the definitions of p(y_t | x_t) and p(x_t | x_t-1) are strange

[jovo]

in particular, the constants typically go before the \exp part (in my experience)