HMM with time-varying probabilities

[jsocolar]

One more thing:
To the best of my knowledge, Stan’s built-in hmm_marginal function only works with time-invariant transition probabilities. Your title suggests that you are working with time-varying transition probabilities, though I don’t quite grasp how time-varying transition probabilities enter into your description of the problem.

When you say

@Valentyn

I know that Stan uses only forward probabilities to compute the log-likelihood.

I’m led to believe that you are working with hmm_marginal and not your own implementation of the forward algorithm. If that’s true, and if it’s true that you intend to work with time-varying transition probabilities, then I think something unintentional must be happening such that you are not passing the time-varying transition probabilities that you want. It is not too difficult to write reasonably efficient implementations of the forward algorithm in Stan code without using hmm_marginal, and to set these up to accept time-varying transition matrices. Some resources linked here:

Covariates in Hidden Markov Model with hmm_marginal

Hi everyone, I am working on a Hidden Markov Model, which will eventually be used for analysing species detections from camera trap data. Currently I have a model which uses the new hmm_marginal function - the full code for the model is below. The model has two hidden states (1 = unoccupied, 2 = occupied) and a Poisson emission process representing the number of detections at each time step (e.g. the number of camera trap photos taken in each 30-minute block of time). The model takes data from…