Software and reproducible research projects.
These are small scientific Python packages that I maintain as research software. The emphasis is on focused numerical methods, tested examples, documentation, and clear scope boundaries.
cholrot is a focused package for rank-one Cholesky updates, downdates, modified-factor products, and solves.
It may be useful in numerical experiments where a positive-definite matrix changes by a rank-one term and recomputing or materializing every intermediate matrix is unnecessary. The package is intentionally narrow: it does not replace LAPACK or dense Cholesky factorization, but provides routines for rank-one modification workflows.
Implementation: Python API with C++/pybind11 kernels and pure Python fallbacks.
Project page · GitHub · PyPI · Documentation
lattice-dsp is a research-oriented package for experiments with stability-aware recursive signal processing: IIR lattice filters, reflection/PARCOR coefficient parameterizations, adaptive recursive filtering, AR diagnostics, and finite-dimensional MIMO/model-reduction examples.
The scalar filtering pieces are the most direct practical APIs. The MIMO, Schur/Pick, and model-reduction pieces are best read as tested experimental diagnostics rather than production solvers.
Implementation: Python API with C++/pybind11 kernels and optional OpenMP batch processing.
Project page · GitHub · PyPI · Documentation
robustcov is a research-oriented Python/C++ package for robust covariance and scatter estimation, SPD geometry, anomaly diagnostics, and robust kernel/similarity workflows.
It is intended for workflows where classical covariance estimates are unstable: contaminated samples, heavy-tailed data, small-sample regimes, high-dimensional scatter estimation, Mahalanobis-style anomaly screening, and input-metric construction for kernel or similarity methods. The package includes robust scatter estimators, automatic estimator-selection helpers, robust outlier detectors, SPD geometry utilities, diagnostic reports, plotting utilities, benchmarks, and use-case galleries.
Implementation: Python API with C++/pybind11 kernels and optional OpenMP acceleration.
Project page · GitHub · PyPI · Documentation