In this talk, I will discuss recent results on the problem of compressing correlated (pseudo) random strings.

Oblivious linear-function evaluation (OLE) is a secure two-party protocol allowing a receiver to learn any linear combination of a pair of field elements held by a sender. OLE serves as a common building block for secure computation of arithmetic circuits, analogously to the role of oblivious transfer (OT) for boolean circuits. A useful extension of OLE is vector OLE (VOLE), allowing the receiver to learn any linear combination of two vectors held by the sender. In several applications of OLE, one can replace a large number of instances of OLE by a smaller number of long instances of VOLE. This motivates the goal of amortizing the cost of generating long instances of VOLE.

I will present a new approach for generating pseudo-random instances of VOLE via a deterministic local expansion of a pair of short correlated seeds and no interaction. This provides the first example of compressing a non-trivial and cryptographically useful correlation with good concrete efficiency. This result enjoys several applications; I will discuss in particular an application to the construction of non-interactive zero-knowledge proofs with reusable preprocessing.

Our VOLE generators are based on a novel combination of function secret sharing (FSS) for multi-point functions and linear codes in which decoding is intractable. Their security can be based on variants of the learning parity with noise (LPN) assumption over large fields that resist known attacks. I will also discuss recent exciting developments toward the 'dream goal' of compressing arbitrary pseudorandom bilinear correlations.

We show a deterministic simulation (or lifting) theorem for composed problems $f \circ \EQ_n$ where the inner function (the gadget) is Equality on $n$ bits. When $f$ is a total function on $p$ bits, it is easy to show via a rank argument that the communication complexity of $f\circ \EQ_n$ is $\Omega(\deg(f) \cdot n)$. However, there is a surprising counter-example of a partial function $f$ on $p$ bits, such that any completion $f'$ of $f$ has $\deg(f') = \Omega(p)$, and yet $f \circ \EQ_n$ has communication complexity $O(n)$. Nonetheless, we are able to show that the communication complexity of $f \circ \EQ_n$ is at least $D(f) \cdot n$ for a complexity measure $D(f)$ which is closely related to the $\AND$-query complexity of $f$ and is lower-bounded by the logarithm of the leaf complexity of $f$. As a corollary, we also obtain lifting theorems for the set-disjointness gadget, and a lifting theorem in the context of parity decision-trees, for the $\NOR$ gadget.

In this talk, I will talk about simulations or lifting theorems in general from a previous work of Chattopadhyay et al., and lifting theorem for Equality in particular—especially why the general recipe of Chattopadhyay et al. does not work for Equality. I will also mention an application of this technique to prove tight communication lower bound for Gap-Equality problem. This is a joint work with Bruno Loff.

We present a new model of probabilistically checkable proof (PCP), which we call “Distributed PCP”: A satisfying assignment (x in {0,1}^n) to a SAT instance is shared between two parties (Alice knows x_1, … x_{n/2}, and Bob knows x_{n/2+1},…,x_n). Their goal is to jointly write a PCP for x, while exchanging little or no information.

Using our new framework, we obtain, for the first time, PCP-like hardness of approximation results for problems in P.

Based on joint works with Amir Abboud, Lijie Chen, Shafi Goldwasser, Kaifeng Lyu, Guy Rothblum, and Ryan Williams.

It is a fundamental problem in the study of computational complexity to understand if access to more resources would mean strictly more computational power. In classical complexity, we have seen such examples in terms of Time Hierarchy and Space Hierarchy theorems. Here, time and space are the resources. It is natural to ask such a question in the setting of algebraic complexity setting. Here, access to more resources translates to allowing the model to perform more number of operations which in turn is allowing the “algebraic formulas” to have larger size.

Arithmetic formulas are directed trees where the leaves are labelled by variables and elements of the field, and internal vertices are labelled by + or x. Every internal vertex computes a polynomial by operating on its inputs. It is easy to see that they are a natural model of computation of polynomials, syntactically (symbolically). The size of the arithmetic formulas refers to the number of + and x operations needed to compute the polynomial.

Rephrasing the question in the algebraic setting we ask the following: for any s, \varepsilon >0, are there polynomial computations that can be efficiently computed by arithmetic formulas of size s but not by any arithmetic formula of size s^{1-\varepsilon}?

In this talk, we will restrict ourselves to arithmetic formulas where computation at every vertex is a multilinear polynomial and here we show explicit separations between the expressive powers of multilinear formulas of small-depth and all polynomial sizes. The multilinear restriction is a reasonable one since most polynomials of interest are indeed multilinear.

Formally, we will show that for any s = poly(n) and \delta > 0, we show that there are explicit polynomial families that have multilinear formulas of size at most s but no 'small'-depth multilinear formulas of size less than s^{0.5 - \delta}. Our proof can be viewed as a derandomization of a lower bound technique of Raz (JACM 2009) using \epsilon-biased spaces.

(Joint work with Nutan Limaye and Srikanth Srinivasan. Appeared in the proceedings of ICALP 2018.)

We consider popular matching problems in a stable marriage instance G. A matching M is popular if it does not lose a head-to-head election against any matching, where each vertex casts a vote for the matching where it gets a better assignment. Popular matchings always exist in G since every stable matching is popular. Algorithmic questions for several popular matching problems have been studied in the last few years: these include finding a max-size popular matching, finding a popular matching with our favorite edge(s) in it, finding a max-weight popular (mixed) matching when there are edge weights. We will see efficient algorithms for some of these problems.

Interestingly, all tractability results in popular matchings seem to reduce the popular matching problem to an appropriate question in either stable matchings or dominant matchings. Dominant matchings always exist in G and form a subclass of max-size popular matchings while stable matchings form a subclass of min-size popular matchings. We recently showed that deciding if G admits a popular matching that is neither stable nor dominant is NP-hard. This implies a tight 1/2-factor approximation for the max-weight popular matching problem in G and the hardness of finding a popular matching in a non-bipartite graph.

[Joint work with Agnes Cseh, Yuri Faenza, Chien-Chung Huang, Vladlena Powers, and Xingyu Zhang.]

Following the works of Drineas et al. (2004), Randomised Numerical Linear Algebra (RNLA) applications have enjoyed a increasing surge of interest in the classical algorithms community. Following their success we recently demonstrated a classical algorithm based on the Nystroem method which allows us to efficiently simulate dynamics of quantum system with specific structural conditions. We briefly introduce the most basic RNLA algorithm by Drineas et al. to give an overview of the topic and then outline our own recent work. We follow with a discussion of potential application of related methods, including spectral sparsification, in quantum algorithms for Hamiltonian Simulation and quantum linear systems; We finish with a short description of own current work related to spectral sparsification (Spielmann & Teng 2011) for Hamiltonian simulation.

Computational learning theory provides a mathematical framework for rigorously formulating learning problems from both a statistical and computational perspective. During this talk I will present two independent results at the interplay of learning theory and quantum computation. First, I will address the problem: “can we learn faster with a quantum computer?”. In particular, I will discuss the learnability of the class of boolean functions that can be expressed as polynomial size formulae in disjunctive normal form (DNF) and show that this is efficiently quantum PAC learnable under product distributions. The learnability of DNFs is a central problem in learning theory and the best known classical algorithm (using standard ways to access the function) runs in superpolynomial time. Second, I will discuss the problem “what it means to learn a quantum state and can we do that efficiently?”. Here, building on a model developed by Aaronson in 2007, I will present a class of states, namely stabiliser states, that can be learned using only a linear number of measurements and polynomial - classical - computation. Part of the results are joint work with Varun Kanade and Simone Severini.

The work of Korman and others, based on investigations of navigational techniques of ‘crazy ants’, leads to the following problem: Let Q be a network with given arc lengths, and let H be a ‘home’ node. At each node a permanent direction is given (one of the adjacent arcs) which leads to a shortest path path from that node to H with probability p less than one. Starting from a node S, a searcher chooses the given direction at each reached node with probability q; otherwise he chooses randomly. He arrives home at node H in expected time T=T(S,H,p,q). How should q be chosen to minimize T? Perhaps when reaching a node the searcher sees its degree k so chooses the given direction with probability q(k). Related problems have been studied for tree from a graph theoretical perspective.

I will describe recent results in the QRAM model that improve the sparsity dependance for quantum linear system solvers to \mu(A) = \min (||A||_F, ||A||_1) where ||A||_1 is the maximum l-1 norm of the rows and columns of A and ||A||_F is the Frobenius norm. These results achieve a worst case quadratic speedup over the HHL algorithm and its improvements and exponential speedups for some cases. I will also present some applications of the improved linear system solvers to quantum machine learning.

We present a collection of results about the clock in Feynman's computer construction and Kitaev's Local Hamiltonian problem. First, by analyzing the spectra of quantum walks on a line with varying endpoint terms, we find a better lower bound on the gap of the Feynman Hamiltonian, which translates into a less strict promise gap requirement for the QMA-complete Local Hamiltonian problem. We also translate this result into the language of adiabatic quantum computation. Second, introducing an idling clock construction with a large state space but fast Cesaro mixing, we provide a way for achieving an arbitrarily high success probability of computation with Feynman's computer with only a logarithmic increase in the number of clock qubits. Finally, we tune and thus improve the costs (locality, gap scaling) of implementing a (pulse) clock with a single excitation.

Joint work with Zeph Landau and Daniel Nagaj

We give an efficient (polylog(n) rounds) algorithm that colors a graph with few colors in the distributed setting (LOCAL model). Among other things, we will see that this algorithm 6-colors planar graphs (using 1 less color than the previous best algorithm of Goldberg, Plotkin, and Shannon) and will show that one cannot 4-color planar graph in o(n) rounds.

This is a joint work with Bousquet, Bonamy, and Esperet.

An online algorithm is a well-known computational model for solving optimization problems. The defining property of this model is following. An algorithm reads an input piece by piece and should return output variables after some of the input variables immediately, even if the answer depends on the whole input. An online algorithm should return output for minimizing an objective function.

We consider streaming algorithms and two-way automata as models for online algorithms. We compare quantum and classical models in case of logarithmic memory and sublogarithmic memory.

We get the following results for online streaming algorithms: - a quantum online streaming algorithm with 1 qubit of memory and 1 advice bit can be better than a classical online streaming algorithm with $o(\log n)$ bits of memory and $o(\log n)$ advice bits. - Quantum online streaming algorithm with a constant size of memory and $t$ advice bits can be better than deterministic online streaming algorithms with $t$ advice bits and unlimited computational power. - In a case of a polylogarithmic size of memory, quantum online streaming algorithms can be better than classical ones even if they have advice bits.

We get the following results for two way automata as an online algorithm for solving online minimization problems: - a two way automata with quantum and classical states for online minimization problems with a constant size of memory can be better than classical ones even if they have advice bits.

In defective coloring, we are given a graph G and two integers c, ∆* and are asked if we can c-color G so that the maximum degree induced by any color class is at most ∆*. This problem is a generalization of proper coloring, for which Δ* = 0.

Defective coloring, like proper coloring, has many applications, for example in scheduling (assign jobs to machines which can work on up to Δ* jobs in parallel), or in radio networks (assign frequencies to antennas with some slack, that is, an antenna can be assigned the same frequency with up to Δ* other antennas within its range without a signal conflict).

We will present some algorithmic and complexity results on this problem in classes of graphs where proper coloring is easy: on the one hand, we will consider subclasses of perfect graphs, namely cographs and chordal graphs; on the other hand we will talk about structural parameterizations, such as treewidth and feedback vertex set.

Joint work with Rémy Belmonte (University of Electro-Communications, Tokyo) and Michael Lampis (Lamsade, Université Paris Dauphine) that appeared in WG '17 and in STACS '18.

We show that quantum expander codes, a constant-rate family of quantum LDPC codes, with the quasi-linear time decoding algorithm of Leverrier, Tillich and Zémor can correct a constant fraction of random errors with very high probability. This is the first construction of a constant-rate quantum LDPC code with an efficient decoding algorithm that can correct a linear number of random errors with a negligible failure probability. Finding codes with these properties is also motivated by Gottesman’s construction of fault tolerant schemes with constant space overhead. In order to obtain this result, we study a notion of α-percolation: for a random subset E of vertices of a given graph, we consider the size of the largest connected α-subset of E, where X is an α-subset of E if |X ∩ E| ≥ α|X|.

Local-search is an intuitive approach towards solving combinatorial optimization problems: start with any feasible solution, and try to improve it by local improvements. Like other greedy approaches, it can fail to find the global optimum by getting stuck on a locally optimal solution. In this talk I will present the ideas and techniques behind the use of local-search in the design of provably good approximation algorithms for some combinatorial problems, such as independent sets, vertex cover, dominating sets in geometric intersection graphs. The key unifying theme is the analysis of local expansion in planar graphs. Joint work with Norbert Bus, Shashwat Garg, Bruno Jartoux and Saurabh Ray.

We study the online maximum matching problem with recourse in a model in which the edges are associated with a known recourse parameter k. An online algorithm for this problem has to maintain a valid matching while edges of the underlying graph are presented one after the other. At any moment the algorithm can decide to include an edge into the matching or to exclude it, under the restriction that at most k actions per edge take place, where k is typically a small constant. This problem was introduced and studied in the context of general online packing problems with recourse by [Avitabile, Mathieu, Parkinson, 2013], whereas the special case k=2 has been studied by [Boyar et al., 2017].

In the first part of this paper, we consider the edge arrival model, in which an arriving edge never disappears from the graph. Here, we first show an improved analysis on the performance of the algorithm given in [AMP, 2013], by exploiting the structure of the specific problem. In addition, we extend the result of [Boyar et al., 2017] and show that the greedy algorithm has ratio 3/2 for every even positive k and ratio 2 for every odd k. Moreover, we present and analyze L-Greedy — an improvement of the greedy algorithm — which for small values of k outperforms the algorithm of [AMP, 2013]. In terms of lower bounds, we show that no deterministic algorithm better than 1+1/(k−1) exists, improving upon the lower bound of 1+1/k shown in [AMP, 2013].

The second part of the paper is devoted to the edge arrival/departure model, which is the fully dynamic variant of the online matching problem with recourse. The analysis of L-Greedy carries through in this model; moreover we show a general lower bound of (k2−3k+3)/(k2−4k+5) for all even k≥4 and provide the stronger bound of 10/7 for k=4. For k∈{2,3}, the competitive ratio is 3/2.

Joint work with Spyros Angelopoulos and Christoph Dürr.

Eighth lecture from Claire Mathieu at Collège de France on Algorithms (at 10am), followed by a talk from Allison Bishop on Algorithms Operating in Adversarial Conditions (at 11am).

Additional information: http://www.college-de-france.fr/site/claire-mathieu/course-2018-01-30-10h00.htm

Seventh lecture from Claire Mathieu at Collège de France on Algorithms (at 10am), followed by a talk from Tim Roughgarden on Game Theory Through the Computational Lens (at 11am).

Additional information: http://www.college-de-france.fr/site/claire-mathieu/course-2018-01-23-10h00.htm

We study a variant of the k-server problem, the infinite server problem, in which infinitely many servers reside initially at a particular point of the metric space and serve a sequence of requests. In the framework of competitive analysis, we show a surprisingly tight connection between this problem and the (h,k)-server problem, in which an online algorithm with k servers competes against an offline algorithm with h servers. Specifically, we show that the infinite server problem has bounded competitive ratio if and only if the (h,k)-server problem has bounded competitive ratio for some k=O(h). We give a lower bound of 3.146 for the competitive ratio of the infinite server problem, which implies the same lower bound for the (h,k)-server problem even when k/h→∞ and holds also for the line metric; the previous known bounds were 2.4 for general metric spaces and 2 for the line. For weighted trees and layered graphs we obtain upper bounds, although they depend on the depth. Of particular interest is the infinite server problem on the line, which we show to be equivalent to the seemingly easier case in which all requests are in a fixed bounded interval away from the original position of the servers. This is a special case of a more general reduction from arbitrary metric spaces to bounded subspaces. Unfortunately, classical approaches (double coverage and generalizations, work function algorithm, balancing algorithms) fail even for this special case.

Sixth lecture from Claire Mathieu at Collège de France on Algorithms (at 10am), followed by a talk from Jon Kleinberg on Algorithms and Fairness (at 11am).

Additional information: http://www.college-de-france.fr/site/claire-mathieu/course-2018-01-16-10h00.htm

Last spring, I visited François Bergeron and we worked on his favorite objects: the spaces H(n,k) of diagonal harmonic polynomials in k sets of n variables. Those spaces connect to many areas of algebraic combinatorics, representation theory and beyond, and the case H(n,2) became famous a decade ago with the n! conjecture and its proof by Haiman.

To fuel his ongoing studies François needed to compute the structure of H(5,6). This is a space of dimension 6.10^5 made of polynomials in 30 variables of degree up to 15, each having thousands of terms.

In this talk, I'll explain how the calculation can now be completed in 45 minutes with a dozen cores and ~15Go of memory. This exploits a combination of strategies (symmetries, representation theory of the symmetric and general linear group, …), each of which reduces the complexity in time and memory by one or two orders of magnitude.

There will be little prerequisites and it's my hope that some strategies (and maybe the code!) could be used in other contexts.

Fifth lecture from Claire Mathieu at Collège de France on Algorithms (at 10am), followed by a talk from Mark Jerrum on Sampling and Approximate Counting (at 11am).

Additional information: http://www.college-de-france.fr/site/claire-mathieu/course-2018-01-09-10h00.htm