How I count simplices for each dimension ( python GUDHI )

[Siwon-S-Kim]

Hi

I just started using GUDHI recently and I am having great time with GUDHI. Thank you developers!

I thought just giving thanks is not productive, so I decided to show something that I just come up with. I believe there is no python method to count simplices for each dimension for right now and I also believe it will be there any soon. I think this way, we can count simplices faster than brute force by for loop with SimplexTree.get_simplices() for each dimension. I used .num_simplices() and .prune_above_dimension method.

numbers = {}
for dimension in range(max_dimension,0,-1):
    numbers[dimension] = complex.num_simplices()
    complex.prune_above_dimension(dimension-1)
    numbers[dimension] = numbers[dimension] - complex.num_simplices()

where complex is GUDHI SimplexTree simplicial complex. I am not aware of usual usage of GUDHI actually, but for my case, this seems to work faster. Since I am yet to know actual structure of Simplex Tree, I cannot prove that this is efficient than brute force by for loop over get_simplices().

But anyway thank you developers.

[mglisse]

Thanks for sharing!
This is a clever trick indeed. In some sense it does more work that something like

r=[0]*(st.dimension()+1)
    for s,_ in st.get_simplices():
    r[len(s)-1] += 1

But your code ends up being faster because most of the work is done in C++ (it is often hard to believe how slow Python is).

In the future, we are planning to have a C++ function that returns exactly those numbers (#912) and should be significantly faster without destroying the complex.