Claude Gravel
Claude Gravel has a total of 25 co-authors. Here are the top co-authors by number of shared publications:
- Daniel Panario
Carleton University
7 shared publications - David Thomson
Carleton University
2 shared publications - Luc Devroye
McGill University
2 shared publications - Andriy Miranskyy
Toronto Metropolitan University, Toronto, ON, Canada
1 shared publication - Anila Mjeda
Munster Technological University, Cork, Ireland
1 shared publication Asuka Wakasugi
EAGLYS Inc., Tokyo, Japan
1 shared publication
Latest Publications
- Open Access
- 3 July 2026
- Published by Elsevier in Finite Fields and Their Applications
- Vol. 117
Cycle structure of discrete functions extended through linear transformations
- 3 July 2026
- Published by Elsevier in Finite Fields and Their Applications
- Vol. 117
Open AccessCycle structure of discrete functions extended through linear transformations
- 31 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Computer
- Vol. 59 (4) , 135-138
Software Testing in the Quantum World
As quantum software grows in complexity, the classical simulation of quantum computers becomes infeasible. This article presents the key challenges in testing large-scale quantum software and offers software engineering perspectives for addressing them.
- 31 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Computer
- Vol. 59 (4) , 135-138
Software Testing in the Quantum World
As quantum software grows in complexity, the classical simulation of quantum computers becomes infeasible. This article presents the key challenges in testing large-scale quantum software and offers software engineering perspectives for addressing them.
- Open Access
- 16 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 69980-69997
An NVMe-Based Secure Computing Platform With FPGA-Based TFHE Accelerator
In this study, we introduce a new approach to secure computing by implementing a platform that utilizes a non-volatile memory express (NVMe)-based system with an FPGA-based Torus fully homomorphic encryption (TFHE) accelerator, solid state drive (SSD), and middleware on the host-side. Our platform is the first to offer completely secure computing capabilities for TFHE by using an FPGA-based accelerator. We defined secure computing instructions to evaluate 14-bit to 14-bit functions using TFHE. Our middleware allows for the communication of ciphertexts, keys, and secure computing programs while invoking secure computing programs through NVMe commands with metadata. Our performance evaluation demonstrates that our secure computing platform outperforms CPU-based and GPU-based platforms by 15 to 120 times and 2.5 to 3 times, respectively, in gate bootstrapping execution time. Additionally, our platform uses 7 to 12 times less electric energy consumption during the gate bootstrapping execution time than CPU-based platforms and 4.95 times less than a GPU-based platform. The performance of a machine learning application running on our platform shows that bootstrapping accounts for more than 80% of ciphertext learning time.
- 16 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 69980-69997
Open AccessAn NVMe-Based Secure Computing Platform With FPGA-Based TFHE Accelerator
In this study, we introduce a new approach to secure computing by implementing a platform that utilizes a non-volatile memory express (NVMe)-based system with an FPGA-based Torus fully homomorphic encryption (TFHE) accelerator, solid state drive (SSD), and middleware on the host-side. Our platform is the first to offer completely secure computing capabilities for TFHE by using an FPGA-based accelerator. We defined secure computing instructions to evaluate 14-bit to 14-bit functions using TFHE. Our middleware allows for the communication of ciphertexts, keys, and secure computing programs while invoking secure computing programs through NVMe commands with metadata. Our performance evaluation demonstrates that our secure computing platform outperforms CPU-based and GPU-based platforms by 15 to 120 times and 2.5 to 3 times, respectively, in gate bootstrapping execution time. Additionally, our platform uses 7 to 12 times less electric energy consumption during the gate bootstrapping execution time than CPU-based platforms and 4.95 times less than a GPU-based platform. The performance of a machine learning application running on our platform shows that bootstrapping accounts for more than 80% of ciphertext learning time.
- Open Access
Evaluating the generalized Buchshtab function and revisiting the variance of the distribution of the smallest components of combinatorial objects
Let $n\geq 1$ and $X_{n}$ be the random variable representing the size of the smallest component of a random combinatorial object made of $n$ elements. A combinatorial object could be a permutation, a monic polynomial over a finite field, a surjective map, a graph, and so on. By a random combinatorial object, we mean a combinatorial object that is chosen uniformly at random among all possible combinatorial objects of size $n$. It is understood that a component of a permutation is a cycle, an irreducible factor for a monic polynomial, a connected component for a graph, etc. Combinatorial objects are categorized into parametric classes. In this article, we focus on the exp-log class with parameter $K=1$ (permutations, derangements, polynomials over finite field, etc.) and $K=1/2$ (surjective maps, $2$-regular graphs, etc.) The generalized Buchstab function $\Omega_{K}$ plays an important role in evaluating probabilistic and statistical quantities. For $K=1$, Theorem $5$ from \cite{PanRic_2001_small_explog} stipulates that $\mathrm{Var}(X_{n})=C(n+O(n^{-\epsilon}))$ for some $\epsilon>0$ and sufficiently large $n$. We revisit the evaluation of $C=1.3070\ldots$ using different methods: analytic estimation using tools from complex analysis, numerical integration using Taylor expansions, and computation of the exact distributions for $n\leq 4000$ using the recursive nature of the counting problem. In general for any $K$, Theorem $1.1$ from \cite{BenMasPanRic_2003} connects the quantity $1/\Omega_{K}(x)$ for $x\geq 1$ with the asymptotic proportion of $n$-objects with large smallest components. We show how the coefficients of the Taylor expansion of $\Omega_{K}(x)$ for $\lfloor x\rfloor \leq x < \lfloor x\rfloor+1$ depends on those for $\lfloor x\rfloor-1 \leq x-1 < \lfloor x\rfloor$. We use this family of coefficients to evaluate $\Omega_{K}(x)$.Open AccessEvaluating the generalized Buchshtab function and revisiting the variance of the distribution of the smallest components of combinatorial objects
Let $n\geq 1$ and $X_{n}$ be the random variable representing the size of the smallest component of a random combinatorial object made of $n$ elements. A combinatorial object could be a permutation, a monic polynomial over a finite field, a surjective map, a graph, and so on. By a random combinatorial object, we mean a combinatorial object that is chosen uniformly at random among all possible combinatorial objects of size $n$. It is understood that a component of a permutation is a cycle, an irreducible factor for a monic polynomial, a connected component for a graph, etc. Combinatorial objects are categorized into parametric classes. In this article, we focus on the exp-log class with parameter $K=1$ (permutations, derangements, polynomials over finite field, etc.) and $K=1/2$ (surjective maps, $2$-regular graphs, etc.) The generalized Buchstab function $\Omega_{K}$ plays an important role in evaluating probabilistic and statistical quantities. For $K=1$, Theorem $5$ from \cite{PanRic_2001_small_explog} stipulates that $\mathrm{Var}(X_{n})=C(n+O(n^{-\epsilon}))$ for some $\epsilon>0$ and sufficiently large $n$. We revisit the evaluation of $C=1.3070\ldots$ using different methods: analytic estimation using tools from complex analysis, numerical integration using Taylor expansions, and computation of the exact distributions for $n\leq 4000$ using the recursive nature of the counting problem. In general for any $K$, Theorem $1.1$ from \cite{BenMasPanRic_2003} connects the quantity $1/\Omega_{K}(x)$ for $x\geq 1$ with the asymptotic proportion of $n$-objects with large smallest components. We show how the coefficients of the Taylor expansion of $\Omega_{K}(x)$ for $\lfloor x\rfloor \leq x < \lfloor x\rfloor+1$ depends on those for $\lfloor x\rfloor-1 \leq x-1 < \lfloor x\rfloor$. We use this family of coefficients to evaluate $\Omega_{K}(x)$. - Open Access
Feedback linearly extended discrete functions
We study a new flexible method to extend linearly the graph of a non-linear, and usually not bijective, function so that the resulting extension is a bijection. Our motivation comes from cryptography. Examples from symmetric cryptography are given as how the extension was used implicitly in the construction of some well-known block ciphers. The method heavily relies on ideas brought from linear coding theory and secret sharing. We are interested in the behaviour of the composition of many extensions, and especially the space of parameters that defines a family of equations based on finite differences or linear forms. For any linear extension, we characterize entirely the space of parameters for which such equations are solvable in terms of the space of parameters that render those equations for the corresponding non-linear extended functions solvable. Conditions are derived to assess the solvability of those kind of equations in terms of the number of compositions or iterations. We prove a relation between the number of compositions and the dimensions of vector spaces that appear in our results. The proofs of those properties rely mostly on tools from linear algebra.
Open AccessFeedback linearly extended discrete functions
We study a new flexible method to extend linearly the graph of a non-linear, and usually not bijective, function so that the resulting extension is a bijection. Our motivation comes from cryptography. Examples from symmetric cryptography are given as how the extension was used implicitly in the construction of some well-known block ciphers. The method heavily relies on ideas brought from linear coding theory and secret sharing. We are interested in the behaviour of the composition of many extensions, and especially the space of parameters that defines a family of equations based on finite differences or linear forms. For any linear extension, we characterize entirely the space of parameters for which such equations are solvable in terms of the space of parameters that render those equations for the corresponding non-linear extended functions solvable. Conditions are derived to assess the solvability of those kind of equations in terms of the number of compositions or iterations. We prove a relation between the number of compositions and the dimensions of vector spaces that appear in our results. The proofs of those properties rely mostly on tools from linear algebra.
- 17 February 2021
- Published by Springer Nature
- p. 174-192
Finding Linearly Generated Subsequences
- Open Access
Random variate generation using only finitely many unbiased, independently and identically distributed random bits
For any discrete probability distributions with bounded entropy, we can generate exactly a random variate using only a finite expected number of perfect coin flips. A perfect coin flip is the outcome of an unbiased Bernoulli random variable. Coin flips are unbiased, independently and identically distributed in all our work. We survey well-known algorithms for the discrete case such as the one from Knuth and Yao as well as the one from Han and Hoshi. We also discuss briefly about a practical implementation for the algorithm proposed by Knuth and Yao. For the continuous case, only approximations can be hoped for. The freedom to choose the accuracy for the approximations matters, and, for that, we propose to measure accuracy in terms of the Wasserstein $L_\infty$-metric. We derive a universal lower bound for the expected number of perfect coin flips required to reach a desired accuracy. We also provide several algorithms for absolutely continuous distributions that come within our universal lower bound.
Open AccessRandom variate generation using only finitely many unbiased, independently and identically distributed random bits
For any discrete probability distributions with bounded entropy, we can generate exactly a random variate using only a finite expected number of perfect coin flips. A perfect coin flip is the outcome of an unbiased Bernoulli random variable. Coin flips are unbiased, independently and identically distributed in all our work. We survey well-known algorithms for the discrete case such as the one from Knuth and Yao as well as the one from Han and Hoshi. We also discuss briefly about a practical implementation for the algorithm proposed by Knuth and Yao. For the continuous case, only approximations can be hoped for. The freedom to choose the accuracy for the approximations matters, and, for that, we propose to measure accuracy in terms of the Wasserstein $L_\infty$-metric. We derive a universal lower bound for the expected number of perfect coin flips required to reach a desired accuracy. We also provide several algorithms for absolutely continuous distributions that come within our universal lower bound.
- Open Access
Finding linearly generated subsequences
We develop a new algorithm to compute determinants of all possible Hankel matrices made up from a given finite length sequence over a finite field. Our algorithm fits within the dynamic programming paradigm by exploiting new recursive relations on the determinants of Hankel matrices together with new observations concerning the distribution of zero determinants among the possible matrix sizes allowed by the length of the original sequence. The algorithm can be used to isolate \emph{very} efficiently linear shift feedback registers hidden in strings with random prefix and random postfix for instance and, therefore, recovering the shortest generating vector. Our new mathematical identities can be used also in any other situations involving determinants of Hankel matrices. We also implement a parallel version of our algorithm. We compare our results empirically with the trivial algorithm which consists of computing determinants for each possible Hankel matrices made up from a given finite length sequence. Our new accelerated approach on a single processor is faster than the trivial algorithm on 160 processors for input sequences of length 16384 for instance.
Open AccessFinding linearly generated subsequences
We develop a new algorithm to compute determinants of all possible Hankel matrices made up from a given finite length sequence over a finite field. Our algorithm fits within the dynamic programming paradigm by exploiting new recursive relations on the determinants of Hankel matrices together with new observations concerning the distribution of zero determinants among the possible matrix sizes allowed by the length of the original sequence. The algorithm can be used to isolate \emph{very} efficiently linear shift feedback registers hidden in strings with random prefix and random postfix for instance and, therefore, recovering the shortest generating vector. Our new mathematical identities can be used also in any other situations involving determinants of Hankel matrices. We also implement a parallel version of our algorithm. We compare our results empirically with the trivial algorithm which consists of computing determinants for each possible Hankel matrices made up from a given finite length sequence. Our new accelerated approach on a single processor is faster than the trivial algorithm on 160 processors for input sequences of length 16384 for instance.
- 19 July 2019
- Published by Springer Nature in Cryptography and Communications
- Vol. 11 (6) , 1211-1231
Unicyclic strong permutations
- 19 July 2019
- Published by Springer Nature in Cryptography and Communications
- Vol. 11 (6) , 1211-1231
Unicyclic strong permutations
- Open Access
Unicyclic Strong Permutations
In this paper, we study some properties of a certain kind of permutation $\sigma$ over $\mathbb{F}_{2}^{n}$, where $n$ is a positive integer. The desired properties for $\sigma$ are: (1) the algebraic degree of each component function is $n-1$; (2) the permutation is unicyclic; (3) the number of terms of the algebraic normal form of each component is at least $2^{n-1}$. We call permutations that satisfy these three properties simultaneously unicyclic strong permutations. We prove that our permutations $\sigma$ always have high algebraic degree and that the average number of terms of each component function tends to $2^{n-1}$. We also give a condition on the cycle structure of $\sigma$. We observe empirically that for $n$ even, our construction does not provide unicylic permutations. For $n$ odd, $n \leq 11$, we conduct an exhaustive search of all $\sigma$ given our construction for specific examples of unicylic strong permutations. We also present some empirical results on the difference tables and linear approximation tables of $\sigma$.Open AccessUnicyclic Strong Permutations
In this paper, we study some properties of a certain kind of permutation $\sigma$ over $\mathbb{F}_{2}^{n}$, where $n$ is a positive integer. The desired properties for $\sigma$ are: (1) the algebraic degree of each component function is $n-1$; (2) the permutation is unicyclic; (3) the number of terms of the algebraic normal form of each component is at least $2^{n-1}$. We call permutations that satisfy these three properties simultaneously unicyclic strong permutations. We prove that our permutations $\sigma$ always have high algebraic degree and that the average number of terms of each component function tends to $2^{n-1}$. We also give a condition on the cycle structure of $\sigma$. We observe empirically that for $n$ even, our construction does not provide unicylic permutations. For $n$ odd, $n \leq 11$, we conduct an exhaustive search of all $\sigma$ given our construction for specific examples of unicylic strong permutations. We also present some empirical results on the difference tables and linear approximation tables of $\sigma$.