PROGRAMME

Saint-Girons V will be held on June 29 – July 3, 2026. There will be a welcome reception on the evening of Monday, 29th June, and the meeting will finish at lunchtime on Friday, 3rd July.  Saint-Girons V will expand the Saint-Girons Conference series, by focusing on emerging topics in HPC. Topics to be discussed include:

    • numerical linear and multi-linear algebra
    • numerical optimization
    • numerical solution of PDEs
    • AI for scientific computing and HPC for AI
    • quantum computing
    • approximate computing involving mixed precision
    • low precision computation
    • randomized algorithms
    • low rank approximation
Monday

09:00 – 16:00     Workshop on Quantum Computing organized jointly with Fondation N7-Dev in Toulouse

      20:00             Welcome buffet at the Le Carré de l’Ange in Saint-Lizier

Tuesday

Main themes: Quantum Computing and Data Science, Artificial Intelligence and Data Science.

09:00 – 09:45     Marc Baboulin (INRIA and Université Paris-Saclay), invited speaker. Toward hybrid quantum-classical linear algebra solvers.

09:45 – 10:15     Simon Martiel (IBM Quantum), invited speaker. Quantum advantage with confidence.

10:15 – 10:45     — coffee break —

10:45 – 11:15     Julien Zylberman (CERFACS). Quantum algorithm for anisotropic diffusion and convection equations with vector norm scaling.

11:15 – 11:30     Youssef Achari Berrada (FAU and CERFACS). Enabling the solution of PDEs using Quantum Neural Networks and novel Trainable Frequency Feature Maps.

11:30 – 11:45     Siwar Badreddine (CERFACS). An Optimal Framework for Constructing Lie-Algebra Generator Pools: Application to Variational Quantum Eigensolvers for Chemistry.

11:45 – 12:00     Jean-Philippe Valois (Université de Lille). Scalable Parallel Branch-and-Bound for Optimal Qubit Mapping.

12:00 – 13:30     — lunch —

13:30 – 14:15     Rick Stevens (Argonne National Laboratory, University of Chicago), keynote speaker. Agentic Science and the DOE Genesis Mission.

14:15 – 14:45     Alena Kopanicakova (Toulouse-INP), invited speaker. Training of Deep Neural Networks Using Multilevel and Domain-Decomposition Methods.

14:45 – 15:15     Sherry Li (Lawrence Berkeley National Laboratory). Domain-aware Gaussian Process for Quantitative Learning in Scientific Computing.

15:15 – 15:45     — coffee break —

15:45 – 16:00     Florin Dobrian (Conviva). Commoditizing Combinatorial Code through Large Language Models: A Maximum Cardinality Matching Case Study.

16:00 – 16:15     Hans Johansen (Lawrence Berkeley National Laboratory). Adaptive Mesh Higher-Order Embedded Boundary Algorithms.

16:15 – 16:30     Takanori Kotama (Nagoya University). HPC-AutoResearch: An Agentic Research System for HPC Through Split-Phase Execution.

16:30 – 16:45     Sixin Zhang (IRIT). Supervised and unsupervised learning for Data assimilation.

17:00 – 18:00     Poster blitz (two-minute presentation per poster)

18:00 – 19:00     Formal poster session with all posters manned for last hour of schedule. Posters will also be kept up all week, and accessible during breaks.

Wednesday

Main themes: HPC and Numerical Linear Algebra.

09:00 – 09:45     Jack Dongarra (Innovative Computing Laboratory), keynote speaker. HPC in Transition.

09:45 – 10:00     Julien Langou (University of Colorado Denver). Variations on an algorithm to compute the Q factor after a Householder QR factorization.

10:00 – 10:15     Joost Rommes (NVIDIA). Accelerated linear algebra on AI-era GPUs.

10:15 – 10:30     Alexis Montoison (AMD). Krylov.jl: Empowering Performance and Portability in High-Performance Computing.

10:30 – 10:45     Miroslav Tuma (Charles University). Avoiding the Normal Equations: Row-Splitting ILU Preconditioners for Sparse Least Squares.

10:45 – 11:15     — coffee break —

11:15 – 12:00     Satoshi Matsuoka (Riken Center for Computational Science), keynote speaker. FP8 is All You Need: Debunking Hardware FP64 as the HPC Holy Grail, A Tensor-Memory Equilibrium Model and Implementation Strategy for Ozaki Scheme II on Memory-Bound Workloads in the Post-FP64 Era.

12:00 – 12:15     Michaël Krajecki (GENCI). Presentation to be announced.

12:15 – 12:30     Alfredo Buttari (CNRS-IRIT). The NumPEx French PEPR Research Program.

12:30 – 12:45     Biswajeet Rath (IRIT-INP Toulouse). Analyzing the use of accelerators and mixed precision algorithms on the performance of sparse direct solvers.

12:45 – 13:00     Youssef Aly (Siemens Industry Software Egypt). Design and Implementation of a Reduced and Mixed Precision AMG Solver on GPUs for Large Industrial Circuit Verification Tasks.

13:00 – 13:15     Marc Marot-Lassauzaie (TUM). A mixed- and variable-precision implementation of Discontinuous Galerkin solvers.

13:14 – 14:45     — lunch —

Touristic and other activities in the afternoon followed by the gala dinner at the Le Carré de l’Ange in Saint-Lizier

Thursday

Main Themes: Numerical Linear Algebra and Optimization.

09:00 – 09:45     Per-Gunnar Martinsson (The University of Texas at Austin, ILAS Keynote Speaker). Randomized Linear Algebra at Scale: From Sketches to Operators.

09:45 – 10:00     Andy Wathen (Rutherford Appleton Laboratory). Self-adjointness and complex eigenvalues of a real nonsymmetric matrix.

10:00 – 10:15     Philip Knight (University of Strathclyde). A Framework for Fast Balancing.

10:15 – 10:30     David Goudin (Bull). Grand Cru Project: Bull’s strategy to develop expertise and innovation for the synthetic benchmarks.

10:30 – 10:45     Sri Hari Krishna Narayanan (Argonne National Laboratory). diffblas and difflapack: derivatives of BLAS and LAPACK routines.

10:45 – 11:00     Jean-Claude Yakoubsohn (Institut de Mathématiques de Toulouse). Unified Newton’s Method For Matrix Decompositions.

11:00 – 11:30     — coffee break —

11:30 – 12:00     Mike Saunders (Stanford University). MinAres: An Iterative Solver for Symmetric Linear Systems.

12:00 – 12:15     Petr Tichy (Charles University). On Dubrulle’s Block Conjugate Gradient Method.

12:15 – 12:30     Gérard Meurant (retired). Block conjugate gradient methods for least squares problems.

12:30 – 12:45     Bastien Vieublé (Academy of Mathematics and Systems Science). A comparison of the convergence rates of iterative refinement and GMRES.

12:45 – 13:00     Stephen Thomas (Lehigh University). Spectral Measures, Christoffel Functions, and Edge Instability in Krylov Methods.

13:00 – 13:15     Oussama Mouhtal (CERFACS, Montréal Polytechnique). A Spectral Preconditioner for the Conjugate Gradient Method with Iteration Budget.

13:15 – 14:30     — lunch —

14:30 – 15:15     Eulogies to Cleve Moler

15:15 – 15:30     Alex Pothen (Purdue University). A Half Century of Algorithm Design: The Matching Story.

15:30 – 15:45     Bob Lucas (Synopsys). Porting multifrontal computations to GPUs using the OpenMP API.

15:45 – 16:00     Esmond Ng (Lawrence Berkeley National Laboratory). On Solving General Sparse Symmetric Linear Systems.

16:00 – 16:15     Edgar Solomonik (University of Illinois Urbana-Champaign). Sparse LDL, LU, and Inverse Butterfly Factorization via Tree Decomposition.

16:15 – 16:45     — coffee break —

16:45 – 17:30     Sven Leyffer (Argonne National Laboratory), keynote speaker. Unifying nonlinearly constrained optimization.

17:30 – 18:00     Hussam Al Daas (STFC, Rutherford Appleton Laboratory), invited speaker. Extended-Krylov-subspace methods for trust-region and norm-regularization subproblems.

18:00 – 18:15     Jennifer Scott (University of Reading and STFC Rutherford Appleton Laboratory). How Low Can You Go? Mixed-Precision Incomplete Factorization Preconditioners for Sparse Least-Squares Problems.

18:15 – 18:30     Thomas Bake (Charles University). Forward and backward error bounds for a mixed precision preconditioned conjugate gradient algorithm.

18:30 – 18:45     Joanna Bisch (INRIA). Bridging the gap between Adaptive Cross Approximation (ACA) and Rank-Revealing LU: a path for BLAS3 ACA.

18:45 – 19:00     Antoine Jego (Sorbonne Université). Memory Accessor for Mixed Precision GMRES-Based Iterative Refinement.

Friday

Main themes: PDEs, Multigrid, Applications.

09:00 – 09:30     Françoise Tisseur (University of Manchester), invited speaker. Computing Accurate Eigenvalues and Singular Values Using Mixed-Precision Algorithms.

09:30 – 09:45     Brieuc Nicolas (INRIA). Series of tensor contractions with an STF runtime.

09:45 – 10:00     Alycia Lisito (Bull). Exploiting recursive tasks to submit data dependent graph for the LU factorization with partial pivoting.

10:00 – 10:15     Matthieu Robeyns (IRIT). A parallel approach for solving the Green’s function based on the nested-dissection and selected-inverse methods.

10:15 – 10:30     Aydin Buluc (NVIDIA). GPU-Accelerated Direct and Iterative Sparse Linear Solvers.

10:30 – 11:00     — coffee break —

11.00 – 11:15     Petr Vacek (IRIT). Multigrid methods with approximate coarsest-level solvers.

11:15 – 11:30     Jun Cao (Toronto Metropolitan University). An Immersed-Boundary-Based Framework for Study of Tornado Dynamics.

11:30 – 11:45     Sophie Moufawad (American University of Beirut). Direct and Inverse Problem for Gas Diffusion in Polar Firn.

11:45 – 12:00     Suraj Kumar (Inria, Lyon). Communication Lower Bounds and Algorithms for Sketching with Random Dense Matrices.

12:00 – 12:30     Serge Petiton (Paris), invited speaker. Distributed and Parallel Sparse Matrix Computation in the Age of AI.

12:30 – 14:00     — lunch —

Post conference activities in the afternoon. People who can stay later can benefit from the renowned hospitality of the people of Saint-Girons as has happened in earlier meetings. (This would be later on Friday and into the weekend.)

POSTERS

  • Nathan Aubert (CERFACS), Quantum algorithm for nonlinear iterative schemes
  • Tom Caruso (INRIA), Mixed precision domain decomposition preconditioner
  • Sai Aakash Dasari (CERFACS), Extrapolated multigrid solver for elliptic PDEs on non-polygonal domains
  • Hugo Dodelin (INRIA), HPC Matrix Multiplication with a MapReduce Model
  • Ouassel El Habti (Sorbonne Université, CNRS, LIP6 & EDF), Compressing sparse direct solvers with block low-rank matrices with shared bases
  • Takeshi Fukaya (Hokkaido University), Performance Evaluation of CholeskyQR-Type Algorithms for Tall-Skinny QR Factorization
  • Julien Gaupp (INRIA), STF++ : Sequential Task Flow with a C++ interface and its application to dense linear algebra
  • Shun-ichiro Hayashi (Nagoya University), VIBECODEHPC: HPC Auto-Tuning with CLI-based Multi-agents and Local LLM Support
  • Guillaume Helbecque (Université de Lille), PGAS-based parallel branch-and-bound for ultra-scale supercomputers
  • Karmijn Hoogveld (CNRS), RandQRCP-ID: a QR-based randomized interpolative decomposition method for fixed-accuracy low-rank approximations
  • Davit Khalatyan (American University of Armenia), Implementation of Quantum Atomic Logic Gate in the Cesium D2 Line
  • Suraj Kumar (INRIA), Minimizing Communication for Parallel Symmetric Matricized-Tensor Times Khatri-Rao Product
  • Alexandre Malhéné (INRIA), When Should GMRES Stop? Practical Stopping Criteria for GMRES
  • Samuel Mendoza (INRIA), Toward a high-level of expressiveness for sparse direct methods
  • Clément Peaucelle (Airbus & INRIA), Introducing Floating-Point Compression In An Industrial, Task-Based Hierarchical Solver
  • Gabriel Berk Pereira (University of Oxford), Krylov Subspace Acceleration for First-Order Splitting Methods in Convex Quadratic Programming
  • Jordan Ramassamy (Toulouse INP – IRIT), Auto-differentiable ensemble Kalman filters with adaptive inflation
  • Rony Ramia (CERFACS), Quantum Measurement Protocols for Scientific Computing
  • Ehouarn Simon (Toulouse INP – IRIT), A kernel extension of the ETKF and a localization strategy
  • Alexandre Tabouret (Sorbonne Université, CNRS, LIP6), Adaptive Precision Restarted GMRES for Solving Sparse Linear Systems
  • Zoran Tomljanovic (School of Applied Mathematics and Informatics, University of Osijek), A Scalable Framework for Simultaneous Optimization of Damper Placement and Viscosities
  • Dimitri Walther (CEA CESTA, Univ. Bordeaux, CNRS, Bordeaux INP, INRIA, LaBRI, UMR 5800), Leveraging Graph Partitioning to Accelerate H-matrix Solvers
  • Sijia Yu (Free university of Bolzano-Bozen), Revealing Hidden Structure in Total Least Squares via Structural Decomposition

 


 

Downloadable programme (oral presentations and posters) and list of participants.

Retour en haut