The curvature-aware precision controller adapts between FP32 and FP64 during PINN training to match double-precision accuracy at reduced computational cost.
Nocedal ,\ 10.1090/S0025-5718-1980-0572855-7 journal journal Mathematics of Computation \ volume 35 ,\ pages 773 ( year 1980 ) NoStop
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
citation-role summary
method 1
citation-polarity summary
roles
method 1polarities
use method 1representative citing papers
PyMieDiff provides a new open-source PyTorch library for fully differentiable Mie scattering from layered spheres, with autograd support for efficiencies, angular patterns, and near-fields.
A proximal limited-memory quasi-Newton scheme is developed for nonsmooth nonconvex optimization, with global convergence proven under mild assumptions and rates under the Kurdyka-Lojasiewicz property.
An open-source Snakemake workflow fully automates NEB reaction path calculations with ML potentials and recovers the known HCN-HNC energy profile without manual steps.
citing papers explorer
-
Curvature-aware dynamic precision approach for physics-informed neural networks
The curvature-aware precision controller adapts between FP32 and FP64 during PINN training to match double-precision accuracy at reduced computational cost.
-
PyMieDiff: A differentiable Mie scattering library
PyMieDiff provides a new open-source PyTorch library for fully differentiable Mie scattering from layered spheres, with autograd support for efficiencies, angular patterns, and near-fields.
-
Proximal Limited-Memory Quasi-Newton Methods for Nonsmooth Nonconvex Optimization
A proximal limited-memory quasi-Newton scheme is developed for nonsmooth nonconvex optimization, with global convergence proven under mild assumptions and rates under the Kurdyka-Lojasiewicz property.
-
Reproducible Orchestration of Best Practices for Reaction Path Optimization with the Nudged Elastic Band
An open-source Snakemake workflow fully automates NEB reaction path calculations with ML potentials and recovers the known HCN-HNC energy profile without manual steps.