Nikunj Goel

Hi @andrjohns, I am working on the N-mixture model to estimate abundance from count data. The model relies on the hypergeometric_2F2 and hypergeometric_3F2 functions. Has/will these functions be implemented/exposed in...

I am in the same boat as @bob-carpenter and @wmacnair. A reduce_log_sum_exp function would be great. But in meantime, it will be great if you can provide a working code...

Thank @bob-carpenter for the response. I realize that the reduce_sum function won't work due to a loss in precision. I could not figure out how to implement parallel marginalization using...

Yes @wds15. On the log scale, I am getting lp[n]= -21638. If I take exp(-21638), I get zero. Using @bob-carpenter example, I was using sum_{n=1} ^ {n=n0} exp(lp[n]), but when...

Exactly, reduce_log_sum_exp like function is better because we can pass any kind of argument very easily. There is one point of contention though, unlike map_rect that returns a vector, reduce_sum...

Yes, that makes sense. I would be happy with basically anything

@bob-carpenter and @wds15 would it be reasonable to expect this feature could be released in the next release stan 2.31?

Hi @andrjohns and @spinkney I tried using `log_sum_exp(tuple(vector, vector, vector)) => vector` in my Stan code and I got an error stating the tuples are not supported in `log_sum_exp`. Here...