Dmitriy

The issue is almost surely with the high multiplicity of the points in both diagrams. You could try manually perturbing them by a tiny epsilon and see if that solves...

Well, you can always manually perturb the diagrams by epsilon. This will fix the multiplicity problem and will introduce arbitrarily small error to the Wasserstein distance. As for parallelization, we...

Dionysus uses `float` for simplex data, so the points in the diagram have `float` coordinates.

@grey-narn So what's the right strategy? If I switch simplex data to `double`, then the coordinates of the points in the diagram become double, and I'm guessing we can trigger...

@grey-narn Yeah, the general solution would require a much more complex yoga with exact predicates and such. We are not going there any time soon. I guess the real question...

Thanks. What compiler ran into this problem?

@jvuillamy For me, these changes cause a segfault. [Same on Travis](https://travis-ci.org/mrzv/dionysus/builds/436580012).

Hm, I think I'd rather keep a single function (`bottleneck_distance`), and have it take an extra parameter (`edge`) to indicate whether to compute a matching. It would have to return...

@grey-narn This is another vote for the feature you mentioned.

It would be nice to be able to print only the options, without the usage. I prefer to write my own usage line that's more informative than the default, and...