Factorization Norms and Hereditary Discrepancy

The $\gamma_2$ norm of a real $m\times n$ matrix $A$ is the minimum number $t$ such that the column vectors of $A$ are contained in a $0$-centered ellipsoid $E\subseteq\mathbb{R}^m$ which in turn is contained in the hypercube $[-t, t]^m$. We prove that this classical quantity approximates the \emph{hereditary discrepancy} $\mathrm{herdisc}\ A$ as follows: $\gamma_2(A) = {O(\log m)}\cdot \mathrm{herdisc}\ A$ and $\mathrm{herdisc}\ A = O(\sqrt{\log m}\,)\cdot\gamma_2(A) $. Since $\gamma_2$ is polynomial-time computable, this gives a polynomial-time approximation algorithm for hereditary discrepancy. Both inequalities are shown to be asymptotically tight. We then demonstrate on several examples the power of the $\gamma_2$ norm as a tool for proving lower and upper bounds in discrepancy theory. Most notably, we prove a new lower bound of $\Omega(\log^{d-1} n)$ for the \emph{$d$-dimensional Tusn\'ady problem}, asking for the combinatorial discrepancy of an $n$-point set in $\mathbb{R}^d$ with respect to axis-parallel boxes. For $d>2$, this improves the previous best lower bound, which was of order approximately $\log^{(d-1)/2}n$, and it comes close to the best known upper bound of $O(\log^{d+1/2}n)$, for which we also obtain a new, very simple proof.

[1]  József Beck,et al.  Irregularities of distribution: Index of theorems and corollaries , 1987 .

[2]  József Beck,et al.  Balanced two-colorings of finite sets in the square I , 1981, Comb..

[3]  M. Lacey,et al.  On the small ball inequality in three dimensions , 2006, math/0609815.

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

[5]  R. L. Graham A Note on Irregularities of Distribution , 2013, Integers.

[6]  Wojciech Banaszczyk,et al.  On series of signed vectors and their rearrangements , 2012, Random Struct. Algorithms.

[7]  E. Haacke Sequences , 2005 .

[8]  Aleksandar Nikolov,et al.  Tight hardness results for minimizing discrepancy , 2011, SODA '11.

[9]  Robert F. Tichy,et al.  Sequences, Discrepancies and Applications , 1997 .

[10]  Roy Mathias,et al.  The Hadamard Operator Norm of a Circulant and Applications , 1997 .

[11]  Aleksandar Nikolov,et al.  The geometry of differential privacy: the sparse and approximate cases , 2012, STOC '13.

[12]  Nathan Linial,et al.  Lower bounds in communication complexity based on factorization norms , 2007, STOC '07.

[13]  J. Matousek,et al.  Combinatorial Discrepancy for Boxes via the Ellipsoid-Infinity Norm , 2014 .

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

[15]  N. Biggs GEOMETRIC ALGORITHMS AND COMBINATORIAL OPTIMIZATION: (Algorithms and Combinatorics 2) , 1990 .

[16]  Wojciech Banaszczyk,et al.  Balancing vectors and Gaussian measures of n-dimensional convex bodies , 1998, Random Struct. Algorithms.

[17]  J. Beck A two-dimensional van Aardenne-Ehrenfest theorem in irregularities of distribution , 1989 .

[18]  J. Gawroński Amsterdam , 2008, Water in Times of Climate Change.

[19]  Bernd Grtner,et al.  Approximation Algorithms and Semidefinite Programming , 2012 .

[20]  L. Lovász,et al.  Geometric Algorithms and Combinatorial Optimization , 1981 .

[21]  Kunal Talwar,et al.  On The Hereditary Discrepancy of Homogeneous Arithmetic Progressions , 2013 .

[22]  Michael L. Fredman,et al.  The Complexity of Maintaining an Array and Computing Its Partial Sums , 1982, JACM.

[23]  Aravind Srinivasan,et al.  Improving the discrepancy bound for sparse matrices: better approximations for sparse lattice approximation problems , 1997, SODA '97.

[24]  Hans Ulrich Simon,et al.  Estimating the Optimal Margins of Embeddings in Euclidean Half Spaces , 2004, Machine Learning.

[25]  K. Ball An Elementary Introduction to Modern Convex Geometry , 1997 .

[26]  Bernard Chazelle,et al.  The Discrepancy Method , 1998, ISAAC.

[27]  Hans Ulrich Simon,et al.  Estimating the Optimal Margins of Embeddings in Euclidean Half Spaces , 2001, COLT/EuroCOLT.

[28]  K. F. Roth Remark concerning integer sequences , 1964 .

[29]  József Beck,et al.  Geometric Discrepancy Theory Anduniform Distribution , 2004, Handbook of Discrete and Computational Geometry, 2nd Ed..

[30]  Ji R Matou An L P Version of the Beck-fiala Conjecture , .

[31]  J. Matousek,et al.  Geometric Discrepancy: An Illustrated Guide , 2009 .

[32]  Géza Bohus,et al.  On the Discrepancy of 3 Permutations , 1990, Random Struct. Algorithms.

[33]  Alberto Seeger Calculus rules for combinations of ellipsoids and applications , 1993, Bulletin of the Australian Mathematical Society.

[34]  Aleksandar Nikolov,et al.  Optimal private halfspace counting via discrepancy , 2012, STOC '12.

[35]  Dömötör Pálvölgyi,et al.  Indecomposable Coverings with Concave Polygons , 2010, Discret. Comput. Geom..

[36]  Aleksandar Nikolov,et al.  Beck's Three Permutations Conjecture: A Counterexample and Some Consequences , 2012, 2012 IEEE 53rd Annual Symposium on Foundations of Computer Science.

[37]  J. Bourgain,et al.  Invertibility of ‘large’ submatrices with applications to the geometry of Banach spaces and harmonic analysis , 1987 .

[38]  J. Matousek,et al.  The determinant bound for discrepancy is almost tight , 2011, 1101.0767.

[39]  Jirí Matousek,et al.  Combinatorial Discrepancy for Boxes via the gamma_2 Norm , 2015, Symposium on Computational Geometry.

[40]  Kasper Green Larsen On Range Searching in the Group Model and Combinatorial Discrepancy , 2011, 2011 IEEE 52nd Annual Symposium on Foundations of Computer Science.

[41]  R. Vershynin John's decompositions: Selecting a large part , 1999, math/9909110.

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

[43]  W. Schmidt On irregularities of distribution vii , 1972 .

[44]  M. Lacey,et al.  On the Small Ball Inequality in All Dimensions , 2007, 0705.4619.

[45]  Troy Lee,et al.  A Direct Product Theorem for Discrepancy , 2008, 2008 23rd Annual IEEE Conference on Computational Complexity.

[46]  Jirí Matousek An Lp Version of the Beck-Fiala Conjecture , 1998, Eur. J. Comb..

[47]  K. Ball An elementary introduction to modern convex geometry, in flavors of geometry , 1997 .

[48]  László Lovász,et al.  Discrepancy of Set-systems and Matrices , 1986, Eur. J. Comb..

[49]  Bernard Chazelle,et al.  A trace bound for the hereditary discrepancy , 2000, SCG '00.

[50]  Jiří Matoušek,et al.  Discrepancy in arithmetic progressions , 1996 .

[51]  József Beck Balanced two-colorings of finite sets in the cube , 1989, Discret. Math..

[52]  Anand Srivastav,et al.  Discrepancy of Cartesian Products of Arithmetic Progressions , 2004, Electron. J. Comb..

[53]  J. Halton On the efficiency of certain quasi-random sequences of points in evaluating multi-dimensional integrals , 1960 .

[54]  Adi Shraibman,et al.  Lower bounds in communication complexity based on factorization norms , 2009 .

[55]  Aleksandar Nikolov,et al.  The Geometry of Differential Privacy: The Small Database and Approximate Cases , 2016, SIAM J. Comput..

[56]  Bernard Chazelle,et al.  The Discrepancy of Boxes in Higher Dimension , 2001, Discret. Comput. Geom..

[57]  M. Ziegler Volume 152 of Graduate Texts in Mathematics , 1995 .

[58]  J. Beck,et al.  Irregularities of distribution , 1987 .

[59]  K. Ball Chapter 4 – Convex Geometry and Functional Analysis , 2001 .

[60]  J. Beck,et al.  Discrepancy Theory , 1996 .

[61]  J. Matoussek On the Discrepancy for Boxes and Polytopes , 1999 .

[62]  N. Tomczak-Jaegermann Banach-Mazur distances and finite-dimensional operator ideals , 1989 .

[63]  Gilbert Strang,et al.  Functions of Difference Matrices Are Toeplitz Plus Hankel , 2014, SIAM Rev..

[64]  Andrew McGregor,et al.  Optimizing linear counting queries under differential privacy , 2009, PODS.

[65]  J. Hammersley MONTE CARLO METHODS FOR SOLVING MULTIVARIABLE PROBLEMS , 1960 .

[66]  B. M. Fulk MATH , 1992 .

[67]  Nathan Linial,et al.  Complexity measures of sign matrices , 2007, Comb..

[68]  Nathan Linial,et al.  Learning Complexity vs. Communication Complexity , 2008, 2008 23rd Annual IEEE Conference on Computational Complexity.

[69]  W. Banaszczyk Balancing vectors and Gaussian measures of n -dimensional convex bodies , 1998 .

[70]  K. F. Roth,et al.  On irregularities of distribution IV , 1979 .

[71]  J. Spencer Ten lectures on the probabilistic method , 1987 .

[72]  William W. L. Chen On irregularities of distribution. , 1980 .

[73]  Aleksandar Nikolov,et al.  Approximating Hereditary Discrepancy via Small Width Ellipsoids , 2013, SODA.