Accompanying dataset for the paper entitled: "The semi-explicit nonsmooth Newmark-β time integrator for robust unilateral contact in dynamic fragmentation simulations"

What this dataset contains This dataset contains the parametric study results, analysis notebooks, and reproducibility scripts for the paper "The semi-explicit nonsmooth Newmark-β time integrator for robust unilateral contact in dynamic fragmentation simulations", on an explicit dynamics framework suited to highly non-smooth interface behaviors. The simulations were performed with NS‑Frag1D, a 1D cohesive fracture and contact solver built on the finite-element library Akantu. The canonical code repository is available at https://github.com/tghesquie/nsn_frag1d (tag v1.0.0) and is archived on Software Heritage at swh:1:rev:d07bc4d443bb81f8eef1b6e22ab297f3088caeb9. The dataset is published on Zenodo at https://doi.org/10.5281/zenodo.21738256. Folder structure data/ — parametric study results (one run per subdirectory): bar_length_box_study/ box_size_study/ impact_study/ restitution_coefficient_study/ strain_rate_nsn_nobox_study/ strain_rate_pen_nobox_study/ stress_heatmap_study/ timestep_nobox_study/ notebooks/ — Python helper modules and Jupyter notebooks to inspect the data and regenerate the paper figures. workflows/ — installation, environment, and validation scripts. Each run directory contains: launch.sh — auto-generated launcher that reproduces the run. input/ — archived material files and generated mesh. data/data.h5 — condensed time-series output (energies, fragment counts, etc.). paraview/ — optional ParaView field files when spatial fields were dumped. Quick start: inspect the data without compiling Akantu You only need a standard scientific-Python stack to explore the notebooks and HDF5 data; the Akantu C++ engine is not required. python3 -m venv .venv_data source .venv_data/bin/activate pip install jupyterlab h5py pandas numpy plotly scipy jupyter lab notebooks/ Open notebooks/visualize.ipynb for general inspection or notebooks/visualize_paper.ipynb to regenerate the paper figures. Full reproduction To reproduce individual runs from the dataset or runs with new parameters, first fetch the pinned code and build Akantu: SKIP_SYSTEM_DEPS=true ./workflows/install.sh If you only need the notebooks (no simulation), skip Akantu compilation: SKIP_SYSTEM_DEPS=true SKIP_AKANTU=true ./workflows/install.sh Activate the environment: source workflows/env.sh Run the quick installation test (only when Akantu was built): ./workflows/test_installation.sh Reproduce a single run from its directory: cd data/impact_study/<run_id> ./launch.sh The launcher writes the reproduction to output/<study>/<run_id>/ (with an incrementing _rerun_N suffix if needed), so the original data under data/ is never overwritten. Docker workflow The Docker image is built from the code repository, not from this dataset. You do not need a local copy of the code; the image can be built directly from GitHub. Build the Docker image docker build --network=host -t explicit-nsn:latest \ https://github.com/tghesquie/nsn_frag1d.git#v1.0.0 The --network=host option helps the build reach Debian/GitHub/GitLab mirrors. If your environment forbids it, clone the repository locally and build from there. Run the container Mount only the dataset directories the container needs. Do not mount the whole dataset root to /app, because that would hide the image's source code, compiled Akantu library, and Python environment. mkdir -p output docker run --rm -it \ -v "$(pwd)/data:/app/data:ro" \ -v "$(pwd)/notebooks:/app/notebooks:ro" \ -v "$(pwd)/output:/app/output" \ explicit-nsn:latest bash Reproduce a run inside the container The image entrypoint already activates the environment, so you can run directly: ./workflows/test_installation.sh cd data/bar_length_box_study/nsnfrag1d_<run_id> ./launch.sh Outputs are written to /app/output, which maps to the host ./output. Dependencies Python ≥ 3.10 (see requirements.txt for the notebook environment; the full solver uses the pyproject.toml/uv.lock fetched from the code repository by workflows/install.sh). System packages for compiling Akantu are listed in pkg.txt (Debian/Ubuntu) and are installed via workflows/install.sh. On systems without apt-get, install their equivalents manually. Akantu — FEM infrastructure library, pinned to commit 22adc1e143ca74fdb70af185536d16ff4a3396de. It is cloned and built on demand by workflows/install.sh unless SKIP_AKANTU=true is set. Run-ID naming convention Each run directory encodes the simulation parameters in a compact, parseable form: nsnfrag1d_l<length>_n<n_elements>_md<mesh_density>_p<order>_t<total_time>_r<strain_rate>_s<safety_factor>[_dd<defects_density>][_k<contact_factor> | _e<restitution>_sc<cohesive_factor>][_iv<impact_velocity>][_bc][_box_bf<box_size_factor>]_seed<seed>[_cir<cir>_impact] See data/data/README.md for a full parameter decoding table. Licensing The dataset as a whole (simulation results, metadata, documentation) is released under CC BY 4.0 — see LICENSE.txt. The Python helper files in notebooks/ (query_db.py, plots.py) are derived from the nsn_frag1d software repository and are distributed under GPL-3.0-or-later. Their source and full license text are available in the code repository. DCSM / Solidipes note This dataset was curated with Solidipes and published on Zenodo. It can also be ingested and published through the EPFL DCSM/Solidipes infrastructure. The files pkg.txt and requirements.txt are used by DCSM to pre-install system packages and Python dependencies. So you can start with SKIP_SYSTEM_DEPS=true SKIP_AKANTU=true ./workflows/install.sh and then open the notebooks directly. Contributions Author #1 contributed to Conceptualization, Methodology, Data Curation, Formal Analysis, Software, Validation, Visualization, Writing – original draft, Writing – review & editing Author #2 contributed to Methodology, Data Curation, Funding Acquisition, Software, Supervision, Validation, Writing – review & editing Author #3 contributed to Methodology, Funding Acquisition, Supervision, Validation, Writing – review & editing Author #4 contributed to Methodology, Supervision, Validation, Writing – review & editing Funding sources This research was funded by the Swiss National Science Foundation (SNSF) [Grant N° 212935]