Abstract
We introduce a method for proving Sum-of-Squares (SoS)/Lasserre hierarchy lower bounds when the initial problem formulation exhibits a high degree of symmetry. Our main technical theorem allows us to reduce the study of the positive semidefiniteness to the analysis of “well-behaved” univariate polynomial inequalities. We illustrate the technique on two problems, one unconstrained and the other with constraints. More precisely, we give a short elementary proof of Grigoriev/Laurent lower bound for finding the integer cut polytope of the complete graph. We also show that the SoS hierarchy requires a non-constant number of rounds to improve the initial integrality gap of 2 for the Min-Knapsack linear program strengthened with cover inequalities.
Similar content being viewed by others
Notes
More precisely, Grigoriev considers the positivstellensatz proof system, which is the dual of the SoS hierarchy considered in this paper. For brevity, we will use SoS hierarchy/proof system interchangeably as is customary in theoretical computer science literature. In optimization context the moment matrix formulation considered in this paper is usually called Lasserre hierarchy.
In [4], L(p) is written \(\tilde{ \mathbb {E}}[p]\) and called the “pseudo-expectation” of p.
We define the set-valued permutation by \(\pi (I) = \left\{ \pi (i)~|~i \in I \right\} \).
A quick calculation reveals that \((1-x)^k = \sum _{k=0}^n (-1)^k \left( {\begin{array}{c}n\\ k\end{array}}\right) x^k\). Taking the jth derivative with \(j<n\) on both sides, setting \(x=1\) and simplifying yields \(\sum _{k=0}^n (-1)^k \left( {\begin{array}{c}n\\ k\end{array}}\right) k(k-1)\cdots (k-j+1)=0\). Using this derivation one can show inductively that \(\sum _{k=0}^n (-1)^k\left( {\begin{array}{c}n\\ k\end{array}}\right) k^j = 0\) for every \(0 \le j<n\), and by taking linear combinations of such expressions one obtains that \(\sum _{k=0}^n (-1)\left( {\begin{array}{c}n\\ k\end{array}}\right) Q(k) = 0\) for any polynomial Q of degree at most \(n-1\).
We show that the roots \(r_i\) can be assumed to be real numbers.
Recall that at level n the integrality gap vanishes.
Recall that \(\left( {\begin{array}{c}n\\ -k\end{array}}\right) =\left( {\begin{array}{c}n\\ n+k\end{array}}\right) =0\) for any positive integer k.
References
Bansal, N., Buchbinder, N., Naor, J.: Randomized competitive algorithms for generalized caching. In: STOC, pp. 235–244 (2008)
Bansal, N., Gupta, A., Krishnaswamy, R.: A constant factor approximation algorithm for generalized min-sum set cover. In: SODA, pp. 1539–1545 (2010)
Bansal, N., Pruhs, K.: The geometry of scheduling. In: FOCS, pp. 407–414 (2010)
Barak, B., Brandão, F.G.S.L., Harrow, A.W., Kelner, J.A., Steurer, D., Zhou, Y.: Hypercontractivity, sum-of-squares proofs, and their applications. In: STOC, pp. 307–326 (2012)
Barak, B., Chan, S.O., Kothari, P.: Sum of squares lower bounds from pairwise independence. In: STOC (2015). arXiv:1501.00734
Barak, B., Steurer, D.: Sum-of-squares proofs and the quest toward optimal algorithms. Electron. Colloq. Comput. Complex. (ECCC) 21, 59 (2014)
Bhaskara, A., Charikar, M., Vijayaraghavan, A., Guruswami, V., Zhou, Y.: Polynomial integrality gaps for strong SDP relaxations of densest k-subgraph. In: SODA, pp. 388–405 (2012)
Bienstock, D., Zuckerberg, M.: Subset algebra lift operators for 0–1 integer programming. SIAM J. Optim. 15(1), 63–95 (2004)
Blekherman, G., ao Gouveia, J., Pfeiffer, J.: Sums of squares on the hypercube. CoRR arXiv:1402.4199 (2014)
Blekherman, G., Parrilo, P., Thomas, R.: Semidefinite optimization and convex algebraic geometry. Society for Industrial and Applied Mathematics, Philadelphia, PA (2012). https://doi.org/10.1137/1.9781611972290
Carnes, T., Shmoys, D.B.: Primal-dual schema for capacitated covering problems. In: IPCO, pp. 288–302 (2008)
Carr, R.D., Fleischer, L., Leung, V.J., Phillips, C.A.: Strengthening integrality gaps for capacitated network design and covering problems. In: SODA, pp. 106–115 (2000)
Chakrabarty, D., Grant, E., Könemann, J.: On column-restricted and priority covering integer programs. In: IPCO, pp. 355–368 (2010)
Cheung, K.K.H.: Computation of the Lasserre ranks of some polytopes. Math. Oper. Res. 32(1), 88–94 (2007)
Chlamtac, E., Tulsiani, M.: Convex relaxations and integrality gaps. In: Anjos, M., Lasserre, J. (eds.) Handbook on Semidefinite, Conic and Polynomial Optimization, vol. 166, pp. 139–169. Springer, Berlin (2011)
Cook, W., Dash, S.: On the matrix-cut rank of polyhedra. Math. Oper. Res. 26(1), 19–30 (2001)
Fawzi, H., Saunderson, J., Parrilo, P.: Sparse sum-of-squares certificates on finite Abelian groups. CoRR. arXiv:1503.01207 (2015)
Godsil, C.: Association schemes (2010). Lecture Notes Available at http://quoll.uwaterloo.ca/mine/Notes/assoc2.pdf
Goemans, M.X., Tunçel, L.: When does the positive semidefiniteness constraint help in lifting procedures? Math. Oper. Res. 26(4), 796–815 (2001). https://doi.org/10.1287/moor.26.4.796.10012
Grigoriev, D.: Complexity of positivstellensatz proofs for the knapsack. Comput. Complex. 10(2), 139–154 (2001)
Grigoriev, D.: Linear lower bound on degrees of positivstellensatz calculus proofs for the parity. Theor. Comput. Sci. 259(1–2), 613–622 (2001)
Grigoriev, D., Hirsch, E.A., Pasechnik, D.V.: Complexity of semi-algebraic proofs. In: STACS, pp. 419–430 (2002)
Hong, S., Tunçel, L.: Unification of lower-bound analyses of the lift-and-project rank of combinatorial optimization polyhedra. Discrete Appl. Math. 156(1), 25–41 (2008). https://doi.org/10.1016/j.dam.2007.07.021
Khot, S.: On the power of unique 2-prover 1-round games. In: STOC, pp. 767–775 (2002). https://doi.org/10.1145/509907.510017
Kurpisz, A., Leppänen, S., Mastrolilli, M.: A Lasserre lower bound for the min-sum single machine scheduling problem. In: ESA, pp. 853–864 (2015)
Kurpisz, A., Leppänen, S., Mastrolilli, M.: On the hardest problem formulations for the 0/1 lasserre hierarchy. In: ICALP, pp. 872–885 (2015)
Kurpisz, A., Leppänen, S., Mastrolilli, M.: Tight sum-of-squares lower bounds for binary polynomial optimization problems. pp. 78:1–78:14 (2016). https://doi.org/10.4230/LIPIcs.ICALP.2016.78
Lasserre, J.B.: Global optimization with polynomials and the problem of moments. SIAM J. Optim. 11(3), 796–817 (2001)
Laurent, M.: A comparison of the Sherali-Adams, Lovász-Schrijver, and Lasserre relaxations for 0–1 programming. Math. Oper. Res. 28(3), 470–496 (2003)
Laurent, M.: Lower bound for the number of iterations in semidefinite hierarchies for the cut polytope. Math. Oper. Res. 28(4), 871–883 (2003). https://doi.org/10.1287/moor.28.4.871.20508
Lee, J.R., Raghavendra, P., Steurer, D.: Lower bounds on the size of semidefinite programming relaxations. In: STOC, pp. 567–576 (2015). https://doi.org/10.1145/2746539.2746599
Lee, T., Prakash, A., de Wolf, R., Yuen, H.: On the sum-of-squares degree of symmetric quadratic functions. In: 31st Conference on Computational Complexity, CCC 2016, Tokyo, Japan, pp. 17:1–17:31 (2016). https://doi.org/10.4230/LIPIcs.CCC.2016.17
Lovász, L., Schrijver, A.: Cones of matrices and set-functions and 0–1 optimization. SIAM J. Optim. 1(2), 166–190 (1991). https://doi.org/10.1137/0801013
Meka, R., Potechin, A., Wigderson, A.: Sum-of-squares lower bounds for planted clique. In: STOC, pp. 87–96 (2015). https://doi.org/10.1145/2746539.2746600
O’Donnell, R.: Approximability and proof complexity (2013). Talk at ELC Tokyo. Slides Available at http://www.cs.cmu.edu/odonnell/slides/approx-proof-cxty.pps
Parrilo, P.: Structured semidefinite programs and semialgebraic geometry methods in robustness and optimization. Ph.D. thesis, California Institute of Technology, California (2000)
Raghavendra, P.: Optimal algorithms and inapproximability results for every CSP? In: STOC, pp. 245–254 (2008). https://doi.org/10.1145/1374376.1374414
Rothvoß, T.: The lasserre hierarchy in approximation algorithms (2013). In: Lecture Notes for the MAPSP 2013—Tutorial
Schoenebeck, G.: Linear level Lasserre lower bounds for certain k-csps. In: FOCS, pp. 593–602 (2008)
Stephen, T., Tunçel, L.: On a representation of the matching polytope via semidefinite liftings. Math. Oper. Res. 24(1), 1–7 (1999)
Tulsiani, M.: CSP gaps and reductions in the Lasserre hierarchy. In: STOC, pp. 303–312 (2009)
Wolsey, L.A.: Facets for a linear inequality in 0–1 variables. Math. Program. 8, 168–175 (1975)
Acknowledgements
The authors would like to express their gratitude to Ola Svensson for helpful discussions and ideas regarding this paper.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This is the full version of the paper that was presented at IPCO 2016. Supported by the Swiss National Science Foundation project 200020-169022 “Lift and Project Methods for Machine Scheduling Through Theory and Experiments” and by the project PZ00P2\(\_\)174117 “Theory and Applications of Linear and Semidefinite Relaxations for Combinatorial Optimization Problems”.
Appendices
Appendix
Omitted calculations for Max-Cut
Grigoriev [20] and Laurent [30] proved that the following solution is feasible for any \(\omega \le n/2\) up to round \(t\le \lfloor \omega \rfloor \) for the SoS hierarchy given in Definition 1:
For a graph \(G = (V, E)\), the objective function of the Max-Cut problem in the usual formulation for 0 / 1 variables is \(\sum _{(i,j) \in E} (x_i-x_j)^2\). The cut value of the SoS relaxation of Definition 1 is then \(\sum _{(i,j) \in E} y_{\left\{ i \right\} }+y_{\left\{ j \right\} }-2y_{\left\{ i,j \right\} }\). When G is the complete graph and the solution to the SoS relaxation is taken to be (36), the cut value of the SoS relaxation is
It is straightforward to check that for odd n and \(\omega = n/2\) this value is larger than the maximum cut in the complete graph of n vertices.
Change of basis. Using the change of basis of Lemma 1, solution \(\{y_I\}\) is equivalent to solution \(\{y^N_I\}\):
where we use the identity \(\sum _{\omega =0}^m (-1)^{\omega } \left( {\begin{array}{c}n\\ {\omega }\end{array}}\right) =(-1)^m\left( {\begin{array}{c}n-1\\ m\end{array}}\right) \).
Rights and permissions
About this article
Cite this article
Kurpisz, A., Leppänen, S. & Mastrolilli, M. Sum-of-squares hierarchy lower bounds for symmetric formulations. Math. Program. 182, 369–397 (2020). https://doi.org/10.1007/s10107-019-01398-9
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10107-019-01398-9