Skip to content

Commit d045a8c

Browse files
authored
Update README.md
1 parent e018357 commit d045a8c

File tree

1 file changed

+3
-3
lines changed

1 file changed

+3
-3
lines changed

README.md

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -18,11 +18,11 @@ Each example consists of three main procedures: preprocessing, finite element si
1818
- Parametrize the geometry and compute reduced parameters of geometry
1919

2020
* We provide three .edp here:
21-
- `ReferNS_transient_parallel.edp` is designed for performing finite-element simulations (transient) with various shapes of domains on a reference domain (paralleled with PETSc). This step generates the geometry-informed snapshots for ROM.
21+
- `ReferNS_steady_parallel.edp` is designed for performing finite-element simulations (steady) with various shapes of domains on a reference domain (paralleled with PETSc). This step generates the geometry-informed snapshots for ROM (recommend to use).
2222
Implement with `ff-mpirun -np 16 ReferNS_transient_parallel.edp -v 0` where 16 stands for the number of cores for mpi.
23+
- `ReferNS_transient_parallel.edp` is the transient version. It takes much more time to generate a steady solution, but can be used for time-dependent problems.
2324
- `StandardNS.edp` is the standard finite-element simulation for the flow on its original domain. It is used for validation of the solution achieved on the reference domain
24-
- `ReferNS.edp` is designed for performing finite-element simulations (steady-state) with various shapes of domains on a reference domain. This step generates the geometry-informed snapshots for ROM (currently out-of-date, not using).
2525

2626
* `ROM.py` collects the snapshots data and constructs the ROM based on POD+RBF interpolation. The script also shows the prediction error on the validation and test dataset. The visualization results are saved in `data/error` directory in the form of .vtk.
2727

28-
Note that surface registration and finite element are computationally expensive. It might take a long time
28+
Note that surface registration and finite elements are computationally expensive. It might take a long time

0 commit comments

Comments
 (0)