A5088706134- Numerical Methods and Algorithms
- Digital Filter Design and Implementation
- Parallel Computing and Optimization Techniques
- Polynomial and algebraic computation
- Model Reduction and Neural Networks
- Low-power high-performance VLSI design
- Neural Networks and Applications
- Cryptography and Residue Arithmetic
- Computability, Logic, AI Algorithms
- Matrix Theory and Algorithms
- Advancements in PLL and VCO Technologies
- Logic, programming, and type systems
- Embedded Systems Design Techniques
- Formal Methods in Verification
- History and Theory of Mathematics
- Advanced Electrical Measurement Techniques
- Distributed and Parallel Computing Systems
- Computational Physics and Python Applications
- Advanced Optimization Algorithms Research
- Control Systems and Identification
- Advanced Data Storage Technologies
- Natural Fiber Reinforced Composites
- Iterative Methods for Nonlinear Equations
- Intelligent Tutoring Systems and Adaptive Learning
- Advanced Clustering Algorithms Research
École Normale Supérieure de Lyon
2007-2023
Centre National de la Recherche Scientifique
2005-2023
Université Claude Bernard Lyon 1
2005-2023
Institut national de recherche en informatique et en automatique
2004-2023
Agruicultural Research Institute
2022
Laboratoire de l'Informatique du Parallélisme
2008-2020
STMicroelectronics (India)
2004
Institut des Arts de Diffusion
2003
Université de Lille
2000-2001
Translational Innovation in Medicine and Complexity
1996
Hemp is finding a strong renewal of interest in the production fine fibers for garment textiles. This resource long-line would come as complement to highly demanded flax fibers, whose large north-west Europe cannot be extended. In Normandy, where complete industrial value chain exists flax, it intended adapt hemp, and this was demonstrated from field scutched with chain. region, early harvesting necessary leave enough time dew-retting permit dry storage stems before mid-September. An...
Range-reduction is a key point for getting accurate elementary function routines. We introduce new algorithm that fast input arguments belonging to the most common domains, yet over full double-precision range.
What is called "numerical reproducibility" the problem of getting same result when scientific computation run several times, either on machine or different machines, with types and numbers processing units, execution environments, computational loads etc. This especially stringent for HPC numerical simulations. In what follows, focus parallel implementations interval arithmetic using floating-point arithmetic. For computations, reproducibility course an issue testing debugging purposes....
We present a framework for validated numerical computations with real functions. The is based on formalisation of abstract data types basic floating-point arithmetic, interval arithmetic and function models Banach algebra. As concrete instantiation, we develop an elementary smooth calculus approximated by sparse polynomial models. demonstrate formal verification applied to operations in theorem prover. ultimate aim formalism powerful enough reachability analysis nonlinear hybrid systems.
Questions whether numerical simulation is reproducible or not have been reported in several sensitive applications. Numerical reproducibility failure mainly comes from the finite precision of computer arithmetic. Results floating-point computation depends on arithmetic and order operations. Massive parallel HPC which merges, for instance, many-core CPU GPU, clearly modifies these two parameters even run to a given computing platform. How trust such computed results? This paper presents how...
Summary As developers of libraries implementing interval arithmetic, we faced the same difficulties when it comes to testing our libraries. What must be tested? How can devise relevant test cases for unit testing? ensure a high (and possibly 100%) coverage? Before considering these questions, briefly recall main features arithmetic and IEEE 1788‐2015 standard arithmetic. After listing different aspects that, in opinion, tested, contribute first step towards offering suite an library. First...
In this paper, we describe implementations of interval matrix multiplication and verified solution to a linear system, using entirely BLAS routines, which are fully optimized parallelized.