arXiv subjects

Ohad Klein

Publications and source records attributed to Ohad Klein.

12 recordsLinked to original sources

arXiv2024

Dimension-free estimates for low degree functions on the Hamming cube

The main result of this paper are dimension-free $L^p$ inequalities, $1 2,$ $\varepsilon>0,$ and $\theta=\theta(\varepsilon,p)\in (0,1)$ satisfying \[ \frac{1}{p}=\frac{\theta}{p+\varepsilon}+\frac{1-\theta}{2} \] we obtain, for any function $f:\{-1,1\}^n\to \mathbb{C}$ whose spectrum is bounded from above by $d,$ the Bernstein-Markov type inequalities \[\|\Delta^k f\|_{p} \le C(p,\varepsilon)^k \,d^k\, \|f\|_{2}^{1-\theta}\|f\|_{p+\varepsilon}^{\theta},\qquad k\in \mathbb{N}.\] Analogous inequalities are also proved for $p\in (1,2)$ with $p-\varepsilon$ replacing $p+\varepsilon.$ As a corollary, if $f$ is Boolean-valued or $f\colon \{-1,1\}^n\to \{-1,0,1\},$ we obtain the bounds \[\|\Delta^k f\|_{p} \le C(p)^k \,d^k\, \|f\|_p,\qquad k\in \mathbb{N}.\] At the endpoint $p=\infty$ we provide counterexamples for which a linear growth in $d$ does not suffice when $k=1$. We also obtain a counterpart of this result on tail spaces. Namely, for $p>2$ we prove that any function $f:\{-1,1\}^n\to \mathbb{C}$ whose spectrum is bounded from below by $d$ satisfies the upper bound on the decay of the heat semigroup $$ \|e^{-t\Delta}f\|_{p} \le \exp(-c(p,\varepsilon) td) \|f\|_{2}^{1-\theta}\|f\|_{p+\varepsilon}^{\theta},\qquad t>0,$$ and an analogous estimate for $p\in (1,2).$ The constants $c(p,\varepsilon)$ and $C(p,\varepsilon)$ depend only on $p$ and $\varepsilon$; crucially, they are independent of the dimension $n$.

math.FA↗

arXiv2023

Covering the hypercube, the uncertainty principle, and an interpolation formula

We show that the minimal number of skewed hyperplanes that cover the hypercube $\{0,1\}^{n}$ is at least $\frac{n}{2}+1$, and there are infinitely many $n$'s when the hypercube can be covered with $n-\log_{2}(n)+1$ skewed hyperplanes. The minimal covering problems are closely related to uncertainty principle on the hypercube, where we also obtain an interpolation formula for multilinear polynomials on $\mathbb{R}^{n}$ of degree less than $\lfloor n/m \rfloor$ by showing that its coefficients corresponding to the largest monomials can be represented as a linear combination of values of the polynomial over the points $\{0,1\}^{n}$ whose hamming weights are divisible by $m$.

math.CO↗

arXiv2023

Dimension-free discretizations of the uniform norm by small product sets

Let $f$ be an analytic polynomial of degree at most $K-1$. A classical inequality of Bernstein compares the supremum norm of $f$ over the unit circle to its supremum norm over the sampling set of the $K$-th roots of unity. Many extensions of this inequality exist, often understood under the umbrella of Marcinkiewicz-Zygmund-type inequalities for $L^p,1\le p\leq \infty$ norms. We study dimension-free extensions of these discretization inequalities in the high-dimension regime, where existing results construct sampling sets with cardinality growing with the total degree of the polynomial. In this work we show that dimension-free discretizations are possible with sampling sets whose cardinality is independent of $\deg(f)$ and is instead governed by the maximum individual degree of $f$; i.e., the largest degree of $f$ when viewed as a univariate polynomial in any coordinate. For example, we find that for $n$-variate analytic polynomials $f$ of degree at most $d$ and individual degree at most $K-1$, $\|f\|_{L^\infty(\mathbf{D}^n)}\leq C(X)^d\|f\|_{L^\infty(X^n)}$ for any fixed $X$ in the unit disc $\mathbf{D}$ with $|X|=K$. The dependence on $d$ in the constant is tight for such small sampling sets, which arise naturally for example when studying polynomials of bounded degree coming from functions on products of cyclic groups. As an application we obtain a proof of the cyclic group Bohnenblust-Hille inequality with an explicit constant $O(\log K)^{2d}$.

math.FA↗

arXiv2023

Quantum and classical low-degree learning via a dimension-free Remez inequality

Recent efforts in Analysis of Boolean Functions aim to extend core results to new spaces, including to the slice $\binom{[n]}{k}$, the hypergrid $[K]^n$, and noncommutative spaces (matrix algebras). We present here a new way to relate functions on the hypergrid (or products of cyclic groups) to their harmonic extensions over the polytorus. We show the supremum of a function $f$ over products of the cyclic group $\{\exp(2\pi i k/K)\}_{k=1}^K$ controls the supremum of $f$ over the entire polytorus $(\{z\in\mathbf{C}:|z|=1\}^n)$, with multiplicative constant $C$ depending on $K$ and $\text{deg}(f)$ only. This Remez-type inequality appears to be the first such estimate that is dimension-free (i.e., $C$ does not depend on $n$). This dimension-free Remez-type inequality removes the main technical barrier to giving $\mathcal{O}(\log n)$ sample complexity, polytime algorithms for learning low-degree polynomials on the hypergrid and low-degree observables on level-$K$ qudit systems. In particular, our dimension-free Remez inequality implies new Bohnenblust--Hille-type estimates which are central to the learning algorithms and appear unobtainable via standard techniques. Thus we extend to new spaces a recent line of work \cite{EI22, CHP, VZ22} that gave similarly efficient methods for learning low-degree polynomials on the hypercube and observables on qubits. An additional product of these efforts is a new class of distributions over which arbitrary quantum observables are well-approximated by their low-degree truncations -- a phenomenon that greatly extends the reach of low-degree learning in quantum science \cite{CHP}.

math.AP↗

arXiv2022

Slicing all Edges of an $n$-cube Requires $n^{2/3}$ Hyperplanes

Consider the $n$-cube graph with vertices $\{-1,1\}^n$ and edges connecting vertices with hamming distance $1$. How many hyperplanes in $\mathbb{R}^n$ are needed in order to dissect all edges? We show that at least $\widetilde{\Omega}(n^{2/3})$ are needed, which improves the previous bound of $\Omega(n^{0.51})$ by Yehuda and Yehudayoff.

math.CO↗

arXiv2022

Locality-Preserving Hashing for Shifts with Connections to Cryptography

Can we sense our location in an unfamiliar environment by taking a sublinear-size sample of our surroundings? Can we efficiently encrypt a message that only someone physically close to us can decrypt? To solve this kind of problems, we introduce and study a new type of hash functions for finding shifts in sublinear time. A function $h:\{0,1\}^n\to \mathbb{Z}_n$ is a $(d,\delta)$ {\em locality-preserving hash function for shifts} (LPHS) if: (1) $h$ can be computed by (adaptively) querying $d$ bits of its input, and (2) $\Pr [ h(x) \neq h(x \ll 1) + 1 ] \leq \delta$, where $x$ is random and $\ll 1$ denotes a cyclic shift by one bit to the left. We make the following contributions. * Near-optimal LPHS via Distributed Discrete Log: We establish a general two-way connection between LPHS and algorithms for distributed discrete logarithm in the generic group model. Using such an algorithm of Dinur et al. (Crypto 2018), we get LPHS with near-optimal error of $\delta=\tilde O(1/d^2)$. This gives an unusual example for the usefulness of group-based cryptography in a post-quantum world. We extend the positive result to non-cyclic and worst-case variants of LPHS. * Multidimensional LPHS: We obtain positive and negative results for a multidimensional extension of LPHS, making progress towards an optimal 2-dimensional LPHS. * Applications: We demonstrate the usefulness of LPHS by presenting cryptographic and algorithmic applications. In particular, we apply multidimensional LPHS to obtain an efficient "packed" implementation of homomorphic secret sharing and a sublinear-time implementation of location-sensitive encryption whose decryption requires a significantly overlapping view.

cs.DS↗

arXiv2021

Fine-Grained Cryptanalysis: Tight Conditional Bounds for Dense k-SUM and k-XOR

An average-case variant of the $k$-SUM conjecture asserts that finding $k$ numbers that sum to 0 in a list of $r$ random numbers, each of the order $r^k$, cannot be done in much less than $r^{\lceil k/2 \rceil}$ time. On the other hand, in the dense regime of parameters, where the list contains more numbers and many solutions exist, the complexity of finding one of them can be significantly improved by Wagner's $k$-tree algorithm. Such algorithms for $k$-SUM in the dense regime have many applications, notably in cryptanalysis. In this paper, assuming the average-case $k$-SUM conjecture, we prove that known algorithms are essentially optimal for $k= 3,4,5$. For $k>5$, we prove the optimality of the $k$-tree algorithm for a limited range of parameters. We also prove similar results for $k$-XOR, where the sum is replaced with exclusive or. Our results are obtained by a self-reduction that, given an instance of $k$-SUM which has a few solutions, produces from it many instances in the dense regime. We solve each of these instances using the dense $k$-SUM oracle, and hope that a solution to a dense instance also solves the original problem. We deal with potentially malicious oracles (that repeatedly output correlated useless solutions) by an obfuscation process that adds noise to the dense instances. Using discrete Fourier analysis, we show that the obfuscation eliminates correlations among the oracle's solutions, even though its inputs are highly correlated.

cs.CR↗

arXiv2021

Probability Mass of Rademacher Sums Beyond One Standard Deviation

Let $a_1, \dots, a_n \in \mathbb{R}$ satisfy $\sum_i a_i^2 = 1$, and let $\varepsilon_1, \ldots, \varepsilon_n$ be uniformly random $\pm 1$ signs and $X = \sum_{i=1}^{n} a_i \varepsilon_i$. It is conjectured that $X = \sum_{i=1}^{n} a_i \varepsilon_i$ has $\Pr[X \geq 1] \geq 7/64$. The best lower bound so far is $1/20$, due to Oleszkiewicz. In this paper we improve this to $\Pr[X \geq 1] \geq 6/64$.

math.CO↗

arXiv2020

Proof of Tomaszewski's Conjecture on Randomly Signed Sums

We prove the following conjecture, due to Tomaszewski (1986): Let $X= \sum_{i=1}^{n} a_{i} x_{i}$, where $\sum_i a_i^2=1$ and each $x_i$ is a uniformly random sign. Then $\Pr[|X|\leq 1] \geq 1/2$. Our main novel tools are local concentration inequalities and an improved Berry-Esseen inequality for Rademacher sums.

math.CO↗

arXiv2019

Quantum speedups need structure

We prove the following conjecture, raised by Aaronson and Ambainis in 2008: Let $f:\{-1,1\}^n \rightarrow [-1,1]$ be a multilinear polynomial of degree $d$. Then there exists a variable $x_i$ whose influence on $f$ is at least $\mathrm{poly}(\mathrm{Var}(f)/d)$. As was shown by Aaronson and Ambainis, this result implies the following well-known conjecture on the power of quantum computing, dating back to 1999: Let $Q$ be a quantum algorithm that makes $T$ queries to a Boolean input and let $ε,δ> 0$. Then there exists a deterministic classical algorithm that makes $\mathrm{poly}(T,1/ε,1/δ)$ queries to the input and that approximates $Q$'s acceptance probability to within an additive error $ε$ on a $1-δ$ fraction of inputs. In other words, any quantum algorithm can be simulated on most inputs by a classical algorithm which is only polynomially slower, in terms of query complexity.

cs.CC↗

arXiv2019

Biased halfspaces, noise sensitivity, and local Chernoff inequalities

A halfspace is a function $f\colon\{-1,1\}^n \rightarrow \{0,1\}$ of the form $f(x)=\mathbb{1}(a\cdot x>t)$, where $\sum_i a_i^2=1$. We show that if $f$ is a halfspace with $\mathbb{E}[f]=ε$ and $a'=\max_i |a_i|$, then the degree-1 Fourier weight of $f$ is $W^1(f)=Θ(ε^2 \log(1/ε))$, and the maximal influence of $f$ is $I_{\max}(f)=Θ(ε\min(1,a' \sqrt{\log(1/ε)}))$. These results, which determine the exact asymptotic order of $W^1(f)$ and $I_{\max}(f)$, provide sharp generalizations of theorems proved by Matulef, O'Donnell, Rubinfeld, and Servedio, and settle a conjecture posed by Kalai, Keller and Mossel. In addition, we present a refinement of the definition of noise sensitivity which takes into consideration the bias of the function, and show that (like in the unbiased case) halfspaces are noise resistant, and, in the other direction, any noise resistant function is well correlated with a halfspace. Our main tools are 'local' forms of the classical Chernoff inequality, like the following one proved by Devroye and Lugosi (2008): Let $\{ x_i \}$ be independent random variables uniformly distributed in $\{-1,1\}$, and let $a_i\in\mathbb{R}_+$ be such that $\sum_i a_{i}^{2}=1$. If for some $t\geq 0$ we have $\Pr[\sum_{i} a_i x_i > t]=ε$, then $\Pr[\sum_{i} a_i x_i>t+δ]\leq \fracε{2}$ holds for $δ\leq c/\sqrt{\log(1/ε)}$, where $c$ is a universal constant.

math.CO↗

arXiv2019

A structure theorem for almost low-degree functions on the slice

The Fourier-Walsh expansion of a Boolean function $f \colon \{0,1\}^n \rightarrow \{0,1\}$ is its unique representation as a multilinear polynomial. The Kindler-Safra theorem (2002) asserts that if in the expansion of $f$, the total weight on coefficients beyond degree $k$ is very small, then $f$ can be approximated by a Boolean-valued function depending on at most $O(2^k)$ variables. In this paper we prove a similar theorem for Boolean functions whose domain is the `slice' ${{[n]}\choose{pn}} = \{x \in \{0,1\}^n\colon \sum_i x_i = pn\}$, where $0 \ll p \ll 1$, with respect to their unique representation as harmonic multilinear polynomials. We show that if in the representation of $f\colon {{[n]}\choose{pn}} \rightarrow \{0,1\}$, the total weight beyond degree $k$ is at most $ε$, where $ε= \min(p, 1-p)^{O(k)}$, then $f$ can be $O(ε)$-approximated by a degree-$k$ Boolean function on the slice, which in turn depends on $O(2^{k})$ coordinates. This proves a conjecture of Filmus, Kindler, Mossel, and Wimmer (2015). Our proof relies on hypercontractivity, along with a novel kind of a shifting procedure. In addition, we show that the approximation rate in the Kindler-Safra theorem can be improved from $ε+ \exp(O(k)) ε^{1/4}$ to $ε+ε^2 (2\ln(1/ε))^k/k!$, which is tight in terms of the dependence on $ε$ and misses at most a factor of $2^{O(k)}$ in the lower-order term.

math.CO↗