Thank you very much. The quick fix works well. But after computing n diagrams, I want to compute a n*n distance matrix in parallel. As you mentioned, the persistence diagrams are not pickleable, making it hard for parallel computing in python.
I put the request on github.