This article was accepted into the corpus but its outbound wikilinks were never NER-processed — typical at the deepest BFS hop or when the run's entity cap was reached. No expansion funnel to show.
| Sandia's Trilinos | |
|---|---|
| Name | Trilinos |
| Developer | Sandia National Laboratories |
| Released | 2000s |
| Programming language | C++, Fortran, Python |
| Operating system | Linux, macOS, Microsoft Windows |
| License | Modified BSD |
Sandia's Trilinos is a collection of open-source software libraries for the solution of large-scale, complex multi-physics and multi-scale computational science and engineering problems. Developed at Sandia National Laboratories, Trilinos provides modular C++-based packages for linear algebra, nonlinear solvers, optimization, preconditioning, mesh infrastructure, discretizations, and data management that integrate with scientific projects at institutions such as Los Alamos National Laboratory, Lawrence Livermore National Laboratory, Argonne National Laboratory, and universities including Stanford University, Massachusetts Institute of Technology, University of California, Berkeley, University of Michigan.
Trilinos targets high-performance computing workflows used by teams at National Energy Research Scientific Computing Center, Oak Ridge National Laboratory, Argonne National Laboratory, Princeton University, and Caltech. The project supports solver stacks leveraging technologies from MPI, OpenMP, CUDA, HIP, and Kokkos to enable portability on systems such as Summit, Frontier, Perlmutter, Fugaku, and cloud platforms operated by Amazon Web Services, Microsoft Azure, and Google Cloud Platform. Trilinos integrates with meshing and visualization tools like ParaView, VisIt, Gmsh, and VTK.
Trilinos originated from initiatives at Sandia National Laboratories to consolidate solver research from groups including teams associated with DOE campaigns and programs such as the Advanced Simulation and Computing Program, Exascale Computing Project, and collaborations with National Science Foundation projects. Early development drew on methods from researchers at Rice University, University of California, Los Angeles, University of Texas at Austin, and industrial partners like IBM and Cray. Over time Trilinos incorporated algorithms influenced by work from mathematicians affiliated with Society for Industrial and Applied Mathematics, American Mathematical Society, and contributors from projects tied to awards such as the Turing Award winners and recipients of the SIAM Fellowship.
Trilinos uses a package-oriented architecture enabling composable stacks where packages interoperate through abstract interfaces. Core infrastructures include adapter layers for Epetra and Tpetra object models, memory spaces backed by Kokkos for device portability, and communication via MPI. Interfacing packages allow coupling to external projects such as PETSc, Trilinos-adjacent ecosystems like Amesos, AztecOO, and mesh frameworks including STK, Albany, and MOOSE. Build and packaging integrate with CMake, Spack, and container systems like Docker and Singularity.
Notable Trilinos packages provide numerical building blocks: Belos for Krylov methods, Ifpack2 and MueLu for preconditioning and multigrid, NOX for nonlinear solvers, Rythmos for time integration, Piro for stability and continuation, OptiPack and ROL for optimization, Tpetra for distributed linear algebra, Intrepid2 for finite element discretizations, Zoltan2 for load balancing and partitioning, Shylu for scalable solvers, and DataTransferKit for mesh-mesh coupling. Supporting utilities include Teuchos for parameter lists and memory management, TrilinosApps-style examples, and testing harnesses tied to CDash dashboards.
Trilinos underpins simulations in domains pursued at institutions like NASA, NOAA, US Geological Survey, and academic groups at Columbia University, University of Illinois Urbana-Champaign, Cornell University, and University of Texas at Austin. Use cases include computational fluid dynamics projects at Rolls-Royce, aerospace simulations for Boeing, structural mechanics in civil engineering collaborations with AECOM, subsurface flow modeling in energy research for ExxonMobil and Chevron, and multi-physics coupling in fusion experiments at Princeton Plasma Physics Laboratory. Trilinos also serves inverse problems and uncertainty quantification efforts tied to Lawrence Berkeley National Laboratory and machine learning pipelines at Google Research and IBM Research.
Trilinos emphasizes strong scaling and weak scaling on leadership-class systems including Oak Ridge Leadership Computing Facility, Argonne Leadership Computing Facility, and NERSC. Performance engineering leverages architecture-aware abstractions from Kokkos to target CPUs from Intel, AMD, GPUs from NVIDIA, and accelerators from Fujitsu. Benchmarks and studies published in venues like SC Conference, International Conference on Computational Science, and journals such as SIAM Journal on Scientific Computing demonstrate algorithmic performance for sparse linear systems, multigrid preconditioners, and parallel eigenvalue solvers.
Trilinos development is coordinated by teams at Sandia National Laboratories with contributions from national labs, universities, and industry collaborators including Los Alamos National Laboratory, Lawrence Livermore National Laboratory, Argonne National Laboratory, Oak Ridge National Laboratory, Microsoft Research, and Intel Labs. The project uses an open governance model with contribution processes integrating with GitHub, issue tracking, and continuous integration. Trilinos is distributed under a Modified BSD license compatible with collaborations involving entities such as DOE laboratories and commercial partners including Schlumberger, Siemens, and General Electric.
Category:Scientific simulation software