Thanks! Sadly, I haven’t had the chance to use this stuff in the real world yet, but I’m planning on adding a few more realistic examples. Pull requests or issues with examples or ideas would be very welcome!
Yeah, there are quite a few projects out there. Here’s a quick overview of what I know about:
- Ripser.jl: my wrapper of the original C++ program. Very bare-bones and a bit outdated, but does essentially the same thing. Being a C++ wrapper, things like cubical homology are out of the question. According to my benchmarks, it’s around 30% faster. I’m still working on getting that number lower and from what I’ve seen Ripserer will actually outperform it once Julia 1.5 comes out. I haven’t tested enough to be sure.
- Sparips.jl: this is a preprocessor that comes with a different wrapper of Ripser. The preprocessor allows you to compute persistent homology of very large datasets. Stealing its code and adding it to Ripserer is on my TODO list.
- ComputationalHomology.jl: haven’t used it yet, but from the looks of it, it does quite a few things besides persistent homology. From what I’ve heard, the persistent homology algorithm is orders of magnitude slower than Ripserer.
- PersistentCohomology.jl: does essentially the same thing, but with a different algorithm, that is orders of magnitude slower.
- Eirene.jl: Not really a package, but more like an interactive tool for computing persistent homology. I tried wrapping my head around the algorithm it uses and failed. Has heaps of features. From what I heard its actually pretty fast. I still haven’t taken the time to learn to use it.