Persistence Image for H0, H1, and H2 plus dropping inf values about persim HOT 2 CLOSED

Hi @tydymy

1. I don't think there is any way to drop features that extend to infinity automatically using PersistenceImager (but I could be mistaken). You could do something like the following before passing your diagrams into the transformer:

diagrams_h1_finite = diagrams_h1[~np.any(diagrams_h1 == np.inf, axis=1)]

where diagrams_h1 is a numpy array of your diagrams for a fixed homological degree. If you had ripser output, you'd have to slice it first, as in diagrams_h1 = ripser_output[1] or diagrams_h0 = ripser_output[0].

2. Typically, persistence images are used for a fixed homological degree. But there is nothing in the code that verifies this--the fit method takes an iterable of numpy arrays. So you can feed in persistence diagrams from any homological degree, not just degree 1. In particular, you could concatenate your persistence diagrams from different homological degrees into a single list and compute the persistence image of that (but I'm not sure why you would).

from persim.

If you have more questions about this, feel free to open another issue.

from persim.