Closed jasondk closed 3 years ago
@AlbertoCasasOrtiz I see in #158 that you rescaled for visualization. I think the larger problem was that the integration cutoff was being set to be no larger than 1e-3 - so most of the probability mass wasn't being computed. I suspect that with my commit c50d8cf5a142d555c586c8a82c2868f5c1ccb917 this is no longer an issue (where I increased the max value for the integration cutoff to 1)
It still not working for me, what parameters are you using?
I think the key thing is to change the transition probability to something like 1e-3 so that the transition occurs relatively early