This repository contains the accompanying code for the thesis.

It contains the neural network code for the thesis, as well as the scripts to run the experiments on an HPC cluster using SLURM. The original problem comes from the field of material sciences [1], the dataset was enhanced with artificial data in [2].

Code to generate the plots from data is also included.

[1] Matthias Zscheyge, Robert Böhm, Andreas Hornig, Johannes Gerritzen, and Maik Gude. Rate-dependent non-linear mechanical behaviour of continuous fibre-reinforced thermoplastic composites – experimental characterisation and viscoelastic-plastic damage modelling. Materials & Design, 193:108827, 2020

[2] Peter Winkler, Norman Koch, Andreas Hornig, and Johannes Gerritzen. Omniopt – a tool for hyperparameter optimization on HPC. In Heike Jagode, Hartwig Anzt, Hatem Ltaief, and Piotr Luszczek, editors, High Performance Computing, pages 285–296, Cham, 2021. Springer International Publishing.