SciComp

Status: frozen Category: research Owner: alawein Visibility: public Purpose: Scientific computing utilities and shared numerical tooling. Next action: continue

SciComp is a cross-platform scientific computing suite that keeps Python, MATLAB, and Mathematica in the same repo on purpose. The value here is not just having many numerical topics in one place. The value is maintaining a shared scientific vocabulary across those implementations and exposing where GPU, symbolic, and teaching workflows diverge.

The Python package surface is unusual by normal packaging standards: Python/ is the canonical import boundary. That constraint is deliberate and should stay visible.

Public value

SciComp is a research-portfolio candidate because it demonstrates breadth across numerical methods, languages, GPU workflows, symbolic tooling, and teaching examples. Public polish should make the unusual cross-language layout feel intentional: explain which surface to use, what each language contributes, and how examples can be reproduced.

Core surfaces

• Python/: canonical Python package and CLI entrypoint
• MATLAB/: MATLAB implementation surface
• Mathematica/: Mathematica notebooks and symbolic workflows
• examples/: runnable demos
• tests/: cross-surface verification
• docs/: API, theory, installation, and troubleshooting

Focus areas

• Quantum mechanics and quantum computing
• Thermal transport and heat-transfer numerics
• Physics-informed machine learning
• GPU acceleration with explicit CPU fallback
• Spintronics, optics, control, and related computational physics modules

Quick start

git clone https://github.com/alawein/scicomp.git
cd scicomp
pip install -e ".[dev]"
python scripts/validate_framework.py
pytest

CLI

berkeley-scicomp --help
bsc --help

Development

pytest --cov=Python
ruff check Python/
mypy Python/
python scripts/validate_framework.py
matlab -batch "run('tests/matlab/test_heat_transfer.m')"

Publication boundaries

Keep machine-specific MATLAB/Mathematica paths, GPU environment assumptions, large generated outputs, and unpublished teaching/research data out of public examples. If a notebook or script supports a claim, document the required runtime and expected output.

Documentation

Start with docs/README.md for installation notes, API reference, GPU guidance, and theory material.