Traceability
The code should make it possible to trace a result back to a method and a set of assumptions.
Software
Code is part of the argument.
A research repository should make it easier to understand what was computed, which assumptions were used, and how another person could inspect or repeat the calculation.
Software principles
What I want public code to provide.
Not every repository is a polished package, but its role should still be clear.
Reviewability
Important numerical conventions should be visible rather than hidden behind a black box.
Reproducibility
Inputs, parameters, versions, tests, and validation material should be explicit when they matter.
Topological data analysis
fast-ballmapper
Efficient cover construction for the Ball Mapper algorithm, with exact and approximate range query backends and reproducible performance experiments.
Bone image analysis
bone_morphometry
Region aware morphometry for trabecular bone images in three dimensions, including conventional measurements and advanced structural descriptors.
Bone strength modelling
boneStrengthPrediction
Scripts and experiments for structural and topological analysis related to apparent bone strength.
Earlier topology research
tda-centrality
Code associated with earlier work on cycle centrality measures built from homology and the master’s thesis that preceded the published paper.
Ball Mapper research
probabilistic_ball_mapper
Experimental work on probabilistic and approximate variants of Ball Mapper style constructions.
A note on maturity
Public does not always mean finished.
Some repositories are maintained software, some are active research code, and some are archives. The status shown above is meant to make that distinction visible rather than presenting every repository as equally mature.