Closed mclements closed 1 month ago
I now see that a point mass is sampled using:
double trandom(double lower, double upper) const {
return lower + std::min(est_, upper - lower);
}
so that this is a constant time in state -- which is arguably clock-reset. Is this the desired semantics? If so, I will close this issue.
I think it's the behavior that is most useful and practical. What do you think? It could certainly be better documented though.
Thanks for these comments. Some further documentation would be helpful -- after the pull request:)? I will close this issue.
The fifth transition is fixed at one year, which makes sense for a clock-reset transition. However, the one-year transition could be in the past for a Markov model:(. Should we change the transition to being exponential (say)?
Kindly, Mark.