You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Returning a list of strategy profiles from Nash solvers is quite thin. It discards useful information about the set of equilibria - including whether the list is known to be complete, whether there are positive-dimensional components, what algorithm parameters were used, and possibly interesting intermediate steps in the algorithm.
We should develop Result classes for Nash solvers, both in Python, and, eventually, at the C++ level. One place to look is at scipy, where e.g. there are Result classes for the results of statistical tests (although there, at present, these are just instances of namedtuple).
The text was updated successfully, but these errors were encountered:
I should mention that it would be particularly helpful to include as comments on this issue, any examples of result classes from algorithms or similar operations from existing libraries!
Returning a list of strategy profiles from Nash solvers is quite thin. It discards useful information about the set of equilibria - including whether the list is known to be complete, whether there are positive-dimensional components, what algorithm parameters were used, and possibly interesting intermediate steps in the algorithm.
We should develop Result classes for Nash solvers, both in Python, and, eventually, at the C++ level. One place to look is at scipy, where e.g. there are Result classes for the results of statistical tests (although there, at present, these are just instances of namedtuple).
The text was updated successfully, but these errors were encountered: