ADELIA is the first AD-enabled INLA system that computes exact hyperparameter gradients via a structure-exploiting multi-GPU backward pass, delivering 4.2-7.9x per-gradient speedups and 5-8x better energy efficiency than finite differences on models with up to 1.9 million latent variables.
SIAM Journal on Matrix Anal
7 Pith papers cite this work. Polarity classification is still indexing.
7
Pith papers citing it
citation-role summary
background 2
method 2
citation-polarity summary
representative citing papers
NSPOD is a multigrid-like preconditioner using DeepONet-learned POD subspaces that dramatically cuts Krylov solver iterations for solid mechanics PDEs on unstructured CAD geometries, outperforming algebraic multigrid.
Approximate one-time preconditioning in face for MPGP algorithms yields error bounds, a sharp condition-number estimate, and large observed speedups on quadratic programs with constraints.
Strong enforcement of symmetry in H(div)-conforming finite elements yields material-robust stress approximations independent of the constitutive law, whereas weak enforcement produces arbitrarily poor results even in zero-stress cases with anisotropic laws.
Polfed.jl provides an efficient implementation of polynomially filtered Lanczos diagonalization for mid-spectrum eigenpairs in quantum many-body systems, supporting larger sizes via on-the-fly polynomial transformations and GPU acceleration.
A new solver combines MGRIT parallel-in-time stepping, sparse-grid combination technique, and space-filling-curve domain decomposition to produce an embarrassingly parallel method for parabolic PDEs up to six dimensions.
Review chapter summarizing advances in parallel sparse direct solvers along communication reduction and data-sparse compression axes.
citing papers explorer
-
Parallel Sparse and Data-Sparse Factorization-based Linear Solvers
Review chapter summarizing advances in parallel sparse direct solvers along communication reduction and data-sparse compression axes.