Workshop on Approximate Computing in Numerical Linear Algebra

›8:00 (30min)

Welcome coffee

8:00 - 8:30 (30min)

Welcome coffee

›8:30 (1h40)

Session 5
Chair: Julien Langou

8:30 - 10:10 (1h40)

Session 5

Chair: Julien Langou

› Fast randomized algorithms for rank structured matrices - Per-Gunnar Martinsson, University of Texas at Austin 08:30-08:50 (20min)

› A novel fixed-accuracy variant of the Randomized Interpolative Decomposition algorithm - Karmijn Hoogveld, IRIT 08:50-09:10 (20min)

› Sketching strategies as NLA companion - Valeria Simoncini, Università di Bologna 09:10-09:30 (20min)

› Adaptive Sketching Based Construction of H2 Matrices on GPUs - Xiaoye Sherry Li, LBNL 09:30-09:50 (20min)

› Accelerating an ensemble of variational data assimilations with randomized preconditioning - Ieva Daužickaitė, CERFACS 09:50-10:10 (20min)

›10:10 (40min)

Coffee break

10:10 - 10:50 (40min)

Coffee break

›10:50 (1h20)

Session 6
Chair: Rio Yokota

10:50 - 12:10 (1h20)

Session 6

Chair: Rio Yokota

› Mixed precision algorithms with neural operator preconditioning for solving some parametric PDEs - Yanfei Xiang, IRMA 10:50-11:10 (20min)

› A Matrix Factorization Framework for Community Detection under the Degree-Corrected Block Model - Alexandra Dache, Université de Mons 11:10-11:30 (20min)

› Integrative Co-Embedding of Multi-View Data Sets - Haesun Park, Georgia Institute of Technology 11:30-11:50 (20min)

› Minimum-norm solution of the non-symmetric semidefinite Procrustes problem - Stefano Sicilia, Université de Mons 11:50-12:10 (20min)

›12:10 (2h20)

Lunch + poster session

12:10 - 14:30 (2h20)

Lunch + poster session

› Orthogonalization Avoidant Davidson - Ioannis Thanasis, Charles University 12:10-14:30 (2h20)

› Mixed precision iterative refinement for least squares with linear equality constraints and generalized least squares problems - Bowen Gao, Fudan university 12:10-14:30 (2h20)

› Mixed-Precision and Energy-Efficient Computations - Yanxiang Chen, Umeå University 12:10-14:30 (2h20)

› PROMISE: Dynamic mixed precision analysis for numerical algorithms - Xinye Chen, LIP6 12:10-14:30 (2h20)

› Three examples of reduced precision with sparsity in high-fidelity simulations - Hans Johansen, LBNL 12:10-14:30 (2h20)

› Do you precondition on the left or on the right ? - Nicole Spillane, CMAP 12:10-14:30 (2h20)

›14:30 (1h40)

Session 7

14:30 - 16:10 (1h40)

Session 7

› Software emulation of 3Sum, the FMA and other operations. - Jean-Michel Muller, LIP 14:30-14:50 (20min)

› Accuracy of Mathematical Functions in Julia - Mantas Mikaitis, University of Leeds 14:50-15:10 (20min)

› Fast and accurate sub-cubic matrix multiplication - Clément Pernet, Grenoble INP - UGA, LJK 15:10-15:30 (20min)

› Probabilistic Error Analysis of Limited-Precision Stochastic Rounding - El-Mehdi El Arar, IRISA 15:30-15:50 (20min)

› Mixed-Precision Strategies in Spectral Element Codes: Accuracy-Aware Approximation for Energy-Efficient Simulations - Roman Iakymchuk, Umeå University 15:50-16:10 (20min)

›16:10 (40min)

Coffee break

16:10 - 16:50 (40min)

Coffee break

›16:50 (1h40)

Session 8
Chair: Patrick Amestoy

16:50 - 18:30 (1h40)

Session 8

Chair: Patrick Amestoy

› < T >LAPACK: a new linear algebra software package with mixed-precision capabilities - Julien Langou, University of Colorado Denver 16:50-17:10 (20min)

› A composable abstraction of hierarchical methods for matrix-vector product acceleration - Antoine Gicquel, LABRI 17:10-17:30 (20min)

› Mixed Precision Strategies for Solving Sparse Linear Systems with BiCGStab - Hugo Dorfsman, LIP6 - IFPEN 17:30-17:50 (20min)

› Solving Big Problems with Little Numbers - Hatem Ltaief, KAUST 17:50-18:10 (20min)

› Approximate solutions to (generalized) matrix chains - Paolo Bientinesi, Umeå University 18:10-18:30 (20min)