Field characteristic

[mbr085]

It seems that multipers does all computations over the field with two elements.

Add the homology_coeff_field=11 parameter present in the Gudhi SimplexTree compute_persistence method.

import multipers as mp
from multipers.data.synthetic import noisy_annulus
X = noisy_annulus(n1=500, n2=0)
from multipers.filtrations.density import KDE
density = KDE(bandwidth=0.2).fit(X).score_samples(X)
# a bit of renormalization
density -= density.min()
density /= density.max()
from multipers.filtrations import RipsLowerstar, RipsCodensity
simplextree = RipsLowerstar(points=X, function = 1-density, threshold_radius=2.5) # this is a simplicial complex
simplextree.collapse_edges(-2)                             # Removes some unnecessary edges (while staying quasi isomorphic)
simplextree.expansion(2)                                   # Adds the 2-simplices (necessary for $H_1$ computations)
bimod = mp.module_approximation(simplextree, homology_coeff_field=11)

Checks.

• I have checked the issue tracker and the example zoo (in the documentation), and the base field is never mentioned.

[DavidLapous]

Hi @mbr085
Not all computations are in Z/2Z, but most of them are indeed. I didn't put that in the library since multipersistence computations are already slow enough, and some of the dependencies are only implemented in F_2 as well.
Do you have a specific usage in mind? adding that to the whole library would be a big work, but some part would only require changing a few variables in the cpp templates.

[mbr085]

I had an example where I wondered if I was confused by orientability, but it turned out not be the case. So no, I do not have a strong usage case in mind.