hu limit was solved in mostly the same manner that hu postflop solvers work. It just uses a ton of compression to fit the game in ~10tb and then stores memory on disk (distributed across nodes) instead of ram. One of the authors created jesolver
http://jeskola.net/jesolver_beta/
The hu limit solution has a nash distance that is just like the nash distance that commercial solvers have. They said that the nash distance is small enough that you could play best response vs it for a lifetime and still not be able to say that you are beating it with statistical significance. They also showed some math that an exploitability of 0.1bb/100 is small enough to meet for that to be true.
multiway limit could be done in the same manner that pluribus did no limit, but i don't think anyone has done so
Last edited by valuecutting; 10-07-2021 at 11:16 PM.