 1 ard E ∈ S 1 .
Chapter 4: Localization of Frames, Banach
Frames, and the Invertibility of the Frame Operator
C. Chui, W. Czaja, M. Maggioni and G. Weiss, Characteriza-

tion of general tight frames with matrix dilations and tight-
edness preserving oversampling, Journal of Fourier Analysis
and Applications, 8 (2) (2002), 173 – 200
Chapter 4
Introduction
Localization of Frames, Banach Frames, and the Invari-
ability of the Frame Operator
Introduction
In this chapter, we collect the main properties of polyno-

mials and sub-exponential weight function and present an
axiomatic theorem of the localization of frames. We define
a family of Banach Spaces associated to each Hilbert Space
and Riesz basis and investigate the behavior of the frame
operator on these space.
N and X are countable index sets in some R d and we may
assume that both X and N are separated. This means that
inf |x − y | ≥ δ > 0
x ,y ∈X ,x ̸=y
and likewise for N .

When X is used to index a family of functions on R d , the
index x in fx (t) indicates that the essential support of fx is
centered at x .
Localization of the Frames, Banach Frames, and
the Invariability of the Frame Operator
Polynomial and Sub-Exponential Weight Functions

A weight is a nonnegative function on Rd which we may assume
without loss of generality to be continuous. A weight m is called
s-moderate, if there are constants C , s ≥ 0 such that
m(t + x ) ≤ C (1 + |t|)s m(x ) for all t, x ∈ Rd (4)
The weight function occurring in (4),
vs (t) = (1 + |t|)s for t ∈ Rd , (5)
is submultiplicative, i.e., vs satisfies vs (t + x ) ≤ vs (t)vs (x ).
Polynomial and Sub-Exponential Weight Functions
(Contd.)
A weight function m is called sub-exponential if there are constants C , γ > 0 and
0 ≤ β < 1 such that
β
m(t + x ) ≤ Ce γ|t| m(x ) for all t, x ∈ Rd . (6)
By setting x = 0 in (4) and in (6) we see that a an s-moderate weight m grows at

most polynomially, i.e., m(t) ≤ C (1 + |t|)s , and a sub-exponential weight grows
β
at most like Ce α|t| .
The weighted ℓp -space ℓpm (X ) on the index set X is defined by the norm
!1/p
X p p
∥c∥ℓpm = |cx | m(x ) , (7)
x ∈X
with the usual modification for p = ∞.

In dealing with s-moderate and sub-exponential weights, we will repeatedly use the
following lemmas.
Lemma
If X ⊆ Rd is separated, then for any s > d
X
sup (1 + |x − v |)−s = Cs < ∞ (8)
v ∈Rd
x ∈X
Proof
Fix v ∈ Rd . Since |x − y | ≥ δ for x ̸= y ∈ X , each box of side length α = δ
√
2 d
contains
at most one point x − v . Therefore we may write x − v = αnx + µx for unique nx ∈ Zd
and µx ∈ [−α, α)d , and by the choice of α we have nx ̸= ny for x ̸= y ∈ X . Since
(1 + |x + y |)−s ≤ (1 + |x |)s (1 + |y |)−s for x , y ∈ Rd and s ≥ 0, we find that
X X X
(1 + |x − v |)−s = (1 + |αnx + µx |)−s ≤ (1 + |µx |)s (1 + |αnx |)−s
x ∈X x ∈X x ∈X
!  
X −s −s
X −s
≤ C1 1+ |αnx | ≤ C1 1 + α |n| <∞
x ∈X :nx ̸=0 n∈Z\{0}
Clearly this estimate is independent of v , hence the assertion.
Lemma
Assume that X ⊆ Rd is separated
(a) If s > d, then
(1 + |x − n|)−s (1 + |x − m|)−s ≤ C (1 + |n − m|)−s

X
m, n ∈ N .
x ∈X
(9)
(b) There exists α′ ∈ (0, α) such that

′
e −α|x −n| e −α|x −m| ≤ Ce −α |m−n| .
X
(10)
x ∈X
Density, Overcompleteness, and Localization of
Frames
This chapter is based on the paper:

R. Balan, P.G. Casazza, C. Heil and Z. Landau, Deficits and
excesses of frames, Adv. Comput Math. 18, 93 – 116
Chapter 5
Introduction
Density, Overcompleteness and Localization of Frames
Introduction
The fundamental structural feature of a frame that is not a Riesz

basis is the overcompleteness of its elements. To date, even
partial understanding of this overcompleteness has been restricted
to limited examples, such as finite-dimensional frames.
Our approach begins with two frames F = {fi }i∈I and E =
{ej }j∈G , where G is a discrete abelian group; we then introduce
a notion of the localization of F with respect to E. The idea of
localization is that it describes the decay of the coefficients of the
expansion of elements of F in terms of the elements of E. To
make this notion of decay meaningful, a map a from the index
set I into the index set G is introduced.
Density Overcompleteness, and Localization of
Frames
Definition
(a) Let F = {fi }i∈I and E = {ej }j∈G be frame sequences in H, and let
a : I → G be an associated map. Let PF , PE denote the orthogonal
projections of H onto span(F) and span(E), respectively. Then given
a free ultrafilter p and a sequence of centers c = (cN )N∈N in G, we
define the relative measure of F with respect to E, p, and c to be
1
⟨PE fi , f¯i ⟩
X
ME (F; p, c) = pN∈N
|IN (cN )| i∈IN (cN )
The relative measure of E with respect to F, p, and c is
1 X
MF (E; p, c) = p − lim ⟨PF ẽj , ej ⟩.
N∈N |SN (cN )|
j∈SN (cN )
Definition (Contd.)
(b) If span(E) ⊃ span(F) then PE is the identity map and E plays no role in
determining the value of Mε(F; p, c). Therefore, in this case we define the
measure of F with respect to p and c to be
1
⟨fi , f˜i ⟩
X
M(F; p, c) = pN∈N
|IN (cN ) | i∈I (c )
N N
2
Since ⟨fi , f¯i ⟩ = S −1/2 fi , we have that M(F; p, c) is real. Additionally,

since S −1/2 (F) is a Parseval frame, we have 0 ≤ ⟨fi , f˜i ⟩ ≤ 1 for all i, and
therefore
0 ≤ M(F; p, c) ≤ 1
For this case we further define the lower and upper measures of F by
1
M− (F) = lim inf inf ⟨fi , f¯i ⟩
X
N→∞ j∈G |IN (j)|
i∈I (j) N
Definition (Contd.)
1
⟨fi , f˜i ⟩.
X
M+ (F) = lim sup sup
N→∞ j∈G |IN (j)| i∈I (j)
N
It can be seen that there exist free ultrafilters p − , p + and sequences of centers
c − , c + such that M− (F) = M (F; p − , c − )and M+ (F) = M (E; p + , c + )
(c) When span(F) ⊃ span(E), we define the measures M(E; p, c) and
M± (E) in an analogous manner.
Theorem
Let F = {fi }i∈I be a frame for H, and let a : I → G be
an associated map such that D + (I, a) < ∞. Let F be the
canonical dual frame and S −1/2 (F) the canonical Parseval
frame. If (F, a) is ℓ1 -self-localized, then:
(a) (F, a) isn ℓo1 -localized with respect to its canonical dual
frame F̃ = f˜i i∈I ,
(b) (F, a) is ℓ1 -self-localized, and

(c) S −1/2 (F), a is ℓ1 -self-localized.
References I
R. Aceska, A. Aldroubi, J. Davis and A. Petersyan, Dynamical

Sampling in shift-invariant spaces, in Commutative and non-
commutative harmonic analysis and applications, vol. 603 of
Contemp. Math., Amer. Math. Soc., Providence, RI, 2013,
pp. 139-148.
A. Aldroubi, C. Cabrelli, A. Cakmak, U. Molter and A. Pet-
rosyan, Iterative actions of normal operators, J. Funct. Anal.,
272 (2017), pp. 1121-1146.
A. Aguilera, C. Cabrelli, D. Carbajal and V. Paternostro,
Dyanmical sampling for shift-preserving operators. Applied
and Computational Harmonic Analysis, 51 (2001), pp. 258-
274.
References II
A. Aldroubi, C. Cabrelli, U. Molter and S. Tang, Dynamical
sampling, Applied and Computational Harmonic Analysis, 42
(2017), pp. 378401, doi: 10.1016/j.acha.2015.08.014.
A. Aldroubi, J. Davis and I. Krishtal, Dynamical Sam-
pling time-space trade-off, Appl. Comput. Harmon. Anal., 34
(20130), pp. 495-503.
A. Aldroubi, J. Davis and I. Krishtal, Exact reconstruction of
signals in evolutionary systems via spatiotemporal trade-off,
Journal of Fourier Analysis and Applications, 21 (2015), pp.
11-31.
A. Aldroubi and K. Grochenig, Nonuniform sampling and re-
construction in shift-invariant spaces, SIAM Rev., 43 (2001),
pp. 585-620 (electronic).
Thank you!

Some Aspects of Localization Frames: Bhabhuti Chauhan

Uploaded by

Copyright:

Available Formats

Some Aspects of Localization Frames: Bhabhuti Chauhan

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

Some Aspects of Localization Frames: Bhabhuti Chauhan

Uploaded by

Copyright:

Available Formats

Some Aspects of Localization Frames

August 30, 2022

1 Chapter 1: Introduction and Preliminaries

1 Chapter 4: Localization of Frames, Banach Frames

The science of localization started in 1940s with Claude

The science of localization started in 1940s with Claude

both side take inner product with eJ , J = 1, 2, . . . , y then

We say that F is localized with respect E if the magnitudes of the

has a decay of the diagonal.

and like wise for N , whenX is sued to index a family of func-

with the usual modification for p = ∞. The weight m is a

⟨fx , ey ⟩| ≤ raX (x )−aY (y ) .

O. Christensen, An introduction to frames and Riesz

In this chapter, we study localized frames and its analysis an

|⟨ex · gn | ≤ c(1 + |x − n|)−s

max{|⟨ex , gn ⟩|, |⟨ex , g̃n ⟩|} ≤ ce −α|x −n| ∀ x ∈ X, n ∈ N

{sup |aX−1 (J)|, sup |aY−1 (g)|} ≤ K < ∞

and r ∈ ℓP (G) such that for all x ∈ X , y ∈ Y

|⟨fx , ey ⟩| ≤ rax (x )−ay (y ) (2)

F ∈ S ′ and F is a symmetrically ℓ′ -self-localized frame of H, then

∀ i, j ∈ I |⟨fi , f j ⟩| ≤ ra(i) − a(j) (3)

comparing eq. (2) and (3) we get

This chapter is based on the paper

In this chapter we study the Banach frames, decay in the

we say that F is localized with respect to E if the magnitude of the

|⟨fx , ey ⟩| ≤ C (1 + |x − y |)−5 and |⟨fx , ẽy ⟩| ≤ C (1 + |x − y |)−s .

Likewise, F is called exponentially localized with exponent α > 0 if

|⟨fx , ey ⟩| ≤ Ce−α|x −y | and |⟨fx , ẽy ⟩| ≤ Ce −α|x −y | .

To prove that every polynomially localized frame is symmetrically lo-

C. Chui, W. Czaja, M. Maggioni and G. Weiss, Characteriza-

In this chapter, we collect the main properties of polyno-

and likewise for N .

Polynomial and Sub-Exponential Weight Functions

m(t + x ) ≤ C (1 + |t|)s m(x ) for all t, x ∈ Rd (4)

The weight function occurring in (4),

vs (t) = (1 + |t|)s for t ∈ Rd , (5)

is submultiplicative, i.e., vs satisfies vs (t + x ) ≤ vs (t)vs (x ).

By setting x = 0 in (4) and in (6) we see that a an s-moderate weight m grows at

with the usual modification for p = ∞.

Clearly this estimate is independent of v , hence the assertion.

(1 + |x − n|)−s (1 + |x − m|)−s ≤ C (1 + |n − m|)−s

(b) There exists α′ ∈ (0, α) such that

This chapter is based on the paper:

The fundamental structural feature of a frame that is not a Riesz

(b) (F, a) is ℓ1 -self-localized, and

R. Aceska, A. Aldroubi, J. Davis and A. Petersyan, Dynamical

You might also like