Lower Bounds on the Size of Semidefinite Programming Relaxations

被引:93
作者
Lee, James R. [1 ]
Raghavendra, Prasad [2 ]
Steurer, David [3 ]
机构
[1] Univ Washington, Seattle, WA 98195 USA
[2] Univ Calif Berkeley, Berkeley, CA USA
[3] Cornell, Ithaca, NY USA
来源
STOC'15: PROCEEDINGS OF THE 2015 ACM SYMPOSIUM ON THEORY OF COMPUTING | 2015年
关键词
semidefinite programming; sum-of-squares method; lower bounds on positive-semidefinite rank; approximation complexity; quantum learning; polynomial optimization; OPTIMIZATION; INAPPROXIMABILITY; ALGORITHMS; COMPLEXITY; PROOFS;
D O I
10.1145/2746539.2746599
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
We introduce a method for proving lower bounds on the efficacy of semidefinite programming (SDP) relaxations for combinatorial problems. In particular, we show that the cut, TSP, and stable set polytopes on n-vertex graphs are not the linear image of the feasible region of any SDP (i.e., any spectrahedron) of dimension less than 2(n delta), for some constant delta > 0. This result yields the first super-polynomial lower bounds on the semidefinite extension complexity of any explicit family of polytopes. Our results follow from a general technique for proving lower bounds on the positive semidefinite rank of a matrix. To this end, we establish a close connection between arbitrary SDPs and those arising from the sum-of-squares SDP hierarchy. For approximating maximum constraint satisfaction problems, we prove that SDPs of polynomial-size are equivalent in power to those arising from degree-O(1) sum of -squares relaxations. This result implies, for instance, that no family of polynomial-size SDP relaxations can achieve better than a 7/8-approximation for MAX 3-SAT.
引用
收藏
页码:567 / 576
页数:10
相关论文
共 44 条
[21]   Linear lower bound on degrees of Positivstellensatz calculus proofs for the parity [J].
Grigoriev, D .
THEORETICAL COMPUTER SCIENCE, 2001, 259 (1-2) :613-622
[22]  
Halperin E., 2003, STOC, P585, DOI [10.1145/780542.780628, DOI 10.1145/780542.780628]
[23]   Approximate graph coloring by semidefinite programming [J].
Karger, D ;
Motwani, R ;
Sudan, M .
JOURNAL OF THE ACM, 1998, 45 (02) :246-265
[24]   Optimal inapproximability results for MAX-CUT and other 2-variable CSPs? [J].
Khot, S ;
Kindler, G ;
Mossel, E ;
O'Donnell, R .
45TH ANNUAL IEEE SYMPOSIUM ON FOUNDATIONS OF COMPUTER SCIENCE, PROCEEDINGS, 2004, :146-154
[25]  
Khot Subhash, 2002, P 34 ANN ACM S THEOR, P767, DOI [DOI 10.1145/509907.510017, 10.1109/CCC.2002.1004334, DOI 10.1109/CCC.2002.1004334]
[26]   Global optimization with polynomials and the problem of moments [J].
Lasserre, JB .
SIAM JOURNAL ON OPTIMIZATION, 2001, 11 (03) :796-817
[27]  
Laurent M, 2009, IMA VOL MATH APPL, V149, P157
[28]   On the Power of Symmetric LP and SDP Relaxations [J].
Lee, James R. ;
Raghavendra, Prasad ;
Steurer, David ;
Tan, Ning .
2014 IEEE 29TH CONFERENCE ON COMPUTATIONAL COMPLEXITY (CCC), 2014, :13-21
[29]  
Lund C., 1993, Proceedings of the Twenty-Fifth Annual ACM Symposium on the Theory of Computing, P286, DOI 10.1145/167088.167172
[30]  
Nemirovsky Arkadii Semenovich, 1983, Problem complexity and method efficiency in optimization