BK:Section 3: Difference between revisions

From Polymath Wiki
Jump to navigationJump to search
No edit summary
Olof (talk | contribs)
 
(9 intermediate revisions by 2 users not shown)
Line 1: Line 1:
Parent page: [[Improving the bounds for Roth's theorem]]


Parent page: [[Improving the bounds for Roth's theorem]]
One of the take-away results from Section 3 of the Bateman-Katz paper is Proposition 3.1, an important part of which is in some places referred to as the "nd-estimate". The rough reason for this terminology is that it says that a set <math>A</math> in <math>\mathbb{F}_3^n</math> of density about <math>1/n</math> either has a `good' density increment on a subspace of codimension <math>d</math>, or else the <math>(1/n)</math>-large spectrum of <math>A</math> intersects any <math>d</math>-dimensional subspace in at most about <math>nd</math> points. We shall say later on why this is significant.
 
==The nd-estimate==
 
Here is the precise result, stated in slightly different terms to the paper in order to illustrate how it relates to other results. For a subspace <math>V \leq \mathbb{F}_3^n</math> we write
:<math>V^{\perp} = \{ \gamma \in \widehat{\mathbb{F}_3^n} : \gamma(x) = 1 \ \forall x \in V \}</math>
for its annihilator (cf. [[Basic facts about Bohr sets|the section on Bohr sets]]).
 
:'''Proposition 1''' Let <math>A \subset \mathbb{F}_3^n</math> be a set with density <math>\alpha</math>, and let <math>0 \leq \delta, \eta \leq 1</math> be parameters. Set
:<math>\Delta = \{ \gamma \in \widehat{G} : | \widehat{1_A}(\gamma) | \geq \delta \alpha \} \setminus \{ 0_{\widehat{\mathbb{F}_3^n}} \}</math>.
:Let <math>V \leq \mathbb{F}_3^n</math> be a subspace. Then
:* either <math>A</math> has density at least <math>\alpha(1 + \eta)</math> on a translate of <math>V</math>,
:* or <math>|\Delta \cap V^{\perp}| \leq 3\eta \delta^{-2}</math>; in fact <math>\sum_{\gamma \in V^{\perp}} |\widehat{(1_A - \alpha)}(\gamma)|^2 \leq 3\eta \alpha^2</math>.


One of the take-away results from Section 3 of the Bateman-Katz paper is Proposition 3.1, which is in some places referred to as the "nd-estimate". The rough reason for this terminology is that it says that a set <math>A</math> in <math>\mathbb{F}_3^n</math> of density about <math>1/n</math> either has a `good' density increment on a subspace of codimension <math>d</math>, or else the <math>(1/n)</math>-large spectrum of <math>A</math> intersects any <math>d</math>-dimensional subspace in at most about <math>nd</math> points.
In other words, if the large spectrum of a set is somewhat concentrated in some subspace then one can find a density increment on a translate of the annihilator of that subspace.


Here is the precise result, stated in slightly different terms to the paper in order to illustrate how it relates to other results.
==Proof==
Let us write <math>\mu_V = \frac{|\mathbb{F}_3^n|}{|V|}1_V</math> for the scaled indicator function of <math>V</math> normalized so that <math>\mathbb{E}_x \mu_V(x) = 1</math>. If
:<math>1_A*\mu_V(x) > \alpha(1 + \eta)</math>
for some <math>x \in \mathbb{F}_3^n</math> then we are in the first case of the conclusion, so let us assume that <math>1_A*\mu_V \leq \alpha(1+\eta)</math>. Write <math>f = 1_A - \alpha</math> for the balanced function of <math>A</math>. Then
:<math> | \Delta \cap V^{\perp} | \delta^2 \alpha^2 \leq \sum_{\gamma \in V^{\perp}} |\widehat{f}(\gamma)|^2 = \sum_{\gamma \in \widehat{\mathbb{F}_3^n}} |\widehat{f}(\gamma)|^2 |\widehat{\mu_V}(\gamma)|^2.</math>
By Parseval's identity, this equals
:<math> \mathbb{E}_{x \in \mathbb{F}_3^n} f*\mu_V(x)^2 = \mathbb{E}_{x \in \mathbb{F}_3^n} 1_A*\mu_V(x)^2 - \alpha^2 \leq \alpha^2(2\eta + \eta^2),</math>
which proves the result.


:'''Proposition 1''' Let <math>A</math> be a subset of <math>\mathbb{F}_3^n</math> with density <math>\alpha</math>, and let <math>\delta > 0</math> and <math>0 \leq \eta \leq 1</math> be parameters. Set <math>\Delta = \{ \gamma \in \widehat{G} : | \widehat{1_A}(\gamma) | \geq \delta \alpha \} \setminus \{0\}</math>. Then
==Comparison with other results about the large spectrum of a set==
# either there is a subspace of <math>\mathbb{F}_3^n</math> of codimension <math>d</math> on which <math>A</math> has density at least <math>\alpha(1 + \eta)</math>
The main ingredient in deriving the nd-estimate is Parseval's identity. This identity also has the following useful consequence: letting <math>\Delta</math> be as above, we have
# or <math>|\Delta \cap W| \leq \eta \delta^{-2}</math> for each <math>d</math>-dimensional subspace <math>W \leq \widehat{\mathbb{F}_3^n}</math>.
:<math>|\Delta| \delta^2 \alpha^2 \leq \sum_{\gamma \in \widehat{\mathbb{F}_3^n}} |\widehat{1_A}(\gamma)|^2 = \mathbb{E}_x 1_A(x)^2 = \alpha</math>,
whence
:<math>|\Delta| \leq \alpha^{-1} \delta^{-2}</math>,
which should be compared to the bound on <math>| \Delta \cap V^{\perp} |</math> given by the nd-estimate.  


'''Proof''' Choose a subspace <math>H</math> such that <math>W</math> is the annihilator of <math>H</math>, and let <math>V</math> be a subspace transverse to <math>H</math>. Then for any <math>\gamma\neq0\in W</math>,  
There is another useful result about the large spectrum of a set known as Chang's theorem. Informally, this says that the largest size of a linearly independent set in the large spectrum <math>\Delta</math> cannot be too large; formally, the largest independent set has size at most <math>C\log(\alpha^{-1}) \delta^{-2}</math>. Unfortunately this statement becomes trivial with the parameters needed for the Bateman-Katz argument (<math>\delta \sim \alpha \sim 1/n^{1+\epsilon}</math>). Nevertheless, there is [http://arxiv.org/abs/math/0605689 a generalization of Chang's theorem due to Shkredov] that gives a lower bound for the number of additive <math>(2m)</math>-tuples in the large spectrum of a set, which is used in [[BK:Section 4|Section 4]] of the Bateman-Katz paper.
:<math>\widehat{1_A}(\gamma)=3^{-n}\sum_{v\in V}(| A\cap(H+v)|-3^{-d}| A|)\gamma(v)</math>
and hence
:<math>\sum_{\gamma\neq0\in W}|\widehat{1_A}(\gamma)|^2=3^{d-2n}\sum_{v\in V}(| A\cap(H+v)|-3^{-d}| A|)^2.</math>
If we let <math>V^+</math> be the subset of <math>V</math> for which each of the squared summands is positive, then either <math>A</math> has the required density increment on a translate of <math>H</math> (which has codimension <math>d</math>), or
:<math>|| A\cap(H+v)|-3^{-d}| A||\ll 3^{-d}| A|\eta.</math>
Hence
:<math>\sum_{v\in V^+}|| A\cap(H+v)|-3^{-d}| A||\ll| A|\eta</math>
and
:<math>\sum_{v\in V^+}|| A\cap(H+v)|-3^{-d}| A||^2\ll 3^{-d}| A|^2\eta^2.</math>
Furthermore, since
:<math>\sum_{v\in V}|| A\cap(H+v)|-3^{-d}| A||=0</math>
defining <math>V^-</math> similarly and combining the trivial estimate
:<math>|| A\cap(H+v)|-3^{-d}| A||\leq3^{-d}| A|</math>
for <math>v\in V^-</math> with the above gives
:<math>\sum_{v\in V^-}|| A\cap(H+v)|-3^{-d}| A||^2\ll3^{-d}| A|^2\eta.</math>
Combining these sum estimates gives
:<math>\sum_{v\in V}|| A\cap(H+v)|-3^{-d}| A||^2\ll3^{-d}| A|^2\eta</math>
and hence
:<math>\sum_{\gamma\neq0\in W}|\widehat{1_A}(\gamma)|^2\ll \alpha^2\eta.</math>
Recalling the definition of <math>\Delta</math>, we have
:<math>|\Delta\cap W|\delta^2\alpha^2\ll\sum_{\gamma\in\Delta\cap W}|\widehat{1_A}(\gamma)|^2\ll\alpha^2\eta.</math>


By contrast, the nd-estimate is something like a statement in the opposite direction: it says that there are quite a lot of linearly independent characters in <math>\Delta</math>, or else there is a density increment. Specifically, if we have picked <math>\gamma_1, \ldots, \gamma_d</math> from <math>\Delta</math>, then
:<math>| \Delta \cap \langle \gamma_1, \ldots, \gamma_d \rangle | \leq 3\eta \delta^{-2}</math>
unless we get a density increment on a (particular) subspace of codimension at most <math>d</math>.
For suitable parameter choices, this says that there are a lot of characters in the large spectrum that are linearly independent of <math>\gamma_1, \ldots, \gamma_d</math>, which is very important in [[BK:Section 5|Section 5]] of the paper. (Note: for this type of conclusion to hold we need to know that the large spectrum <math>\Delta</math> is quite large, which follows from the assumption that we may make in the <math>\mathbb{F}_3^n</math> context that <math>A</math> has no particularly large non-trivial Fourier coefficients.)


To be added:
==Relation to Lemma 2.8 in Sanders's paper==
* Statement of size bound on <math>\Delta</math> from Parseval alone
* Statement of Chang's theorem
* Relation to Lemma 2.8 in Sanders's paper

Latest revision as of 20:13, 7 February 2011

Parent page: Improving the bounds for Roth's theorem

One of the take-away results from Section 3 of the Bateman-Katz paper is Proposition 3.1, an important part of which is in some places referred to as the "nd-estimate". The rough reason for this terminology is that it says that a set [math]\displaystyle{ A }[/math] in [math]\displaystyle{ \mathbb{F}_3^n }[/math] of density about [math]\displaystyle{ 1/n }[/math] either has a `good' density increment on a subspace of codimension [math]\displaystyle{ d }[/math], or else the [math]\displaystyle{ (1/n) }[/math]-large spectrum of [math]\displaystyle{ A }[/math] intersects any [math]\displaystyle{ d }[/math]-dimensional subspace in at most about [math]\displaystyle{ nd }[/math] points. We shall say later on why this is significant.

The nd-estimate

Here is the precise result, stated in slightly different terms to the paper in order to illustrate how it relates to other results. For a subspace [math]\displaystyle{ V \leq \mathbb{F}_3^n }[/math] we write

[math]\displaystyle{ V^{\perp} = \{ \gamma \in \widehat{\mathbb{F}_3^n} : \gamma(x) = 1 \ \forall x \in V \} }[/math]

for its annihilator (cf. the section on Bohr sets).

Proposition 1 Let [math]\displaystyle{ A \subset \mathbb{F}_3^n }[/math] be a set with density [math]\displaystyle{ \alpha }[/math], and let [math]\displaystyle{ 0 \leq \delta, \eta \leq 1 }[/math] be parameters. Set
[math]\displaystyle{ \Delta = \{ \gamma \in \widehat{G} : | \widehat{1_A}(\gamma) | \geq \delta \alpha \} \setminus \{ 0_{\widehat{\mathbb{F}_3^n}} \} }[/math].
Let [math]\displaystyle{ V \leq \mathbb{F}_3^n }[/math] be a subspace. Then
  • either [math]\displaystyle{ A }[/math] has density at least [math]\displaystyle{ \alpha(1 + \eta) }[/math] on a translate of [math]\displaystyle{ V }[/math],
  • or [math]\displaystyle{ |\Delta \cap V^{\perp}| \leq 3\eta \delta^{-2} }[/math]; in fact [math]\displaystyle{ \sum_{\gamma \in V^{\perp}} |\widehat{(1_A - \alpha)}(\gamma)|^2 \leq 3\eta \alpha^2 }[/math].

In other words, if the large spectrum of a set is somewhat concentrated in some subspace then one can find a density increment on a translate of the annihilator of that subspace.

Proof

Let us write [math]\displaystyle{ \mu_V = \frac{|\mathbb{F}_3^n|}{|V|}1_V }[/math] for the scaled indicator function of [math]\displaystyle{ V }[/math] normalized so that [math]\displaystyle{ \mathbb{E}_x \mu_V(x) = 1 }[/math]. If

[math]\displaystyle{ 1_A*\mu_V(x) \gt \alpha(1 + \eta) }[/math]

for some [math]\displaystyle{ x \in \mathbb{F}_3^n }[/math] then we are in the first case of the conclusion, so let us assume that [math]\displaystyle{ 1_A*\mu_V \leq \alpha(1+\eta) }[/math]. Write [math]\displaystyle{ f = 1_A - \alpha }[/math] for the balanced function of [math]\displaystyle{ A }[/math]. Then

[math]\displaystyle{ | \Delta \cap V^{\perp} | \delta^2 \alpha^2 \leq \sum_{\gamma \in V^{\perp}} |\widehat{f}(\gamma)|^2 = \sum_{\gamma \in \widehat{\mathbb{F}_3^n}} |\widehat{f}(\gamma)|^2 |\widehat{\mu_V}(\gamma)|^2. }[/math]

By Parseval's identity, this equals

[math]\displaystyle{ \mathbb{E}_{x \in \mathbb{F}_3^n} f*\mu_V(x)^2 = \mathbb{E}_{x \in \mathbb{F}_3^n} 1_A*\mu_V(x)^2 - \alpha^2 \leq \alpha^2(2\eta + \eta^2), }[/math]

which proves the result.

Comparison with other results about the large spectrum of a set

The main ingredient in deriving the nd-estimate is Parseval's identity. This identity also has the following useful consequence: letting [math]\displaystyle{ \Delta }[/math] be as above, we have

[math]\displaystyle{ |\Delta| \delta^2 \alpha^2 \leq \sum_{\gamma \in \widehat{\mathbb{F}_3^n}} |\widehat{1_A}(\gamma)|^2 = \mathbb{E}_x 1_A(x)^2 = \alpha }[/math],

whence

[math]\displaystyle{ |\Delta| \leq \alpha^{-1} \delta^{-2} }[/math],

which should be compared to the bound on [math]\displaystyle{ | \Delta \cap V^{\perp} | }[/math] given by the nd-estimate.

There is another useful result about the large spectrum of a set known as Chang's theorem. Informally, this says that the largest size of a linearly independent set in the large spectrum [math]\displaystyle{ \Delta }[/math] cannot be too large; formally, the largest independent set has size at most [math]\displaystyle{ C\log(\alpha^{-1}) \delta^{-2} }[/math]. Unfortunately this statement becomes trivial with the parameters needed for the Bateman-Katz argument ([math]\displaystyle{ \delta \sim \alpha \sim 1/n^{1+\epsilon} }[/math]). Nevertheless, there is a generalization of Chang's theorem due to Shkredov that gives a lower bound for the number of additive [math]\displaystyle{ (2m) }[/math]-tuples in the large spectrum of a set, which is used in Section 4 of the Bateman-Katz paper.

By contrast, the nd-estimate is something like a statement in the opposite direction: it says that there are quite a lot of linearly independent characters in [math]\displaystyle{ \Delta }[/math], or else there is a density increment. Specifically, if we have picked [math]\displaystyle{ \gamma_1, \ldots, \gamma_d }[/math] from [math]\displaystyle{ \Delta }[/math], then

[math]\displaystyle{ | \Delta \cap \langle \gamma_1, \ldots, \gamma_d \rangle | \leq 3\eta \delta^{-2} }[/math]

unless we get a density increment on a (particular) subspace of codimension at most [math]\displaystyle{ d }[/math]. For suitable parameter choices, this says that there are a lot of characters in the large spectrum that are linearly independent of [math]\displaystyle{ \gamma_1, \ldots, \gamma_d }[/math], which is very important in Section 5 of the paper. (Note: for this type of conclusion to hold we need to know that the large spectrum [math]\displaystyle{ \Delta }[/math] is quite large, which follows from the assumption that we may make in the [math]\displaystyle{ \mathbb{F}_3^n }[/math] context that [math]\displaystyle{ A }[/math] has no particularly large non-trivial Fourier coefficients.)

Relation to Lemma 2.8 in Sanders's paper