# OpenStax_CNX

You are here: Home » Content » Compressive Sensing » Thresholding and Greedy Bases

### Lenses

What is a lens?

#### Definition of a lens

##### Lenses

A lens is a custom view of the content in the repository. You can think of it as a fancy kind of list that will let you see content through the eyes of organizations and people you trust.

##### What is in a lens?

Lens makers point to materials (modules and collections), creating a guide that includes their own comments and descriptive tags about the content.

##### Who can create a lens?

Any individual member, a community, or a respected organization.

##### What are tags?

Tags are descriptors added by lens makers to help label content, attaching a vocabulary that is meaningful in the context of the lens.

#### Affiliated with (What does "Affiliated with" mean?)

This content is either by members of the organizations listed or about topics related to the organizations listed. Click each link to see a list of all content affiliated with the organization.
• Rice Digital Scholarship

This collection is included in aLens by: Digital Scholarship at Rice University

Click the "Rice Digital Scholarship" link to see all content affiliated with them.

#### Also in these lenses

• Lens for Engineering

This module and collection are included inLens: Lens for Engineering
By: Sidney Burrus

Click the "Lens for Engineering" link to see all content selected in this lens.

### Recently Viewed

This feature requires Javascript to be enabled.

Inside Collection (Course):

# Thresholding and Greedy Bases

Module by: Jason Laska, Ronald DeVore. E-mail the authors

We shall next discuss some notions related to best nn-term approximation.

## Thresholding

1. Let XX be a Hilbert space. Given ff, let Λ ϵ (f)={j:|c j (f)|>ϵ}Λ ϵ (f)={j:|c j (f)|>ϵ}. The thresholding operator T ϵ T ϵ is defined by
T ϵ f= jΛ ϵ (f) C j (f)φ j .T ϵ f= jΛ ϵ (f) C j (f)φ j .
(1)
It is easy to see that for each ϵϵ, T ϵ fT ϵ f is the best approximation to ff using NN terms where NN is the cardinality of Λ ϵ Λ ϵ :
f-T ϵ X =σ N (f) X .f-T ϵ X =σ N (f) X .
(2)
Thresholding is easily implemented on a computer.
2. The thresholding scheme above can be generalized if XX is not a Hilbert space provided the dictionary has some specific structure. For example, when
1. The dictionary is the wavelet basis and X=L p X=L p , 1<p<1<p<.
2. X=l p X=l p and the dictionary is the canonical basis δ j =φ j δ j =φ j . ex: (0,0,...,1,0)(0,0,...,1,0)
3. For a general Banach space XX and the dictionary (φ j )(φ j ) is a greedy basis.

## Greedy Bases

We briefly describe the notion of greedy basis.

### definition 1

Given XX, we say (φ j )(φ j ) is a greedy basis for XX if for each ϵ>0ϵ>0,

f-T ϵ f X C(X)σ N (f) X f-T ϵ f X C(X)σ N (f) X
(3)

where NN is the cardinality of Λ ϵ Λ ϵ .

### definition 2

A basis φ j φ j is said to be unconditional if

±c j φ j X Cc j φ j X ±c j φ j X Cc j φ j X
(4)

or equivalently

c j φ j X Cd j φ j X where |c j ||d j |.c j φ j X Cd j φ j X where |c j ||d j |.
(5)

This is an older concept from functional analysis. In words, this definition says that if the terms c j c j are rearranged, the series c j φ j c j φ j will still converge. This is not generally true for all bases.

### definition 3

A basis φ j φ j is said to be democratic if

jΛ φ j C jΛ ' φ j , jΛ φ j C jΛ ' φ j ,
(6)

where the cardinality of Λ ' Λ ' equals the cardinality of ΛΛ.

### remark 1

(φ j )(φ j ) greedy (φ j )(φ j ) is both unconditional and democratic.

Some examples involving the last two definitions:

• The fourier basis in L p L p is not democratic, but is unconditional for l<p<l<p<.
• The wavelet basis contains both of these properties, and is therefore greedy.

### Theorem 1

If X=L p X=L p has (φ j )(φ j ) greedy, B={φ j }B={φ j }, f= j=1 c j (f)φ j f= j=1 c j (f)φ j , c j (f)=f,ψ j c j (f)=f,ψ j where ψ j ψ j is a dual basis,

fA r (B)(c j (f))w l τ ,1 τ=r+1 pfA r (B)(c j (f))w l τ ,1 τ=r+1 p
(7)

and

f A r c j (f) l τ f A r c j (f) l τ
(8)

Let us now consider a specific setting that we shall be concerned with a lot in this course. We shall examine some of the concepts we have introduced in the finite dimensional space of of all sequence (points) in R N R N . Recall that we can put many different norms on this space including the p p norms and the weak p p norms.

### remark 2

Given a vector =(x 1 ,x 2 ,...,x N )R N =(x 1 ,x 2 ,...,x N )R N . The best approximation to xx from Σ n Σ n in the p p norm is to take the vector in Σ n Σ n which shares the nn largest values of xx. Its error of approximation satisfies

σ n (x) p Cn -r x w l τ ,1 τ=r+1 pσ n (x) p Cn -r x w l τ ,1 τ=r+1 p
(9)

### remark 3

x w l τ x l τ ,1 τ=r+1 px w l τ x l τ ,1 τ=r+1 p.

### example 1

For p=1p=1 and r=3r=3, σ n (x) 1 Cn -3 x w l τ σ n (x) 1 Cn -3 x w l τ and 1 τ=41 τ=4. In words, this equation shows what kind of ττ is needed for a given decay rate (or given some ττ, what kind of decay rate will be achieved) to approximate with certain ability.

### example 2

Show σ n (x) l p Cn -r x l τ σ n (x) l p Cn -r x l τ holds with C=1C=1.

Proof: Let Λ n :={i:|x i | largest }Λ n :={i:|x i | largest },

σ n (x) l p p = iΛ n |x i | p σ n (x) l p p = iΛ n |x i | p
(10)
iΛ n |x i | p-τ |x i | τ iΛ n |x i | p-τ |x i | τ
(11)
(x w l τ n -1 τ ) p-τ (|x i | τ )(x w l τ n -1 τ ) p-τ (|x i | τ )
(12)
x l τ p-τ x l τ τ n=n -rp x l τ p x l τ p-τ x l τ τ n=n -rp x l τ p
(13)

and so

σ n (x) l p p n -rp x l τ p σ n (x) l p p n -rp x l τ p
(14)

σ n (x) l p n -r x l τ σ n (x) l p n -r x l τ

### remark 4

For X=L p X=L p , {φ j }{φ j } a wavelet basis, we can say wavelet coefficients of ff are in l p l p is equivalent to ff is in a certain Besov class (roughly speaking ff has rr derivatives and f (r) L τ f (r) L τ ). We refer the reader to (Reference) for precise formulations of results of this type.

## Content actions

PDF | EPUB (?)

### What is an EPUB file?

EPUB is an electronic book format that can be read on a variety of mobile devices.

PDF | EPUB (?)

### What is an EPUB file?

EPUB is an electronic book format that can be read on a variety of mobile devices.

#### Collection to:

My Favorites (?)

'My Favorites' is a special kind of lens which you can use to bookmark modules and collections. 'My Favorites' can only be seen by you, and collections saved in 'My Favorites' can remember the last module you were on. You need an account to use 'My Favorites'.

| A lens I own (?)

#### Definition of a lens

##### Lenses

A lens is a custom view of the content in the repository. You can think of it as a fancy kind of list that will let you see content through the eyes of organizations and people you trust.

##### What is in a lens?

Lens makers point to materials (modules and collections), creating a guide that includes their own comments and descriptive tags about the content.

##### Who can create a lens?

Any individual member, a community, or a respected organization.

##### What are tags?

Tags are descriptors added by lens makers to help label content, attaching a vocabulary that is meaningful in the context of the lens.

| External bookmarks

#### Module to:

My Favorites (?)

'My Favorites' is a special kind of lens which you can use to bookmark modules and collections. 'My Favorites' can only be seen by you, and collections saved in 'My Favorites' can remember the last module you were on. You need an account to use 'My Favorites'.

| A lens I own (?)

#### Definition of a lens

##### Lenses

A lens is a custom view of the content in the repository. You can think of it as a fancy kind of list that will let you see content through the eyes of organizations and people you trust.

##### What is in a lens?

Lens makers point to materials (modules and collections), creating a guide that includes their own comments and descriptive tags about the content.

##### Who can create a lens?

Any individual member, a community, or a respected organization.

##### What are tags?

Tags are descriptors added by lens makers to help label content, attaching a vocabulary that is meaningful in the context of the lens.

| External bookmarks