MOOSE 5.0 Brings GPU Acceleration, New Discretizations and AI Tools to Multiphysics Simulation - Bioengineer.org
The Multiphysics Object-Oriented Simulation Environment (MOOSE) framework has released version 5.0, introducing significant GPU acceleration capabilities, new discretization techniques, and AI-powered tools for multiphysics simulations.
The framework, which allows researchers to build tightly coupled simulations of interacting physical processes, has been continuously and collaboratively developed. This latest version reports on features implemented since the previous update in mid-2025, with the headline change being a dramatic widening of GPU support.
Dedicated Kokkos systems now accelerate the native libMesh finite-element backend, while the MFEM backend has been expanded. Both new capabilities remain compatible with existing CPU workflows, ensuring smooth adoption for current users.
GPU Acceleration Advances
The Kokkos-MOOSE framework achieves residual object polymorphism statically through a C++ template pattern rather than traditional virtual-keyword-based dynamic polymorphism. This approach maintains familiar user interfaces while enabling GPU acceleration.
Beyond residual objects, developers have accelerated materials and traditional postprocessing systems, including postprocessors, vector postprocessors and auxiliary kernels. Initial Kokkos support has also been merged for MOOSE's linear finite-volume system.
this new compilation pathway improves GPU performance by up to 4.5 times compared with the previous TorchScript/JIT implementation
Discretization Techniques
MOOSE 5.0 introduces new discretization methods, including interior penalty HDG (IP-HDG) discretizations alongside the existing local hybridized discontinuous Galerkin (L-HDG) methods. The IP-HDG capability was leveraged to build a field-split preconditioner for solving high-Reynolds-number Navier-Stokes flow problems.
The framework also includes a new Shifted Boundary Method module for solving partial differential equations on complex geometries without requiring boundary-fitted meshes, preserving accuracy while avoiding difficult meshing processes.
AI Integration
Artificial intelligence features make their debut in this release through MOOSEnger, an agentic tool developed by Idaho National Laboratory that helps users create simulation inputs, address issues and process outputs.
MOOSEnger leverages deterministic input checks, retrieval-augmented generation, integration of the MOOSE language server and version-controlled artifact management. Designed as an extendable framework, it can be adapted for domain-specific applications.
Recent developments in meshing capabilities, 3D mortar contact methods and the Kokkos linear finite-volume discretization have also heavily leveraged AI-assisted development, which was deployed to create patches for user support as part of routine maintenance.
Related
