8

I am curious whether it is significantly harder (computational-wise) to compute persistent homology as compared to computing homology. Or is it the same time complexity.

I am aware that there is a reduction algorithm for homology. I vaguely recall that it has polynomial complexity, though I can't find the reference now. I would be grateful if someone can point me to any reference.

From the description of persistent homology, it is clear that it can't be easier than computing homology. But for computing one homology group, say $H_n(X)$, vs computing one persistent homology group, say $H_n^{i,p}$, I am not sure which is easier. By Carlsson's theory, persistent homology is the same as the usual homology of a graded module, hence I am not sure if this implies that the complexity is the same as computing homology.

Thanks for any help. Any references will be greatly appreciated.

Alexander Gruber
  • 28,037
yoyostein
  • 20,428

1 Answers1

1

Over finite fields, its the same complexity, matrix multiplication time. See "Zig-zag persistence in matrix multiplication time" Over the integers, aside from some ambiguity of how to define persistent homology, the problem lies in the explosion of coefficient size (due to torsion).

user1536
  • 111
  • 1