+ All Categories
Home > Documents > Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian...

Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian...

Date post: 22-Dec-2015
Category:
Upload: juliana-gardner
View: 215 times
Download: 0 times
Share this document with a friend
46
Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science
Transcript
Page 1: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Lower Bounds for Depth Three Circuits with small bottom fanin

Neeraj Kayal Chandan SahaIndian Institute of Science

Page 2: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

A lower bound

Theorem: Consider representations of a degree d polynomial of the form

If the ’s have degree one and at most variables each then there is an explicit (family) of polynomials on variables such that s is at least .

Page 3: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

A lower boundTheorem: Consider representations of a degree d polynomial of the form

If the ’s have degree one and at most variables each then there is an explicit (family) of polynomials on variables such that s is at least .

Remark: For a generic , s must be at least

Any asymptotic improvement in the exponent will imply .

Page 4: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

A lower boundCorollary (via Kumar-Saraf): Consider representations of a degree d polynomial of the form

If the ’s have degree one and at most variables each then there is an explicit (family) of polynomials on variables such for s is at least .

Remark: Bad news.

Good news.

Page 5: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Background/Motivation

Page 6: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Arithmetic Circuits

…𝑥2𝑥1𝑥𝑛

Page 7: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

     …

-1

𝑥1+𝑥2 𝑥1−𝑥2

𝑥2𝑥1𝑥𝑛

Arithmetic Circuits

Page 8: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

     …

     …

-1

𝑥1+𝑥2 𝑥1−𝑥2

𝑥2𝑥1𝑥𝑛

Arithmetic Circuits

Page 9: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

     …

     …

-1

𝑥2𝑥1𝑥𝑛

Arithmetic Circuits

Page 10: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

     …

     …

-1

𝑥2𝑥1𝑥𝑛

Arithmetic Circuits

Page 11: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

     …

     …

-1

 

)(xHn

𝑥2𝑥1𝑥𝑛

Page 12: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

     …

     …

-1

 

)(xHn

𝑥2𝑥1𝑥𝑛

Size = Number of Edges

Page 13: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

     …

     …

-1

 

)(xHn

𝑥2𝑥1𝑥𝑛

Depth

Page 14: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

     …

     …

-1

 

)(xHn

𝑥2𝑥1𝑥𝑛

This talk.

> Field is .

> Gates have unbounded fanin

>

Page 15: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Two Fundamental Questions

Can explicit polynomials be efficiently computed?

Can computation be efficiently parallelized?

Page 16: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Two Fundamental Questions

Can explicit polynomials be efficiently computed? Does VP equal VNP?

Can computation be efficiently parallelized? Can every efficient computation be also done by small, shallow circuits?

Page 17: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Can computation be efficiently parallelized?

Question: How efficiently can we simulate circuits of

size by circuits of depth ?

Page 18: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Can computation be efficiently parallelized?

Theorem (Hyafil79, VSBR83, AV-08, Koiran-12,

GKKS13, Tavenas-13, Wigderson/Tavenas): Any

circuit of size s and degree can be simulated

by a -depth circuit of size

Page 19: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Can computation be efficiently parallelized?

Theorem (Hyafil79, VSBR83, AV-08, Koiran-12,

GKKS13, Tavenas-13, Wigderson/Tavenas): Any

circuit of size s and degree can be simulated

by a -depth circuit of size … also by a regular,

homogeneous -depth circuit of size

Page 20: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Can computation be efficiently parallelized?

Theorem (Hyafil79, VSBR83, AV-08, Koiran-12,

GKKS13, Tavenas-13, Wigderson/Tavenas): Any

circuit of size s and degree can be simulated

by a -depth circuit of size … also by a regular,

homogeneous -depth circuit of size

Question: Is this optimal?

Page 21: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Can computation be efficiently parallelized?

Theorem (Hyafil79, VSBR83, AV-08, Koiran-12,

GKKS13, Tavenas-13, Wigderson/Tavenas): Any

circuit of size s and degree can be simulated

by a -depth circuit of size … also by a regular,

homogeneous -depth circuit of size

Question: Is this optimal?

(KS15 + Ramprasad): For yes, but with caveats.

Page 22: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Can computation be efficiently parallelized?

Theorem (Hyafil79, VSBR83, AV-08, Koiran-12,

GKKS13, Tavenas-13, Wigderson/Tavenas): Any

circuit of size s and degree can be simulated

by a -depth circuit of size … also by a regular,

homogeneous -depth circuit of size

Corollary: Strong enough lower bounds for low-depth circuits imply VP VNP.

Page 23: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

A possible way to approach VP vs VNP

• Strong enough lower bounds for low-depth circuits imply VP VNP.

Low Depth Circuit (’s simple)

Low-depth circuits are easy to analyze..

Page 24: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

A possible way to approach VP vs VNP

• Strong enough lower bounds for low-depth circuits imply VP VNP.

Low Depth Circuit (’s simple)

Low-depth circuits are easy to analyze..

Lots of work on lower bounds for low depth

arithmetic circuits in recent years – hope to

discover general patterns and technical

ingredients

Page 25: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

’s are: Lower Bound

has degree ~ in VNP GKKS13+KSS14

has degree ~ IMM FLMS14

is sparse ~ in VNP KLSS14

is sparse ~ IMM KS14

has degree one and arity IMM This work

have homogeneous depth three circuits of bottom fanin

in VNP This work

have homogeneous depth five circuits of bottom fanin

IMM Next talk

Page 26: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

A possible way to approach VP vs VNP

Prove strong enough lower bounds for low depth

circuits.

Low Depth Circuit (’s simple)

Low-depth circuits are easy to analyze..

Lots of work on low depth arithmetic circuits recently

This talk:

• common pattern/proof strategy

• Technical ingredients

Page 27: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

A common Proof Strategy and some technical

ingredients

Page 28: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Proof Strategy (’s simple ). Let .

shallow circuit C

Find a geometric property GP of the ’s.

Express the property GP in terms of rank of a “big” matrix M: if T has the property than rank(M(T)) is “relatively small”

Show that rank(M()) is “relatively large”.

Page 29: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Proof Strategy (’s simple ). Let .

shallow circuit C

Find a geometric property GP of the ’s.

Express the property GP in terms of rank of a “big” matrix M: if T has the property than rank(M(T)) is “relatively small”

Show that rank(M()) is “relatively large”.

Page 30: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Lower Bounding rank of large matrices

• If a matrix M(f) has a large upper triangular submatrix, then it has large rank

• (Alon): If the columns of M(f) are almost orthogonal then M(f) has large rank.

Page 31: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

When the ’s have low degree (’s have low degree). Let .

shallow circuit C

Find a geometric property GP of the ’s.

Express the property GP in terms of rank of a “big” matrix M: if T has the property than rank(M(T)) is “relatively small”

Show that rank(M()) is “relatively large”.

Page 32: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Finding a geometric property GP of T

is a product of low degree polynomials

V(T) is a union of low-degree hypersurfaces

V(T) has lots of high-order singularities

Page 33: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Finding a geometric property GP of T

𝑇 (𝑥 , 𝑦 )=𝑔(𝑥 , 𝑦) ∙ h(𝑥 , 𝑦)

Page 34: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Finding a geometric property GP of T

is a product of low degree polynomials

V(T) is a union of low-degree hypersurfaces

V(T) has lots of high-order singularities

V( T) has lots of points

Page 35: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

When the ’s have low degree (’s have low degree). Let .

shallow circuit C

Find a geometric property GP of the ’s.

Express the property GP in terms of rank of a “big” matrix M: if T has the property than rank(M(T)) is “relatively small”

Show that rank(M()) is “relatively large”.

Page 36: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

When the ’s have low degree (’s have low degree). Let .

shallow circuit C

Find a geometric property GP of the ’s.

Express the property GP in terms of rank of a “big” matrix M: if T has the property than rank(M(T)) is “relatively small”

Show that rank(M()) is “relatively large”.

Page 37: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Expressing largeness of a variety in terms of rank

is a variety.

Let = set of degree- polynomials.

Page 38: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Expressing largeness of a variety in terms of rank

is a variety.

Let = set of degree- polynomials. Let = set of degree- polynomials which vanish at every point of V.

Hilbert’s Theorem (Informal): If V is “large” then has small dimension.

Page 39: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Expressing largeness of a variety in terms of rank

is a variety.

Let = set of degree- polynomials. Let = set of degree- polynomials which vanish at every point of V.

Hilbert’s Theorem (Informal): If V is “large” then has small dimension.

Let = { () of deg } .

Hilbert’s Theorem (Formal): If V has dimension r then has asymptotic dimension .

Page 40: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

When the ’s have low degree (’s have low degree). Let .

shallow circuit C

Find a geometric property GP of the ’s.

Express the property GP in terms of rank of a “big” matrix M: if T has the property than rank(M(T)) is “relatively small”

Show that rank(M()) is “relatively large”.

Page 41: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Restrictions

Example: is a function. f(0, 1, z) is a restriction of f. Geometrically, f(0, 1, z) is same as restricting f to the axis-parallel line W = V(x, y-1). Algebraically, f(0, 1, z) is same as mod

More generally: Let be any ideal and a polynomial. Call mod as an algebraic restriction of f.

Obs: Hilbert’s theorem can be generalized for algebraic restrictions of polynomials.

Page 42: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Employing restrictions

Lemma (KLSS14): If Q is a sparse polynomial then for a suitable random algebraic restriction mod is a low degree polynomial.

Yields lower bounds for homogeneous depth four (KLSS14 and KS14).

Page 43: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Lemma (KLSS14): If T is a sparse polynomial a sum of product of low arity polynomials then for a suitable random algebraic restriction mod is a low degree polynomial.

Employing Restrictions

Page 44: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Lemma: If T is a sparse polynomial a sum of product of low arity polynomials then for a suitable random algebraic restriction mod is a low degree polynomial.

Employing Restrictions

Yields lower bounds for homogeneous depth five with low bottom fanin (KS15 and BC15).

Page 45: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Lemma (Shpilka-Wigderson): A depth three circuit C of size s can be converted to a homogeneous depth circuit C’ of size . Further if C has bottom fanin t then C’ also has bottom fanin t.

A lemma by Shpilka and Wigderson

Yields lower bounds mentioned earlier.

Page 46: Lower Bounds for Depth Three Circuits with small bottom fanin Neeraj Kayal Chandan Saha Indian Institute of Science.

Conclusion

Proving lower bounds for low depth circuits is a potential way to prove lower bounds for more general circuits.

There is a meta-strategy common to many recent (and older) lower bounds. We don’t understand the power or limitations of this meta-strategy.

Open: lower bounds for homogeneous depth three circuits for polynomials of degree .


Recommended