Bruno Levy

- You are curious about Optimal Transport ?
- You'd like to know what is "Semi-Discrete" ?
- You'd like to know how it helps computational cosmology ?

Short version of the story:
https://github.com/BrunoLevy/GraphiteThree/wiki/Transport_intro

P.S. post-doc position available very soon, stay tuned (or DM me if you can't wait).

Mar 19 at 9:30amWeb
Show threadj_bertolotti1d ago
@BrunoLevy01 Now I need to find the time to translate your LUA tutorial into Mathematica 😉
Show threadBruno Levy1d ago
@j_bertolotti we are currently working on a Python/numpy tutorial with @pltc. The most difficult part is computing the Laguerre diagram, the rest is probably easy to port.
Show threadj_bertolotti1d ago
@BrunoLevy01 @pltc The big advantage of Mathematica is that it already has baked-in functions do to this kind of stuff.
Show threadj_bertolotti1d ago
@BrunoLevy01 @pltc Ok, Mathematica has a fast Voronoi digram function, but no Laguerre diagram one. It is easy to find online simple codes that do the trick though.
Plotting is very slow right now, but I don't think I will have to plot a lot for this to work.
Show threadBruno Levy1d ago
@pltc good to know, thank you @j_bertolotti ! I'll consider porting the tutorial into several languages (but once Python is finished, we have Julia to do, because one of my collaborator friend is using it !)
Show threadBruno Levy1d ago
@j_bertolotti discussing with @pltc, it seems that what your figure shows is already a Laguerre diagram (or additively-weighted Voronoi diagram), so Matlab probably already have everything that we need ! We'll publish a github repo soon, PRs will be welcome !
Show threadj_bertolotti1d ago
@BrunoLevy01 @pltc It is, but it was not out of the box. Had to code a function that did it.
(Btw: Mathematica, not MatLab 😉 )
Show thread🇺🇦 haxadecimal 🚫👑1d ago
@BrunoLevy01
As it happens, I've been trying to do a 2D least action recently. Before looking at your GitHub repo, I didn't know the terms "least action" or "minimal transport".
So far, my crude attempt at it is far too computationally intensive. I was trying to come up with some heuristics. Now I'll dig deeper into what you've written.