Download presentation

Presentation is loading. Please wait.

Published byDonte Parmalee Modified over 2 years ago

1
Approximability & Proof Complexity Ryan O’Donnell Carnegie Mellon

2
Approximability & Proof Complexity Ryan O’Donnell Carnegie Mellon of optimization problems

3
Minimum Vertex-Cover M IN -VC(G) = min {|S| : S ⊆ V, S touches all edges of E}

4
Minimum Vertex-Cover M IN -VC(G) = min {|S| : S ⊆ V, S touches all edges of E}

5
“2-approximating” Min-VC Choose any maximal matching M ⊆ E MIN-VC(G) ≥ |M| Let S = {all endpoints in M}. It’s a vertex-cover (why?) satisfying |S| = 2|M| ≤ 2MIN-VC(G)

6
“Factor 2-certifying” Min-VC Choose any maximal matching M ⊆ E Output “MIN-VC(G) ≥ |M|” 1.Output bound is always correct. 2.Bound is always within factor α of truth. A “factor α-certification” algorithm:

7
Linear programming (LP) relaxation k = minimize: ∑ v ∈ V X v subject to: 0 ≤ X v ≤ 1for all v ∈ V X u + X v ≥ 1 for all (u,v) ∈ E Output “MIN-VC(G) ≥ k”

8
Matching algorithm, LP algorithm are both factor 2-certification algorithms. Are they also 1.99-certification algorithms? No. M IN -VC(K n ) =n−1 maximal |M| = n/2 LP value = n/2 KnKn

9
Matching algorithm, LP algorithm are both factor 2-certification algorithms. Are they also 1.99-certification algorithms? No. M IN -VC(K n ) =n−1 maximal |M| = n/2 LP value = n/2 Is there any poly-time 1.99-certification alg? KnKn

10
We don’t know. Best we know is: ∃ 1.36-certification alg (unless P=NP) [Dinur-Safra’02] Is 1.99-certifying Min-VC NP-hard?

11
Approximability & Proof Complexity

12
Resolution Refutes statements encoded… by boolean disjunctions Cutting planesby integer inequalities Nullstellensatz/ Polynomial Calculus [BIKPP’96,CEI’96] by polynomial equations Positivstellensatz/ Sum-of-Squares (SOS) [Grigoriev-Vorobjov’99] by polynomial inequalities ZFC (“Frege”)in ordinary math language

13
is infeasible if −1 = Q 0 + Q 1 P 1 + Q 2 P 2 + +Q m P m there exist “certifying polynomials” Q 0, …, Q m, each a sum of squares, s.t. we have the identity Positivstellensatz [Krivine’64,Stengle’73,Schmüdgen’91,Putinar’93,Wörmann’98] and only if assuming a mild technical condition

14
“M IN -VC(G) > k” X v 2 ≥ X v X v 2 ≤ X v X u +X v ≥ 1 for all (u,v) ∈ E ∑ v X v ≤ k infeasible −1 = Q 0 + Q 1 (k −∑ X v ) + ∑ Q uv (X u +X v −1) + ∃ certifying SOS polynomials Q such that for all v ∈ V ⇔ ⇔

15
Positivstellensatz / SOS proof system Suggested by Grigoriev and Vorobjov in 1999. The complexity of an SOS proof is the maximum degree of any Q i P i or Q 0. SOS d denotes the proof system restricted to degree d. (No longer complete.)

16
Example proof Theorem: The following system is infeasible: {X 2 ≤ 1, Y 2 ≤ 1, Z 2 ≤ 1, XY+YZ+ZX ≤ −2}. ZFC proof: Let f(X,Y,Z) = XY+YZ+ZX. Suppose X 2 ≤ 1; i.e., X ∈ [−1,1]. Since f is linear in X, it’s maximized if X = ±1. Similarly for Y and Z. Suffices to show f(±1,±1,±1) > −2. If all three inputs same, f is 3. If not all three inputs same, f is −1.

17
Example proof Theorem: The following system is infeasible: {X 2 ≤ 1, Y 2 ≤ 1, Z 2 ≤ 1, XY+YZ+ZX ≤ −2}. SOS d=4 proof: +

18
Show SOS d=4 certifies MIN-VC(K n ) ≥ n−1 (i.e., refutes MIN-VC(K n ) ≤ n−2). Exercise

19
SOS d is ‘automatizable’! Theorem: [Lasserre’00,Parrilo’00, cf. N.Shor’87] If a polynomial inequality system can be refuted in the SOS d proof system, the certifying Q i ’s can be found in n O(d) time (using semidefinite programming).

20
The strongest(?) automatizable proof system that we know SOS d is stronger than: Width-d Resolution Degree-d Nullstellensatz Basic LP relaxations Basic SDP relaxations d/2 rounds of Lovász-Schrijver LP/SDP hierarchy d/2 rounds of Sherali-Adams LP/SDP hierarchy (Doesn’t seem to be stronger than degree-d Polynomial Calculus.)

21
A very powerful poly-time algorithm for Vertex-Cover certification: Output the largest k ∈ [n] such that SOS d=1000 certifies “M IN -VC(G) > k”. Could this be a 1.99-certification algorithm? I.e., is it true that whenever MIN-VC(G) = β, ∃ degree-1000 Q i ’s certifying Min-VC(G) ≥ β/1.99?

22
Partial history of upper and lower bounds for SOS d

23
SOS d upper bounds, 2001-2011 Nothing that we didn’t already know by other means. E.g., SDP is a.878-certification alg for Max-Cut ∵ SDP ≤ SOS d=4 ∴ SOS d=4 also.878-certifies Max-Cut

24
SOS d lower bounds, 1999-2009

25
[Grigoriev’99]: (indep. [Schoenbeck’08]) Consider a random system of O(n) 3-variable equations modulo 2. With very high probability… No assignment sats > 51% of equations Unless d = Ω(n), SOS d cannot refute “the system is totally satisfiable”. Tseitin Tautologies / 3Lin(mod 2)

26
[Grigoriev’01]: See also [Laurent’02], [Cheung’05] “If n is odd and X 1, …, X n satisfy X i 2 = 1, then X 1 + + X n cannot be 0.” Not provable in SOS d unless d ≥ n+1. ‘Knapsack’ (Essentially equivalent: “MAX-CUT(K n ) ≥ ”) Open Problem: Give a pleasant proof that d needs to be at least, say, 6.

27
(A corollary of the 3Lin(mod 2) lower bound.) (Not rigorously proven, but seems true in all cases.) [Tulsiani’09] Rule of Thumb For any factor-α certification problem which we know is NP-hard, there exists instances which require degree-n Ω(1) SOS proofs.

28
A very powerful poly-time algorithm for Vertex-Cover certification: Output the largest k ∈ [n] such that SOS d=1000 certifies “M IN -VC(G) > k”. Could this be a 1.99-certification algorithm?

29
Integrality Gaps [GK’95] SDP does not 2-certify Min-VC:Frankl-Rödl graphs [FS’00] SDP does not.879-certify Max-Cut:Noisy-sphere graphs [KV’05] SDP+∆ does not.879-certify Max-Cut or solve Unique-Games: KV noisy-hypercube graphs [DKSV’06] SDP+∆ does not O(1)-certify Sparsest-Cut (Balanced-Separator): DKSV noisy-hypercube graphs [KS’09,RS’09] Sherali-Adams +, degree-O(1), does not.879-certify Max-Cut or solve Unique-Games: KV noisy-hypercube graphs [BCGM’11] Sherali-Adams +, degree-6, (and probably degree-O(1)) does not 2-certify Min-VC Frankl-Rödl graphs

30
These tricky instances aren’t so hard for SOS!

31
[BBHKSZ’12]: SOS d=4 solves the KV Unique-Games instances! [OZ’13]: SOS d=4 solves the DKSV Balanced-Separator instances. [OZ’13]: SOS d=O(1).95-certifies the KV Max-Cut instances.

32
[BBHKSZ’12]: SOS d=4 solves the KV Unique-Games instances! [OZ’13]: SOS d=4 solves the DKSV Balanced-Separator instances. [DMN’13]: SOS d=O(1) solves the KV Max-Cut instances. [KV’05]: used ZFC to show “MAX-CUT(KV) ≈ 85%” [KS’09,RS’09]: SA + d=O(1) only certify “MAX-CUT(KV) ≥ 75%” [DMN’13] SOS d certifies “MAX-CUT(KV) ≥ 85% − o d (1)”

33
[BBHKSZ’12]: SOS d=4 solves the KV Unique-Games instances! [OZ’13]: SOS d=4 solves the DKSV Balanced-Separator instances. [DMN’13]: SOS d=O(1) solves the KV Max-Cut instances. [KOTZ’13]: SOS d=O(1) solves “most of” the Frankl-Rödl Min-VC instances.

34
The whole result is just that one particular algorithm does well on one particular instance? I have 3 responses.

35
Response 1: an old joke Q:Why did the complexity theorist work on algorithms? A:To get lower bounds on his lower bounds. We basically no longer know any “hard instances”.

36
Response 2: Evidence for algorithmic optimism? [BBHKSZ+’13] points out that as far as we know, SOS d=4 solves the Unique-Games problem (i.e., refutes the UGC). Perhaps SOS d is the killer algorithm for combinatorial optimization.

37
Response 3: New proofs Proving known theorems in restricted proof systems can lead to new insights and proofs. [Razborov’93]: New Switching Lemma proof [BBHKSZ’12, Hypercontractivity insights OZ’13,KOTZ’13]: [BHM’12,KOTZ’13]: New Frankl-Rödl Thm. proof [MN’13,DMN’13]:New Maj. is Stablest proof

38
Let’s take stock Approximation Algs ≤ Proof Complexity: “Is there an efficient algorithm for 100-coloring a 3-colorable graph?”

39
Let’s take stock Approximation Algs ≤ Proof Complexity: “Given a graph that is not 100-colorable, how hard is it to prove that it’s not 3-colorable?” SOS d is a quirky yet strong proof system, automatizable in time n O(d). SOS d=O(1) solves all the trickiest instances we know of Unique-Games, Max-Cut.

40
Three closing thoughts regarding proof complexity 1.Frankl-Rödl graphs and SOS lower bounds 2.The Dynamic SOS proof system 3.My favorite algorithm for Unique-Games

41
Frankl-Rödl graphs FR m (γ):V = {0,1} m E = {(x,y) : Δ(x,y) ≥ (1−γ) m} [Frankl-Rödl’87]: MIN-VC(FR m (γ)) ≥ (1−o(1)) 2 m [KOTZ’13]: SOS d=O(1/γ) can prove this. But perhaps SOS d=O(1) cannot handle

42
A simpler open problem Theorem: (a corollary of [Harper’66]’s Vertex-Isoperimetric Inequality) Let A, B ⊆ {0,1} m satisfy dist(A,B) ≥ Then |A|, |B| aren’t both large: Conjecture: SOS d=4 cannot prove this.

43
Dynamic SOS Lines of the proof are of form P(X 1, …, X n ) ≥ 0. From P ≥ 0 and Q ≥ 0 can derive P + Q ≥ 0 and P Q ≥ 0. Can always derive R 2 ≥ 0. To refute a system {P 1 ≥ 0, …, P m ≥ 0}, derive −1 ≥ 0. Complexity: max degree of any line

44
Dynamic SOS Facts [Grigoriev-Hirsch-Pasechnik’01]: Dynamic SOS d=3 refutes Knapsack Dynamic SOS d=5 refutes any unsatisfiable 3XOR instance Open problem 1 [GHP’01] : Suggest an explicit unsatisfiable boolean formula which SOS d=O(1) might not refute. Open problem 2: Give negative evidence re automatizability.

45
Unique-Games [Khot’02] conjectured that for the “UG” CSP, it’s NP-hard to distinguish -satisfiable instances from (1−)-satisfiable instances. [BBHKSZ’12]: Perhaps SOS d=4 can actually do it. ⇒ UGC is false (assuming NP ≠ P) Perhaps SOS d=log(n) can do it. ⇒ UGC is false (assuming NP ⊈ TIME[n log n ]) Why be so concerned about automatizability?

46
Unique-Games My favorite UG algorithm: Given an -satisfiable instance, nondeterministically guess a poly-length ZFC proof that instance is ≤ (1−)-satisfiable. [Khot’02] conjectured that for the “UG” CSP, it’s NP-hard to distinguish -satisfiable instances from (1−)-satisfiable instances. If this algorithm works, UGC is false. (assuming NP ≠ coNP)

47
ありがとうございました！ Thank you!

Similar presentations

© 2017 SlidePlayer.com Inc.

All rights reserved.

Ads by Google

Ppt on management of natural resources Ppt on sachin tendulkar life Ppt on search engine working Ppt on global warming and acid rain Ppt on weapons of mass destruction bush Ppt on changing face of london Download ppt on civil disobedience movement in united Make a ppt on natural disasters Ppt on preservation of public property auctions Ppt on differential used in automobiles