Download presentation

Presentation is loading. Please wait.

Published byZaria Ledger Modified over 3 years ago

1
Simple Affine Extractors using Dimension Expansion. Matt DeVos and Ariel Gabizon

2
Vague Definition: A pseudorandom object(e.g. graph, function) has some nice property a random object would have with high probability. For example: A graph that has no large cliques or large independent sets. The field of pseudorandomness aims to explicitly construct pseudorandom objects. Pseudorandomness

3
Efficient Det. Alg. Explicitly constructing pseudorandom objects bad objects Universe of exp(n) objects good object

4
Why do we want to explicitly construct pseudorandom objects? -Insight into the computational power(lessnes) of randomness -Useful tools in derandomizing algorithms (good example-expanders!) Still, is constructing pseudorandom objects more meaningful than making money, or trying to become famous? Thm: Pseudorandomness is meaningless Theoretical Computer Science is meaningless

5
NP machine P NP by explicitly constructing pseudorandom objects functions with poly-size circuits functions on n bits function in NP without poly-size circuits

6
The nice property can usually be phrased as avoiding a not too large set of bad events. Example: A function of high circuit complexity avoids the event `being computed by circuit C’ for all small circuits C. Circuits are hard to understand – let’s first work with bad events that are easier to understand. The bad event in this paper – a function that is biased on an affine subspace.

7
Affine Extractors

8
Finite field F, with |F|=q (q=p l for prime p) Vector Space F n An affine extractor is a coloring of F n such that any large enough affine subspace is colored in a balanced way For simplicity assume only 2 colors FnFn

9
Just to make sure.. An affine subspace X µ F n of dim. k Defined by vectors a (1),…,a (k),b 2 F n where a (1),…,a (k) are independent X={ (j=1 to k) t j ¢ a (j) + b|t 1,…,t k 2 F}

10
Now, more formally.. An affine extractor for dim k, field size q and error ² is a function D:F n {0,1} such that for any affine subspace X µ F n of dim k |Pr x X (D(x) =1 ) - ½| · ² (We will omit ² from now on, think of it as 1/100) Intuition: D `extracts’ a random bit for the uniform distribution on X. 1/100

11
Feeling the parameters.. k-dimension of subspace q- field size k larger problem easier (need to be unbiased only on larger subspaces) q smaller problem harder(subspaces have less structure - are closed under scalar multiplication from smaller field) Random function D:F n {0,1} is w.h.p an affine extractor when q=2 and k = 5 ¢ logn

12
Previous results and ours: (explicit) G-Raz: Affine Extractor for all k ¸ 1, when q>n 2. Bourgain: Affine Extractor for k= ® ¢ n, for any constant ® >0, and q=2. (exponentially small error) Our result: Affine Extractor for all k ¸ 1, when q= ((n/k) 2 ) Simple Construction and Proof! However: need char(F)= (n/k) (have weaker result for arbitrary characteristic)

13
Warm Up Suppose q>n. How can we get a function f:F n F that is non-constant on lines? i.e, for every a 0, b 2 F n want g(t), f(a ¢ t + b) = f(a 1 ¢ t + b 1,…,a n ¢ t + b n ) to be a non-constant function

14
Answer: Take f(x 1,..,x n ) = i=1 to n) x i i. g(t), f(a ¢ t + b) = i=1 to n) (a i ¢ t + b i ) i Note: a i 0 for some i. Suppose that a n 0. g(t) is a non-constant polynomial of degree n. as q>n, this is a non-constant function on F. (from G-Raz)

15
Quadratic Residue Function: QR:F {0,1}, QR(a) = 1 $9 b 2 F such that b 2 =a Thm[Weil]: Let F be a field of odd size q. Let g(t) be a non-constant polynomial over F of odd degree d. Choose t 2 F randomly.. QR(g(t)) has bias at most d/q 1/2 works for multivariate g too.. Weil’s Theorem

16
Subspace X of dim k defined by a (1),…,a (k),b For f:F n F, define f| X (t 1,..,t k ) = f( (j=1 to k) t j ¢ a (j) + b ) Using Weil: Poly f(X 1,..,X n ) of degree d such that: f| X constant for all X of dim k Affine Extractor for dim k and q » d 2

17
`trick’: Using this view can multiply vectors x,y 2 (F q ) n - not just add them! Vector Space\Field Duality

18
Fix 1-1 Φ:(F q ) n -->F q n s.t. ∀ a,b ∈ F q n s,t ∈ F q : Φ(at+ bs) = Φ(a)∙t + Φ(b)∙s We identify the source output with an element of F q n : ∑a j ∙t j +b --> Φ[∑ a j ∙t j +b] =∑Φ(a j )∙t j +Φ(b) (as t j ∈ F q ) our source coincides with a multivariate polynomial with coeff in F q n (from now omit Φ and think of a j ∈ F q n ) Viewing the source over the `big’ field

19
Suppose we allow f| X to have coeff. in the `big field’ F q n can take f(x) = x. For any subspace X f| X (t 1,..,t k ) = (j=1 to k) a j ¢ t j + b is non- constant. but to use Weil need f| X with coeff. in F q Idea- if coeff. of f| X span F q n. over F q – we can `project down to F q ’ without becoming zero\constant

20
A,B linear subspaces in F q n Dfn: A ¢ B, span{a ¢ b|a 2 A, b 2 B} (enough to take products of basis elements) [Heur-Lieng-Xiang] Suppose n is prime. Then dim(A ¢ B) ¸ min{dim(A)+dim(B)-1,n} (analogous to the classic Cauchy-Davenport on Z p ) `dimension expansion of products of subspaces’

21
Thm: Suppose n is prime. Let T: F q n F q be any non-trivial F q -linear map. Let d=n/(k-1). Suppose Char(F)>d. Let f(x)=T(x d ). Then for any affine subspace X of dim k, f| X is a non-constant poly of degree d with coeff in F q. Proof idea: When Char(F) is large enough, coefficients of f| X are `independent products’ of basis elements.

24
Open question: Similar results over F 2 Relates to following: n is prime.V a linear subspace of dim k in (F 2 ) n, k>min{100logn,n/100}. t= ┌ 2n/k ┐. V t ={x 1+2+4+..+2^{t} | x 2 V}. Show that V t spans (F 2 ) n over F 2.

25
Cauchy – Davenport A,B ½ Z p A+B, {a+b| a 2 A, b 2 B} C-D: |A+B| ¸ min{|A|+|B|-1,p}

26
Proof: Induction on |A|. |A|=1 : |A+B| = |B| (=|A|+|B|-1) Induction step: Assume first that ; ( A Å B ( A Using Inclusion-Exclusion + Ind. Hyp |A Å B + A [ B| ¸ min{|A Å B| + |A [ B| -1,p} = min{|A| +|B| -1,p} Done as A Å B + A [ B ½ A+B

27
justify assumption ; ( A Å B ( A: w.l.g: 0 2 A,B (can replace A by –a +A, for some a 2 A. This does not change |A+B|) |A|>1, so can fix 0≠a 2 A. If B=Z p we are done. Otherwise, fix first c s.t. c ∙ a ∉ B. Replace B by –(c-1) ∙ a + B. We have 0 2 B but a ∉ B. (which justifies above assumption)

Similar presentations

Presentation is loading. Please wait....

OK

CS151 Complexity Theory Lecture 6 April 15, 2004.

CS151 Complexity Theory Lecture 6 April 15, 2004.

© 2018 SlidePlayer.com Inc.

All rights reserved.

Ads by Google

Ppt on fmcg industry in india Ppt on cross site scripting xss Ppt on fire extinguisher types colors Dentist appt on your birthday Ppt on intelligent manufacturing systems Ad mad show ppt on tv Ppt on south african culture tattoos Ppt on cross site scripting testing Ppt on business environment nature concept and significance of ash Download ppt on school management system