Showing papers by "Oded Regev published in 2016"

The restricted isometry property of subsampled fourier matrices

Abstract: A matrix A ∈ Cq×N satisfies the restricted isometry property of order k with constant e if it preserves the e2 norm of all k-sparse vectors up to a factor of 1 ± e. We prove that a matrix A obtained by randomly sampling q = O(k · log2k · log N) rows from an N ×N Fourier matrix satisfies the restricted isometry property of order k with a fixed e with high probability. This improves on Rudelson and Vershynin (Comm. Pure Appl. Math., 2008), its subsequent improvements, and Bourgain (GAFA Seminar Notes, 2014).

Recovering Short Generators of Principal Ideals in Cyclotomic Rings

Abstract: A handful of recent cryptographic proposals rely on the conjectured hardness of the following problem in the ring of integers of a cyclotomic number field: given a basis of a principal ideal that is guaranteed to have a "rather short" generator, find such a generator. Recently, Bernstein and Campbell-Groves-Shepherd sketched potential attacks against this problem; most notably, the latter authors claimed a polynomial-time quantum algorithm. Alternatively, replacing the quantum component with an algorithm of Biasse and Fieker would yield a classical subexponential-time algorithm. A key claim of Campbell et al. is that one step of their algorithm--namely, decoding the log-unit lattice of the ring to recover a short generator from an arbitrary one--is classically efficient whereas the standard approach on general lattices takes exponential time. However, very few convincing details were provided to substantiate this claim. In this work, we clarify the situation by giving a rigorous proof that the log-unit lattice is indeed efficiently decodable, for any cyclotomic of prime-power index. Combining this with the quantum algorithm from a recent work of Biasse and Song confirms the main claim of Campbell et al. Our proof consists of two main technical contributions: the first is a geometrical analysis, using tools from analytic number theory, of the standard generators of the group of cyclotomic units. The second showsthat for a wide class of typical distributions of the short generator, a standard lattice-decoding algorithm can recover it, given any generator. By extending our geometrical analysis, as a second main contribution we obtain an efficient algorithm that, given any generator of a principal ideal in a prime-power cyclotomic, finds a $$2^{\tilde{O}\sqrt{n}}$$ -approximate shortest vector in the ideal. Combining this with the result of Biasse and Song yields a quantum polynomial-time algorithm for the $$2^{\tilde{O}\sqrt{n}}$$ -approximate Shortest Vector Problem on principal ideal lattices.

Efficient quantum algorithms for (gapped) group testing and junta testing

Abstract: In the k-junta testing problem, a tester has to efficiently decide whether a given function f: {0, 1}n → {0, 1} is a k-junta (i.e., depends on at most fc of its input bits) or is e-far from any k-junta. Our main result is a quantum algorithm for this problem with query complexity O([EQUATION]) and time complexity O(n[EQUATION]). This quadratically improves over the query complexity of the previous best quantum junta tester, due to Atici and Servedio. Our tester is based on a new quantum algorithm for a gapped version of the combinatorial group testing problem, with an up to quartic improvement over the query complexity of the best classical algorithm. For our upper bound on the time complexity we give a near-linear time implementation of a shallow variant of the quantum Fourier transform over the symmetric group, similar to the Schur-Weyl transform. We also prove a lower bound of Ω(k1/3) queries for junta-testing (for constant e).

A Reverse Minkowski Theorem

Abstract: $ ewcommand{\R}{\mathbb{R}} ewcommand{\lat}{\mathcal{L}} $We prove a conjecture due to Dadush, showing that if $\lat \subset \R^n$ is a lattice such that $\det(\lat') \ge 1$ for all sublattices $\lat' \subseteq \lat$, then \[ \sum_{\vec y \in \lat} e^{-\pi t^2 \|\vec y\|^2} \le 3/2 \; , \] where $t := 10(\log n + 2)$. From this we derive bounds on the number of short lattice vectors, which can be viewed as a partial converse to Minkowski's celebrated first theorem. We also derive a bound on the covering radius.

The List-Decoding Size of Fourier-Sparse Boolean Functions

Abstract: A function defined on the Boolean hypercube is k-Fourier-sparse if it has at most k nonzero Fourier coefficients. For a function f: F2n r R and parameters k and d, we prove a strong upper bound on the number of k-Fourier-sparse Boolean functions that disagree with f on at most d inputs. Our bound implies that the number of uniform and independent random samples needed for learning the class of k-Fourier-sparse Boolean functions on n variables exactly is at most O(n · klog k).As an application, we prove an upper bound on the query complexity of testing Booleanity of Fourier-sparse functions. Our bound is tight up to a logarithmic factor and quadratically improves on a result due to Gur and Tamuz [2013].

Towards Strong Reverse Minkowski-Type Inequalities for Lattices

Abstract: We present a natural reverse Minkowski-type inequality for lattices, which gives upper bounds on the number of lattice points in a Euclidean ball in terms of sublattice determinants, and conjecture its optimal form. The conjecture exhibits a surprising wealth of connections to various areas in mathematics and computer science, including a conjecture motivated by integer programming by Kannan and Lovasz (Annals of Math. 1988), a question from additive combinatorics asked by Green, a question on Brownian motions asked by Saloff-Coste (Colloq. Math. 2010), a theorem by Milman and Pisier from convex geometry (Ann. Probab. 1987), worst-case to average-case reductions in lattice-based cryptography, and more. We present these connections, provide evidence for the conjecture, and discuss possible approaches towards a proof. Our main technical contribution is in proving that our conjecture implies the l2 case of the Kannan and Lovasz conjecture. The proof relies on a novel convex relaxation for the covering radius, and a rounding procedure based on "uncrossing" lattice subspaces.

A Counterexample to Monotonicity of Relative Mass in Random Walks

Abstract: For a finite undirected graph $G = (V,E)$, let $p_{u,v}(t)$ denote the probability that a continuous-time random walk starting at vertex $u$ is in $v$ at time $t$. In this note we give an example of a Cayley graph $G$ and two vertices $u,v \in G$ for which the function \[ r_{u,v}(t) = \frac{p_{u,v}(t)} {p_{u,u}(t)} \qquad t \geq 0 \] is not monotonically non-decreasing. This answers a question asked by Peres in 2013.

On the Space Complexity of Linear Programming with Preprocessing

Abstract: It is well known that Linear Programming is P-complete, with a logspace reduction. In this work we ask whether Linear Programming remains P-complete, even if the polyhedron (i.e., the set of linear inequality constraints) is a fixed polyhedron, for each input size, and only the objective function is given as input. More formally, we consider the following problem: maximize c⋅x, subject to Ax ≤ b; x ∈ Rd, where A,b are fixed in advance and only c is given as an input.We start by showing that the problem remains P-complete with a logspace reduction, thus showing that n{o(1)-space algorithms are unlikely. This result is proved by a direct classical reduction.We then turn to study approximation algorithms and ask what is the best approximation factor that could be obtained by a small space algorithm. Since approximation factors are mostly meaningful when the objective function is non-negative, we restrict ourselves to the case where x > 0 and c > 0. We show that (even in this possibly easier case) approximating the value of c⋅x (within any polynomial factor) is P-complete with a polylog space reduction, thus showing nthat 2(log n)o(1)-space approximation algorithms are unlikely.The last result is proved using a recent work of Kalai, Raz, and Rothblum, showing that every language in P has a no-signaling multi-prover interactive proof with poly-logarithmic communication complexity. To the best of our knowledge, our result gives the first space hardness of approximation result proved by a PCP-based argument.

A Note on Koldobsky's Lattice Slicing Inequality

Abstract: $ ewcommand{\R}{{\mathbb{R}}} ewcommand{\Z}{{\mathbb{Z}}} \renewcommand{\vec}[1]{{\mathbf{#1}}} $We show that if $K \subset \R^d$ is an origin-symmetric convex body, then there exists a vector $\vec{y} \in \Z^d$ such that \begin{align*} |K \cap \Z^d \cap \vec{y}^\perp| / |K \cap \Z^d| \ge \min(1,c \cdot d^{-1} \cdot \mathrm{vol}(K)^{-1/(d-1)}) \; , \end{align*} for some absolute constant $c> 0$, where $\vec{y}^\perp$ denotes the subspace orthogonal to $\vec{y}$. This gives a partial answer to a question by Koldobsky.

Counterexamples to a conjecture of Woods

Abstract: A conjecture of Woods from 1972 is disproved: for d≥30, there are well-rounded unimodular lattices in Rd with covering radius greater than that of Zd.

A counterexample to a strong variant of the Polynomial Freiman-Ruzsa conjecture in Euclidean space

Abstract: The Polynomial Freiman-Ruzsa conjecture is one of the central open problems in additive combinatorics. If true, it would give tight quantitative bounds relating combinatorial and algebraic notions of approximate subgroups. In this note, we restrict our attention to subsets of Euclidean space. In this regime, the original conjecture considers approximate algebraic subgroups as the set of lattice points in a convex body. Green asked in 2007 whether this can be simplified to a generalized arithmetic progression, while not losing more than a polynomial factor in the underlying parameters. We give a negative answer to this question, based on a recent reverse Minkowski theorem combined with estimates for random lattices.

The Minrank of Random Graphs.

Abstract: The minrank of a directed graph $G$ is the minimum rank of a matrix $M$ that can be obtained from the adjacency matrix of $G$ by switching some ones to zeros (i.e., deleting edges) and then setting all diagonal entries to one. This quantity is closely related to the fundamental information-theoretic problems of (linear) index coding (Bar-Yossef et al. ), network coding (Effros et al. ), and distributed storage (Mazumdar, ISIT, 2014). We prove tight bounds on the minrank of directed Erdős–Renyi random graphs $G(n,p)$ for all regimes of $p\in [{0,1}]$ . In particular, for any constant $p$ , we show that $\mathsf {minrk}(G) = \Theta (n/\log n)$ with high probability, where $G$ is chosen from $G(n,p)$ . This bound gives a near quadratic improvement over the previous best lower bound of $\Omega (\sqrt {n})$ (Haviv and Langberg), and partially settles an open problem raised by Lubetzky and Stav. Our lower bound matches the well-known upper bound obtained by the “clique covering” solution and settles the linear index coding problem for random knowledge graphs.

Towards Strong Reverse Minkowski-type Inequalities for Lattices

Abstract: We present a natural reverse Minkowski-type inequality for lattices, which gives upper bounds on the number of lattice points in a Euclidean ball in terms of sublattice determinants, and conjecture its optimal form. The conjecture exhibits a surprising wealth of connections to various areas in mathematics and computer science, including a conjecture motivated by integer programming by Kannan and Lovasz (Annals of Math. 1988), a question from additive combinatorics asked by Green, a question on Brownian motions asked by Saloff-Coste (Colloq. Math. 2010), a theorem by Milman and Pisier from convex geometry (Ann. Probab. 1987), worst-case to average-case reductions in lattice-based cryptography, and more. We present these connections, provide evidence for the conjecture, and discuss possible approaches towards a proof. Our main technical contribution is in proving that our conjecture implies the $\ell_2$ case of the Kannan and Lovasz conjecture. The proof relies on a novel convex relaxation for the covering radius, and a rounding procedure for based on "uncrossing" lattice subspaces.

The Minrank of Random Graphs

Abstract: The minrank of a graph $G$ is the minimum rank of a matrix $M$ that can be obtained from the adjacency matrix of $G$ by switching some ones to zeros (i.e., deleting edges) and then setting all diagonal entries to one. This quantity is closely related to the fundamental information-theoretic problems of (linear) index coding (Bar-Yossef et al., FOCS'06), network coding and distributed storage, and to Valiant's approach for proving superlinear circuit lower bounds (Valiant, Boolean Function Complexity '92). We prove tight bounds on the minrank of random Erdős-Renyi graphs $G(n,p)$ for all regimes of $p\in[0,1]$. In particular, for any constant $p$, we show that $\mathsf{minrk}(G) = \Theta(n/\log n)$ with high probability, where $G$ is chosen from $G(n,p)$. This bound gives a near quadratic improvement over the previous best lower bound of $\Omega(\sqrt{n})$ (Haviv and Langberg, ISIT'12), and partially settles an open problem raised by Lubetzky and Stav (FOCS '07). Our lower bound matches the well-known upper bound obtained by the "clique covering" solution, and settles the linear index coding problem for random graphs. Finally, our result suggests a new avenue of attack, via derandomization, on Valiant's approach for proving superlinear lower bounds for logarithmic-depth semilinear circuits.