Code for: Identifiability limits of grain inverse design measured with regression models on a public solid rocket motor archive
Year: 2026
Type: software
Abstract: Python scripts that reproduce every table, figure and reported number of the paper from the public archive "A Multi-Topology Solid Rocket Motor Grain Dataset for Inverse Internal Ballistic Design" (https://doi.org/10.5281/zenodo.18093210). The archive itself is not included and has to be downloaded separately, as described in veri/README.md. The deposit contains the scripts for reading the archive, resampling the burning surface area curves, the model free identifiability ratio, the regression models (conditioned multilayer perceptron, per topology network, ridge regression, random forest, train mean baseline), three sequence networks (residual convolutional network, patch Transformer, selective state space network), the mixture density network and the forward consistency check. It also holds the result files of the published run, so that every value can be checked without retraining. Trained network weights are not included. The scripts recreate them. Environment: Python 3.12, numpy 1.26.4, scipy 1.13.1, scikit-learn 1.9.0, PyTorch 2.6.0, matplotlib 3.9.2. The order of execution and the runtime of each step are given in README.md.
Institution: Bolu Abant İzzet Baysal University
Language: English
Subfield: Aerospace Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7222505810