Python library for Rational Reduced Order Modeling
Bumped version to 1.8
Improved stability of dot and solve operations for tensorized sampling.
Replaced allowRepeatedSamples with uniqueness check for sample points.
Removed batch sampling from examples.
Added tensorized sample management (dot and solve). Removed batch sampling.
Added nearest neighbor. Fixed instability management in rational greedy.
Improved greedy a posteriori estimation, removed broken estimators, added EIM…
Improved hfengine blocks assembly. Added recompression of homogeneized RHS.
Updated examples and tests to symbolic affine dependence. Added first 2d greedy…
Reorganized hfengines using symbolic expressions. Modified MOR methods…
Added symbolic expression support. Moved degree and derivative support to…
Fixed discrepancy-based residual estimator for all polynomial bases.
Added support for discrepancy-based residual estimator. Only works for monomial…
Unified sample generation template. Modified engines, tests, and samples…
Module for the solution and rational model order reduction of parametric PDE-based problem. Coded in Python 3.6.
- numpy and scipy;
- fenics and mshr;
- and other standard Python3 modules (os, typing, time, datetime, abc, pickle, traceback, and itertools).
Most of the high fidelity problem engines already provided rely on FEniCS. If you do not have FEniCS installed, you may want to create an Anaconda3/Miniconda3 environment using the provided conda-fenics.yml environment file by running the command
conda env create --file conda-fenics.yml
This will create an environment where Fenics (and all other required modules) can be used. In order to use FEniCS, the environment must be activated through
source activate fenicsenv
Clone the repository
git clone https://c4science.ch/source/RROMPy.git
enter the main folder and install the package by typing
python3 setup.py install
The installation can be tested with
python3 setup.py test
This project is licensed under the GNU GENERAL PUBLIC LICENSE license - see the LICENSE file for details.
Part of the funding that made this module possible has been provided by the Swiss National Science Foundation through the FNS Research Project No. 182236.