arXiv · 2512.11590
HPRMAT: A high-performance R-matrix solver with GPU acceleration for coupled-channel problems in nuclear physics
Abstract
I present HPRMAT, a self-contained, high-performance R-matrix solver framework for coupled-channel scattering calculations in nuclear physics. It provides the full R-matrix propagation machinery with the same user-supplied-potential interface as standard R-matrix packages, and is additionally a drop-in replacement for the linear algebra routines of Descouvemont's package. It employs direct linear equation solving with optimized libraries instead of traditional matrix inversion, achieving significant performance improvements. The package provides four solver backends: (1) double-precision LU factorization, (2) mixed-precision arithmetic with iterative refinement, (3) a Woodbury formula approach exploiting the kinetic-coupling matrix structure, and (4) GPU acceleration. Benchmark calculations demonstrate that the GPU solver achieves about 15x speedup over the optimized CPU direct solver, and 41x over the legacy inversion-based code, at N=25600. The mixed-precision strategy is particularly effective on consumer GPUs (e.g., NVIDIA RTX 3090/4090), where single-precision throughput exceeds double-precision by a factor of 64:1; by performing the factorization in single precision, with iterative refinement available to recover full double-precision accuracy where needed, HPRMAT overcomes the poor FP64 performance of consumer hardware while retaining the accuracy required for cross-section calculations. This makes large-scale continuum-discretized coupled-channels (CDCC) and coupled-channel calculations accessible to researchers using standard desktop workstations, without requiring expensive data-center GPUs.
Explore related subjects
Keep this discovery
Explore connections, maps & timelines
Jin Lei. 2025-12-12. HPRMAT: A high-performance R-matrix solver with GPU acceleration for coupled-channel problems in nuclear physics. https://doi.org/10.1016/j.cpc.2026.110379
Cite the original work for its findings. Save a collection to share your selection of sources.