Given a heterogenous (discrete) Hidden markov model with hidden states, $x_1, \dots, x_n$ and observed states $o_1, \dots o_n$, known transition matrices $T_1, \dots, T_n$ and emission matrices $E_1, \dots, E_n$ and initial probability vector $I_1$, how can one simulate hidden state sequences from the distribution

$$ (x_1, \dots, x_n) | (o_1, \dots, o_n)?$$

  • $\begingroup$ Check the forward-backward algorithm. $\endgroup$
    – Xi'an
    May 16, 2018 at 16:08
  • $\begingroup$ Hi @svendvn are you looking for a software package to do this or are you asking for an explanation on how the sequence simulation is carried out? $\endgroup$ May 24, 2018 at 7:38
  • $\begingroup$ Hi @ShaunWilkinson. I am looking for an explanation, not a software package $\endgroup$
    – svendvn
    May 24, 2018 at 10:08


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.