New cmdstan ode_solvers and brms

Interfaces brms
6

Nope… brms is just fine. You just need to get used as to how the brms stan source generation works. The make_stancode thing is the way to go… although I don’t use the parse option as it is easier first to dump that into a file and check it out. Here is what samples for me just fine:

fo_model2 <- "
  vector fo2(real t,                  //time
             vector y,                // the state
             real k1){                // the parameter next to A0

             vector[1] dydt;          // dimension of ODE
             dydt[1] = - k1 * y[1];   // ODE
             return dydt;             // returns a 1-dimensional vector
  }

// function called from brms: integration of ODEs
vector fo_ode2(data vector time, vector vA0, vector vk1) {
        vector[1] y0 = rep_vector(vA0[1], 1);        //one initial value
        int N = size(time);
        vector[1] y_hat_ode[N] = ode_rk45(fo2, y0 ,0.0,to_array_1d(time),vk1[1]);
        vector[N] y_hat;
        for(i in 1:N) y_hat[i] = y_hat_ode[i,1];
  return(y_hat);
}

"

df_sim_ode <- df_sim[-1,] #remove the first datapoint at t0

fo_formula <- bf(y_obs~fo_ode2(time,A0,k1),
                  A0~1,
                  k1~1,
                  nl=TRUE, loop=FALSE)

fo_priors <- c(prior(normal(100,10),nlpar=A0),
                prior(normal(0.2,0.1), nlpar=k1),
                prior(cauchy(0,10), class=sigma)
                )

fo_result2 <- brm(data=df_sim_ode,family = gaussian,formula=fo_formula,prior=fo_priors, inits=0,iter=4000,chains=4, cores=4, stanvars = stanvar(scode=fo_model2, block="functions"), backend="cmdstanr", file="fo_result2")
1 Like