Topic 2 Subtopic a Title: Accelerated Deployment of MFEM Based Solvers in Large Scale Industrial Problems

While many Advanced Scientific Computing Research (ASCR) supported software packages are open source, they are often complicated to use, distributed primarily in source-code form targeting HPC systems, and potential adopters lack options for purchasing commercial support, training, and custom- development services. In response to this need, ReLogic Research, Inc., in collaboration with Lawrence Livermore National Laboratory (LLNL), will develop a secure, cloud deployable platform based on the MFEM software termed Minerva. Minerva will feature an integration layer allowing users of commercially available finite element (FE) pre/post-processing software, e.g., Abaqus/CAE, Hypermesh, Femap, etc., for use with the Abaqus solver to run simulation studies with the MFEM discretization library and will strengthen further MFEM implemented solvers to make them applicable for solving large scale industrial design and optimization problems. During Phase I, the Team will 1) develop a software layer allowing access from MFEM to Abaqus input data files including mesh, material, and simulation parameters, and will 2) demonstrate the feasibility of the approach by allowing the use of solvers in MFEM. The software layer will allow users to transfer already developed CAD and FEM models to MFEM seamlessly and run scalable detailed FEM studies and optimizations, combining the strengths of the two suites in a single production environment. The goal of Minerva is to accelerate high-performance scalable finite element discretization research and application development for a wide variety of computational environments, ranging from laptops to supercomputers. It is anticipated that through collaboration with academia and industry throughout the program, these organizations will be early adopters of this platform and will support a customer ecosystem focused on accelerating enhancements to the platform and development of standalone applications to be deployed outside of the platform. By providing a cloud-based platform that significantly reduces the barriers to entry to using and developing MFEM applications, this research will benefit ASCR by contributing fixes for defects in the software, changes needed to make the software function on generally- available platforms, and other non-proprietary enhancements of general utility to the software back to the MFEM project.