persim

persim sits downstream of every engine that produces a diagram. It supplies the distances and vectorizations — bottleneck and Wasserstein, persistence images, landscapes — that turn a variable-length diagram into something a statistical or machine-learning model can consume.

Keeping it as its own fixture separates the computation of persistence from its representation, which are different decisions with different failure modes.