Cognitive Discovery System (CDS) a scientific computing library written in pure Python. No NumPy, no SciPy, no compiled extensions. Zero dependencies, runs anywhere.
18 modules: quantum circuit simulation, FFT/IFFT, ODE solvers, numerical integration, linear algebra, statistics, Monte Carlo, optimization, graph algorithms, symbolic math, a from-scratch NLP stack (BPE → attention → MiniGPT + scalar autograd), and a hypothesis-generation module.
The idea: pure Python can't beat C on raw loops, but it can close the gap algorithmically — O(N log N) FFT, O(N³) LU decomposition (instead of O(N!) determinants), O(1) probabilistic quantum measurement. Not a NumPy replacement; a readable, dependency-light toolkit.
1,441 tests · 100% coverage · on PyPI https://pypi.org/project/cognitive-discovery-system/
Repo: https://github.com/Furox88/cognitive-discovery-system Install: pip install cognitive-discovery-system
If you find it useful, a star on the repo means a lot. The autograd part is the one I'd most appreciate eyes on.
If anyone has time to glance at the from-scratch autograd implementation, that's the part I'm least sure about. Open to feedback
Top comments (0)