The papers above have associated code, wich is publicly available on github, as python libraries.
[Github] Multiparameter Module Approximation,
This library computes descriptors of multiparameter persistence modules, aswell as their vectorizations.
It estimates interval decompositions of these modules, which allows to have interpretable outputs,
and efficiently store the rank invariant on diagonal lines.
Furthermore, the library can make use of several other multiparameter persistence libraries,
and can therefore deal with the majority of existing multi-filtrations
(e.g., Rips-Density-like, or Degree-Rips-like bifiltrations, Function-delaunay, or any one critical multifiltration),
and benefits from various theoretical optimizations (e.g. minimal presentations, multiparameter edge collapse).
We also provide ScikitLearn-style pipeline to use these tools in Machine Learning.
[Github] Multiparameter Signed Barcodes as Signed Measures(which is an extension of the one above)merged with above.
This library has the same goals of the previous one, but uses a very different approach.
It turns out that the majority of multiparameter persistence module invariants,
such as (pointwise dimension / euler characteristic, rank invariant),
can be encoded as discrete signed measures.
By leveraging on this, we provide different strategies, in a ScikitLearn style manner
to use efficiently these mathematical structures in a Machine Learning context.
I'm also working on putting this work into GUDHI, you can follow its status around here.