GridapROMs.jl
GridapROMs provides tools for the reduced order modelling (ROM) of parameterised partial differential equations (PDEs). It is built on top of Gridap: parameterised problems are written as in Gridap, with the parameter μ as an extra argument, and reduced with a few additional lines of code.
The library supports
- steady and transient, linear and nonlinear, single- and multi-field problems
- reduced bases built with POD or TT-SVD, in any norm, with supremizer enrichment
- hyper-reduction with DEIM, S-OPT and radial basis function interpolation
- global and local (clustered) reduced operators
- distributed-memory parallelism of the whole offline and online phases
Installation
pkg> add GridapROMsTutorials
| Tutorial 1: Parameterised PDEs | parameter spaces, parameterised data and FE spaces, steady and transient FE problems |
| Tutorial 2: Reduced order models | offline/online phases, POD, TT-SVD, hyper-reduction, reduced operators, multi-field, nonlinear and transient problems |
| Tutorial 3: Local reduced operators | clustering of the parameter space and local reduced operators |
| Tutorial 4: Distributed-memory ROMs | running the ROM pipeline in parallel with MPI |
Reference
The library is organised in modules. The most common functions and types are exported by GridapROMs; the rest are accessible from the respective module.
| Module | Content |
|---|---|
Utils | integration domains, linear/nonlinear bundles, performance trackers, errors |
DofMaps | DOF orderings of FE spaces and sparse matrices |
TProduct | tensor-product geometries and FE spaces |
ParamDataStructures | parameter spaces, realisations, parameterised arrays and snapshots |
ParamAlgebra | parameterised algebraic operators and solvers |
ParamGeometry | parameterised geometries |
ParamFESpaces | parameterised FE spaces and functions |
ParamSteady | steady parameterised FE operators |
ParamODEs | transient parameterised FE operators and time-marching schemes |
Extensions | extension of solutions from active to background domains |
RBSteady | reduction, hyper-reduction and reduced operators for steady problems |
RBTransient | their space-time extension for transient problems |
Distributed | distributed-memory parallelism |
How to cite
@article{MUELLER2026109985,
title = {GridapROMs.jl: Efficient reduced order modelling in the Julia programming language},
journal = {Computer Physics Communications},
volume = {320},
pages = {109985},
year = {2026},
doi = {https://doi.org/10.1016/j.cpc.2025.109985},
author = {Nicholas Mueller and Santiago Badia}
}