nash_q_learning
nash_q_learning copied to clipboard
tocom242242
Error when using nashpy calculation in replacement environment
When I replace the matrix environment with the small environment I wrote, I use nashpy to calculate the length of pi, and the length of pi often changes. Sometimes there is a case where the length is zero and an error is reported. What does this mean? Does it mean that there are multiple Nash equilibrium points, how should I modify it? Thank you.
[(array([0., 0., 1., 0., 0.]), array([0., 0., 0., 0., 1.])), (array([0.67208459, 0. , 0.32791541, 0. , 0. ]), array([0. , 0. , 0.51307566, 0.48692434, 0. ])), (array([0.51880328, 0. , 0. , 0. , 0.48119672]), array([0. , 0. , 0.11574906, 0.88425094, 0. ]))] 3 This is the calculated pi and its length at a certain occurrence [[0.6559585633182168, 0.497509580230747, 0.4356962170942844, 0.7459217987761908, 0.4440960550343467], [0.594101906637517, 0.14389364237289126, 0.5644707360506582, 0.38696744948674355, 0.5979444623727973], [0.6066314076499245, 0.24148246706700294, 0.5468899220410627, 0.628756196088522, 0.8159105990196025], [0.24635874151293857, 0.6678868022812967, 0.22636373821599765, 0.1195441767231642, 0.5628103686521754], [0.7781361689219078, 0.7911891715865836, 0.025662569218334985, 0.7995954864741698, 0.04006035719713352]] 5 This is the q_1 used by pi and its length
Sorry for not replying sooner @yikeqingli . And thanks for letting me know the issue. Currently I added multi state environment to this repository and confirmed this issue.Please wait a little longer
