The return result of birth-death seems like radius ($r$) rather than epsilon ($\epsilon=2r$), which doesn't keep the same representation of VR-complex.
When the radius reaches to 1.01666667 or 1.11803399, all 1--simplex(es) in Cech filtration (topological space?) looks have disappeared. So why larger radius are traced? Or do I have any misunderstanding of Cech filtration?
The figure with circle radius=1.01666667 is shown as follows.
where the orange edges are edges whose length <=the specific radius 1.01666667
I also tried the same point set in GUDHI 3.8(C++ version), and the output is as follows, which looks more reasonable.
So I guess if my understanding is right, the backend of CechComplex from GUDHI may need to be updated.
When using
EuclideanCechPersistence
to calculate persistence homology, the return result looks wrong.test code:
The return of
diag
is:The problems are:
The figure with circle radius=1.01666667 is shown as follows. where the orange edges are edges whose length <=the specific radius 1.01666667
I also tried the same point set in
GUDHI
3.8(C++ version), and the output is as follows, which looks more reasonable. So I guess if my understanding is right, the backend of CechComplex from GUDHI may need to be updated.with the core codes: