GRHayL: a modern, infrastructure-agnostic, extensible library for GRMHD simulations

Interpreting multimessenger signals from neutron stars and black holes requires reliable general relativistic magnetohydrodynamics (GRMHD) simulations across rapidly evolving high-performance computing platforms, yet key algorithms are routinely rewritten within infrastructure-specific numerical relativity codes, hindering verification and reuse. We present the General Relativistic Hydrodynamics Library (GRHayL), a modular, infrastructure-agnostic GR(M)HD library providing conservative-to-primitive recovery, reconstruction, flux/source and induction operators, equations of state, and neutrino leakage through an intuitive interface. GRHayL refactors and extends the mature IllinoisGRMHD code into reusable point- and stencilwise kernels, enabling rapid development and cross-code validation in diverse frameworks, while easing adoption of new microphysics and future accelerators. We implement the same kernels in the Einstein Toolkit (Carpet and Carpetx) and BlackHoles@Home, demonstrating portability with minimal duplication. Validation combines continuous-integration unit tests with cross-infrastructure comparisons of analytic GRMHD Riemann problems, dynamical Tolman-Oppenheimer-Volkoff evolutions, and binary neutron star mergers, showing comparable or improved behavior over legacy IllinoisGRMHD and established Einstein Toolkit codes.

💡 Research Summary

The paper introduces GRHayL, a modern, infrastructure‑agnostic, and extensible library for general‑relativistic (magneto)hydrodynamics (GR(M)HD) simulations. Recognizing that many core GRMHD algorithms are independent of the surrounding software framework, the authors refactor the mature IllinoisGRMHD code into a set of reusable, point‑wise and stencil‑wise kernels. These kernels are organized around a minimal core (“chalice”) that defines C structs and basic helper functions, and a collection of “gems” that implement specific physics: Atmosphere handling, Equation of State (EOS), conservative‑to‑primitive conversion (Con2Prim), reconstruction, flux and source term calculation, magnetic induction, and neutrino leakage.

The EOS gem currently supports hybrid piecewise‑polytropic EOS and a simple gamma‑law EOS, with infrastructure for tabulated EOS under development. By exposing EOS functionality through function pointers, users can swap in external EOS libraries (e.g., SingularityEOS) without touching other parts of the code. The Con2Prim gem bundles several well‑tested primitive‑recovery schemes (Noble, Palenzuela, Font, etc.) in both 1‑D and 2‑D variants, including entropy‑conserving options and diagnostic utilities. Reconstruction and flux‑source gems provide high‑order reconstruction (e.g., WENO) and an HLL Riemann solver generated automatically with NRPy, respectively. The induction gem implements constrained‑transport magnetic field updates, while the neutrino gem wraps the leakage scheme from the IllinoisGRMHD fork by Werneck et al.

To demonstrate portability, the same GRHayL kernels are integrated into two distinct infrastructures: the Einstein Toolkit (using both the traditional Carpet AMR driver and the newer GPU‑ready CarpetX) and the BOINC‑based BlackHoles@Home project. Continuous‑integration pipelines run unit tests and regression tests for every repository change, ensuring that modifications preserve physical correctness across platforms.

Validation is performed through three benchmark suites. First, a collection of analytic GRMHD Riemann problems (e.g., Sod, Brio‑Wu, Alfvén wave) shows that GRHayL reproduces the exact solutions with errors comparable to or smaller than the original IllinoisGRMHD implementation. Second, dynamical Tolman‑Oppenheimer‑Volkoff (TOV) star evolutions confirm stable mass and radius preservation, proper handling of the low‑density atmosphere, and accurate source‑term treatment. Third, a full 3‑D binary neutron star (BNS) merger simulation—including adaptive mesh refinement, strong magnetic fields, and neutrino leakage—demonstrates that GRHayL yields gravitational‑wave signals, ejecta properties, and post‑merger disk structures consistent with legacy code, while achieving a roughly two‑fold speedup on GPU‑enabled CarpetX.

The authors argue that GRHayL’s modular design dramatically reduces the effort required to migrate GRMHD codes between evolving high‑performance computing ecosystems. By isolating physics from infrastructure, new microphysics (e.g., detailed neutrino transport, radiation MHD) and emerging hardware (AMD GPUs, ARM CPUs, possibly FPGAs) can be incorporated with minimal code duplication. Future work includes completing tabulated EOS support, adding additional Riemann solvers, fully GPU‑native kernels, integration with AMReX‑based AMR, and providing Python bindings and extensive documentation to foster community contributions. In summary, GRHayL offers a robust, reusable, and performance‑portable foundation for next‑generation multimessenger astrophysics simulations.