Algorithmic Obstructions in the Random Number Partitioning Problem

We consider the algorithmic problem of finding a near-optimal solution for the number partitioning problem (NPP). The NPP appears in many applications, including the design of randomized controlled trials, multiprocessor scheduling, and cryptography; and is also of theoretical significance. It possesses a so-called statistical-to-computational gap: when its input $X$ has distribution $\mathcal{N}(0,I_n)$, its optimal value is $\Theta(\sqrt{n}2^{-n})$ w.h.p.; whereas the best polynomial-time algorithm achieves an objective value of only $2^{-\Theta(\log^2 n)}$, w.h.p. In this paper, we initiate the study of the nature of this gap. Inspired by insights from statistical physics, we study the landscape of NPP and establish the presence of the Overlap Gap Property (OGP), an intricate geometric property which is known to be a rigorous evidence of an algorithmic hardness for large classes of algorithms. By leveraging the OGP, we establish that (a) any sufficiently stable algorithm, appropriately defined, fails to find a near-optimal solution with energy below $2^{-\omega(n \log^{-1/5} n)}$; and (b) a very natural MCMC dynamics fails to find near-optimal solutions. Our simulations suggest that the state of the art algorithm achieving $2^{-\Theta(\log^2 n)}$ is indeed stable, but formally verifying this is left as an open problem. OGP regards the overlap structure of $m-$tuples of solutions achieving a certain objective value. When $m$ is constant we prove the presence of OGP in the regime $2^{-\Theta(n)}$, and the absence of it in the regime $2^{-o(n)}$. Interestingly, though, by considering overlaps with growing values of $m$ we prove the presence of the OGP up to the level $2^{-\omega(\sqrt{n\log n})}$. Our proof of the failure of stable algorithms at values $2^{-\omega(n \log^{-1/5} n)}$ employs methods from Ramsey Theory from the extremal combinatorics, and is of independent interest.

[1]  David S. Johnson,et al.  Computers and Intractability: A Guide to the Theory of NP-Completeness , 1978 .

[2]  G. S. Lueker Exponentially small bounds on the expected optimum of the partition and subset sum problems , 1998, Random Struct. Algorithms.

[3]  Alexander S. Wein Optimal Low-Degree Hardness of Maximum Independent Set , 2020, Mathematical Statistics and Learning.

[4]  David Gamarnik,et al.  Sparse High-Dimensional Linear Regression. Algorithmic Barriers and a Local Search Algorithm , 2017, 1711.04952.

[5]  Mark Jerrum,et al.  Large Cliques Elude the Metropolis Process , 1992, Random Struct. Algorithms.

[6]  G. Szekeres,et al.  A combinatorial problem in geometry , 2009 .

[7]  David Gamarnik,et al.  The overlap gap property in principal submatrix recovery , 2019, Probability Theory and Related Fields.

[8]  Stephan Mertens,et al.  Universality in the level statistics of disordered systems. , 2004, Physical review. E, Statistical, nonlinear, and soft matter physics.

[9]  Santosh S. Vempala,et al.  Statistical Algorithms and a Lower Bound for Detecting Planted Cliques , 2012, J. ACM.

[10]  Xin Yang,et al.  Number Balancing is as Hard as Minkowski's Theorem and Shortest Vector , 2016, IPCO.

[11]  Nicholas C. Wormald,et al.  Large independent sets in regular graphs of large girth , 2007, J. Comb. Theory, Ser. B.

[12]  Edward G. Coffman,et al.  Probabilistic analysis of packing and partitioning algorithms , 1991, Wiley-Interscience series in discrete mathematics and optimization.

[13]  Peng Zhang,et al.  Balancing covariates in randomized experiments using the Gram-Schmidt walk , 2019, ArXiv.

[14]  M. Talagrand Mean Field Models for Spin Glasses , 2011 .

[15]  Benjamin Yakir,et al.  The Differencing Algorithm LDM for Partitioning: A Proof of a Conjecture of Karmarkar and Karp , 1996, Math. Oper. Res..

[16]  Christian Borgs,et al.  Proof of the local REM conjecture for number partitioning. I: Constant energy scales , 2009 .

[17]  Andrea Montanari,et al.  Optimization of the Sherrington-Kirkpatrick Hamiltonian , 2018, 2019 IEEE 60th Annual Symposium on Foundations of Computer Science (FOCS).

[18]  David Gamarnik,et al.  Computing the Partition Function of the Sherrington-Kirkpatrick Model is Hard on Average , 2018, 2020 IEEE International Symposium on Information Theory (ISIT).

[19]  Christian Borgs,et al.  Phase transition and finite‐size scaling for the integer partitioning problem , 2001, Random Struct. Algorithms.

[20]  J. Spencer Six standard deviations suffice , 1985 .

[21]  David Gamarnik,et al.  The Landscape of the Planted Clique Problem: Dense subgraphs and the Overlap Gap Property , 2019, ArXiv.

[22]  B. Szegedy,et al.  Limits of locally–globally convergent graph sequences , 2014 .

[23]  Richard M. Karp,et al.  The Differencing Method of Set Partitioning , 1983 .

[24]  Florent Krzakala,et al.  Phase transitions in sparse PCA , 2015, 2015 IEEE International Symposium on Information Theory (ISIT).

[25]  George S. Lueker A note on the average-case behavior of a simple differencing method for partitioning , 1987 .

[26]  B. Derrida Random-Energy Model: Limit of a Family of Disordered Models , 1980 .

[27]  Jonathan Shi,et al.  Tensor principal component analysis via sum-of-square proofs , 2015, COLT.

[28]  Guy Bresler,et al.  Optimal Average-Case Reductions to Sparse PCA: From Weak Assumptions to Strong Hardness , 2019, COLT.

[29]  Gábor Lugosi,et al.  Concentration Inequalities - A Nonasymptotic Theory of Independence , 2013, Concentration Inequalities.

[30]  Florent Krzakala,et al.  MMSE of probabilistic low-rank matrix estimation: Universality with respect to the output channel , 2015, 2015 53rd Annual Allerton Conference on Communication, Control, and Computing (Allerton).

[31]  Avi Levy,et al.  Deterministic Discrepancy Minimization via the Multiplicative Weight Update Method , 2016, IPCO.

[32]  G. S. Lueker,et al.  Probabilistic analysis of optimum partitioning , 1986, Journal of Applied Probability.

[33]  David Conlon,et al.  Recent developments in graph Ramsey theory , 2015, Surveys in Combinatorics.

[34]  A. COJA-OGHLAN,et al.  Walksat Stalls Well Below Satisfiability , 2016, SIAM J. Discret. Math..

[35]  Kevin P. Costello Balancing Gaussian vectors , 2009 .

[36]  Alan M. Frieze,et al.  On the independence number of random graphs , 1990, Discret. Math..

[37]  Santosh S. Vempala,et al.  Integer feasibility of random polytopes: random integer programs , 2011, ITCS.

[38]  Noga Alon,et al.  The Probabilistic Method , 2015, Fundamentals of Ramsey Theory.

[39]  Afonso S. Bandeira,et al.  Notes on computational-to-statistical gaps: predictions using statistical physics , 2018, Portugaliae Mathematica.

[40]  Li-Hui Tsai,et al.  Asymptotic Analysis of an Algorithm for Balanced Parallel Processor Scheduling , 1992, SIAM J. Comput..

[41]  Charles R. Johnson,et al.  Matrix analysis , 1985, Statistical Inference for Engineers and Data Scientists.

[42]  David Gamarnik,et al.  Low-Degree Hardness of Random Optimization Problems , 2020, 2020 IEEE 61st Annual Symposium on Foundations of Computer Science (FOCS).

[43]  S. Mertens Phase Transition in the Number Partitioning Problem , 1998, cond-mat/9807077.

[44]  A. Hoffman,et al.  The variation of the spectrum of a normal matrix , 1953 .

[45]  Pravesh Kothari,et al.  A Nearly Tight Sum-of-Squares Lower Bound for the Planted Clique Problem , 2016, 2016 IEEE 57th Annual Symposium on Foundations of Computer Science (FOCS).

[46]  Daniel M. Kane,et al.  Statistical Query Lower Bounds for Robust Estimation of High-Dimensional Gaussians and Gaussian Mixtures , 2016, 2017 IEEE 58th Annual Symposium on Foundations of Computer Science (FOCS).

[47]  David Gamarnik,et al.  Combinatorial approach to the interpolation method and scaling limits in sparse random graphs , 2010, STOC '10.

[48]  Madhu Sudan,et al.  Limits of local algorithms over sparse random graphs , 2013, ITCS.

[50]  Stefan Boettcher,et al.  Analysis of the Karmarkar-Karp differencing algorithm , 2008, ArXiv.

[51]  Florent Krzakala,et al.  Statistical physics of inference: thresholds and algorithms , 2015, ArXiv.

[52]  Wasim Huleihel,et al.  Reducibility and Computational Lower Bounds for Problems with Planted Sparse Structure , 2018, COLT.

[53]  Dmitry Panchenko,et al.  Suboptimality of local algorithms for a class of max-cut problems , 2017, The Annals of Probability.

[54]  Amin Coja-Oghlan,et al.  Algorithmic Barriers from Phase Transitions , 2008, 2008 49th Annual IEEE Symposium on Foundations of Computer Science.

[55]  Luc Devroye,et al.  Local optima of the Sherrington-Kirkpatrick Hamiltonian , 2017, ArXiv.

[56]  Afonso S. Bandeira,et al.  Notes on Computational Hardness of Hypothesis Testing: Predictions using the Low-Degree Likelihood Ratio , 2019, ArXiv.

[57]  Cristopher Moore,et al.  Random k-SAT: Two Moments Suffice to Cross a Sharp Threshold , 2003, SIAM J. Comput..

[58]  B. Derrida Random-energy model: An exactly solvable model of disordered systems , 1981 .

[59]  Will Perkins,et al.  Storage capacity in symmetric binary perceptrons , 2019, Journal of Physics A: Mathematical and Theoretical.

[60]  Madhu Sudan,et al.  Performance of Sequential Local Algorithms for the Random NAE-K-SAT Problem , 2017, SIAM J. Comput..

[61]  Thierry Mora,et al.  Clustering of solutions in the random satisfiability problem , 2005, Physical review letters.

[62]  Nikhil Bansal,et al.  Constructive Algorithms for Discrepancy Minimization , 2010, 2010 IEEE 51st Annual Symposium on Foundations of Computer Science.

[63]  Thomas Rothvoß,et al.  Constructive Discrepancy Minimization for Convex Sets , 2014, 2014 IEEE 55th Annual Symposium on Foundations of Computer Science.

[64]  Ryan O'Donnell,et al.  Sum of squares lower bounds for refuting any CSP , 2017, STOC.

[65]  Alan M. Frieze,et al.  On the independence and chromatic numbers of random regular graphs , 1992, J. Comb. Theory, Ser. B.

[66]  Michael Kearns,et al.  Efficient noise-tolerant learning from statistical queries , 1993, STOC.

[67]  Prasad Raghavendra,et al.  High-dimensional estimation via sum-of-squares proofs , 2018, Proceedings of the International Congress of Mathematicians (ICM 2018).

[68]  Roman Vershynin,et al.  High-Dimensional Probability , 2018 .

[69]  Roman Vershynin,et al.  Introduction to the non-asymptotic analysis of random matrices , 2010, Compressed Sensing.

[70]  David Gamarnik,et al.  Finding a large submatrix of a Gaussian random matrix , 2016, The Annals of Statistics.

[71]  Amin Coja-Oghlan,et al.  On independent sets in random graphs , 2010, SODA '11.

[72]  Christian Borgs,et al.  Proof of the local REM conjecture for number partitioning. II. Growing energy scales , 2005, Random Struct. Algorithms.

[73]  Andrea Montanari,et al.  Improved Sum-of-Squares Lower Bounds for Hidden Clique and Hidden Submatrix Problems , 2015, COLT.

[74]  Martin E. Hellman,et al.  Hiding information and signatures in trapdoor knapsacks , 1978, IEEE Trans. Inf. Theory.

[75]  Bálint Virág,et al.  Local algorithms for independent sets are half-optimal , 2014, ArXiv.

[76]  Federico Ricci-Tersenghi,et al.  On the solution-space geometry of random constraint satisfaction problems , 2006, STOC '06.

[77]  A. Krieger,et al.  Nearly random designs with greatly improved balance , 2016, Biometrika.

[78]  Shachar Lovett,et al.  Constructive Discrepancy Minimization by Walking on the Edges , 2012, 2012 IEEE 53rd Annual Symposium on Foundations of Computer Science.

[79]  David Gamarnik,et al.  The overlap gap property and approximate message passing algorithms for $p$-spin models , 2019, The Annals of Probability.

[80]  G. B. Arous,et al.  Algorithmic thresholds for tensor PCA , 2018, The Annals of Probability.

[81]  Prasad Raghavendra,et al.  The Power of Sum-of-Squares for Detecting Hidden Structures , 2017, 2017 IEEE 58th Annual Symposium on Foundations of Computer Science (FOCS).

[82]  János Komlós,et al.  A Note on Ramsey Numbers , 1980, J. Comb. Theory, Ser. A.

[83]  C. Martin 2015 , 2015, Les 25 ans de l’OMC: Une rétrospective en photos.

[84]  Avi Wigderson,et al.  Sum-of-squares Lower Bounds for Planted Clique , 2015, STOC.

[85]  Ilias Zadik,et al.  Free Energy Wells and Overlap Gap Property in Sparse PCA , 2020, COLT.

[86]  Konstantinos Panagiotou,et al.  Catching the k-NAESAT threshold , 2011, STOC '12.