README.md

data

This directory contains simulation data and setup details for five mechanical models: OnceClampedBeam, TwiceClampedBeam, TwistingBeam, Liver and LeftVentricle.

The above datasets have been already been processed into a format suitable for running emulation, as described in the parent directory README.md.

For completeness, data processing code is provided here. This code is contained inside data_process_utils.py, which can be called using data_main.py as is detailed below. The requirement formats for the raw unprocessed data to allow the above scripts to run are detailed in DATA_FORMAT_REQUIREMENTS.md. We also provide an example of an unprocessed dataset generated using FEniCS in TwistingBeamUnprocessed.

Note that an "unprocessed" dataset will still require some pre-processing of the results from e.g. FEniCS - for an example of how to apply this pre-processing, see FenicsMeshProcessingLV.ipynb.

For more details on augmented graph graph generation (i.e. the use of virtual nodes/edges), see Section 2.3 of Dalton et. al (2022) and the associated GitHub repository.

Processing Data

All datasets were processes by calling data_main.py with default parameters. For instance, to process the data in TwistingBeamUnprocessed, run:

python -m data_main --data_dir="TwistingBeamUnprocessed"

Directory Layout

Files:

data_main.py

Main script for processing raw data into format suitable for PIGNN emulation

data_process_utils.py

Utility functions for processing raw data

DATA_FORMAT_REQUIREMENTS.md

Details of the required format for raw data

Subdirectories:

OnceClampedBeam and TwiceClampedBeam

Models used in Section 3.1 of the manuscript

TwistingBeam

Model with 343 nodes used in Section 3.2 of the manuscript

Liver

Model used in Section 3.3 of the manuscript

LeftVentricle

Slighly different model to that used in Section 3.4 of the manuscript - for details, see the README in parent directory