INRIA a CCSD electronic archive server
Not a member yet
122212 research outputs found
Sort by
Biology and Clinical Management of Non-V600 BRAF Alterations in NSCLC
No full textInternational audienc
Statistical estimation of a mean-field fitzhugh-nagumo model
No full textWe consider an interacting system of particles with value in R d × R d , governed by transport and diffusion on the first component, on that may serve as a representative model for kinetic models with a degenerate component. In a first part, we control the fluctuations of the empirical measure of the system around the solution of the corresponding Vlasov-Fokker-Planck equation by proving a Bernstein concentration inequality, extending a previous result of [DMH22] in several directions. In a second part, we study the nonparametric statistical estimation of the classical solution of Vlasov-Fokker-Planck equation from the observation of the empirical measure and prove an oracle inequality using the Goldenshluger-Lepski methodology and we obtain minimax optimality. We then specialise on the FitzHugh-Nagumo model for populations of neurons. We consider a version of the model proposed in Mischler et al. [MQT16] an optimally estimate the 6 parameters of the model by moment estimators.</div
Diagonals and algebraicity modulo : a sharper degree bound
No full textTo appear in the Annales scientifiques de l'École normale supérieure. A longer version of this work is available at https://arxiv.org/abs/2306.02640International audienceIn 1984, Deligne proved that for any prime number , the reduction modulo of the diagonal of a multivariate algebraic power series with integer coefficients is algebraic over the field of rational functions with coefficients in . Moreover, he conjectured that the algebraic degrees of these functions should grow at most polynomially in . In this article, we provide a new and elementary proof of Deligne's theorem, which yields the first general polynomial bound on with an explicit and reasonable degree.En 1984, Deligne a montr\'e que pour tout nombre premier~, la r\'eduction modulo de la diagonale d'une s\'erie formelle alg\'ebrique de plusieurs variables \`a coefficients entiers est alg\'ebrique sur le corps des fonctions rationnelles \`a coefficients dans . De plus, il a sugg\'er\'e que les degr\'es d'alg\'ebricit\'e de ces fonctions devaient cro\^itre au plus polynomialement en fonction de . Dans cet article, nous pr\'esentons une nouvelle preuve du th\'eor\`eme de Deligne qui est \'el\'ementaire et permet d'\'etablir la premi\`ere borne g\'en\'erale polynomiale avec un degr\'e raisonnable
Théorème de décomposition de type Smirnov sur les courbes ouvertes simples et régulières et étude de leurs propriétés
No full textWe present a new decomposition of divergence-measure vector fields in dimension two into simple open regular curves with constant speed. Our construction builds on Smirnov's decomposition theorem and adapts it to a framework that allows a careful analysis of the support and regularity of the resulting curves. This provides a rigorous representation of divergence-measure fields via simple constant-speed curves, which can be useful for applications in geometric measure theory and imaging.Nous présentons une nouvelle décomposition des mesures de Radon vectorielles de dimension deux à divergence finie en courbes ouvertes simples et régulières, paramétrées à vitesse constante. Notre construction s’appuie sur le théorème de décomposition de Smirnov et l’adapte à un cadre permettant une analyse fine du support et de la régularité des courbes obtenues. Elle fournit ainsi une représentation rigoureuse des champs à divergence mesure au moyen de courbes simples à vitesse constante, utile pour des applications en théorie géométrique de la mesure et en imagerie
A Global Robust Finite-Time Control for the Heisenberg System: Simultaneous Regulation and Tracking
No full textInternational audienceIn this paper, a novel hybrid control strategy is proposed to ensure the global robust finite-time regulation and tracking of any admissible reference for the Heisenberg system, sometimes also known as the Brockett integrator. The proposed scheme is comprised of a local controller, ensuring the finite-time attractivity of the origin of the error vector; a global controller, providing finite-time convergence to the region of attraction of the local one; and a state-dependent switching condition, which combines both controllers into a single hybrid structure. The resulting strategy provides settling-time estimations and is straightforward to tune. Some simulations considering possible applications, such as the tracking of geodesic curves, are included to illustrate the implementability of the results
Guiding Polyhedral Scheduling for Vectorization through Constraints Generated from an SLP Algorithm
No full textInternational audiencePolyhedral schedulers present well established techniques to extract parallelism, improve data locality, and generate tiled code for statically analyzable loops. However, as the polyhedral model abstracts programs in a mathematical representation detached from language, architectural, and hardware specific constraints, encoding vectorization in an affine form can prove challenging.In this paper, we present an approach to integrate information on vectorization decisions made by an SLP algorithm (Autovesk) into a polyhedral compiler (Pluto) through the addition of constraints to the schedule. We execute the SLP vectorization algorithm preserving annotated statement instance information. From its output, we create a set of constraints aiming to enforce vectorization. Those optional constraints are injected during the scheduling process of the polyhedral compiler. We evaluate the performance and make use of hardware counters to check the relevancy of our method on the Polybench/C suite
Invariants in Linear Optics
No full textLinear optics (LO) prohibits certain transformations. In this paper, we study the conditions for a computation to be possible in LO. We find that there are finitely many polynomials such that each of these polynomials evaluates to the same value on two photonic states if and only if there is a LO circuit transforming one of these states into the other. The proof is non-constructive, so we then focus on methods to find such polynomials
Good quantum codes with addressable and parallelizable non-Clifford gates
No full textIn this work, we prove that for any m > 1, there exists a family of good qudit quantum codes supporting transversal logical C m-1 Z gates that can address specified logical qudits and be largely executed in parallel. Building on the family of good quantum error-correcting codes presented in [HVWZ25a], which support addressable and transversal logical CCZ gates, we extend their framework and show how to perform large sets of gates in parallel. The construction relies on the classical algebraic geometry codes of Stichtenoth [Sti06]. Our results lead to a substantial reduction in the depth overhead of multi-control-Z circuits. In particular, we show that the minimal depth of any logical C m-1 Z circuit involving qudits from m distinct code blocks is upper bounded by O(k m-1 ), where k is the code dimension. While this overhead is optimal for dense C m-1 Z circuits, for sparse circuits we discuss how the depth overhead can be significantly reduced by exploiting the structure of the quantum code
SDManet
No full textThis software is a OMNET++-based simulator for service discovery in heterogeneous networks
Adapting landmark geopositioning for use with imprecise maps
No full textInternational audiencePrecise vehicle geopositioning is essential for autonomous navigation, especially in urban environments where GNSS systems suffer from multipath effects, signal loss, and atmospheric disturbances, making them unreliable for decimeter-level accuracy required for autonomous driving. High-cost solutions like GNSS RTK are impractical for widespread adoption, and traditional mapping methods using specialized vehicle fleets with advanced sensors struggle with scalability and realtime updates in dynamic urban settings. This paper introduces a novel landmark-based geopositioning algorithm designed to achieve robust and accurate vehicle positioning while using a map that may not be totally accurate. Our approach integrates Gaussian estimates of vehicle position and orientation, derived from on-board sensor data, with a digital map containing landmarks known with limited precision. By exploiting the Gaussian nature of the data, the algorithm ensures reliable and efficient geopositioning despite imprecise landmark information. Evaluated in our custom simulator, our method achieves decimeter-level accuracy with a 40 ms update rate, even with complete GNSS signal loss. The algorithm is sensor-agnostic and vehicle-independent, enhancing its applicability. This approach offers a cost-effective, adaptive solution for autonomous navigation, addressing the limitations of traditional geopositioning and mapping methods in complex urban environments