This is related to my post here, but I changed the model and data structure enough to warrant a new post.
The model is as follows:
//This stan code fits a model with lags done ahead of time, rather than in the
// Stan code
//lags should be calculated manually, and the first observation in each time
//series should be removed before data is given to stan
functions{
real partial_sum(matrix[] y_slice,
int start, int end,
matrix lagvars,
vector alpha, matrix beta,
matrix Omega, vector[] tau,
int[] country)
{
for(i in start:end){
return multi_normal_lpdf(to_vector(y_slice[i]) | alpha[country[i]] + to_matrix(beta[country[i]])*lagvars[i]',
quad_form_diag(to_matrix(Omega[country[i]]), tau[country[i]]));
}
}
}
data{
int N; //# observations
int K; //# dimensions of Y
int C; //# of countries
int R; //# of regions
int<lower = 1, upper=C> country[N]; //country id for each obs
int<lower = 1, upper=N> n[N]; //observation id for each obs
matrix[N,K] Y; //the outcome array - each variable's time series stacked by region
matrix[N,K] L; //the lagged variables
}
parameters{
//individual level
vector<lower = 0>[K] tau[C]; //scale for residuals
matrix[K, K] z_beta[C]; //untransformed betas
vector[K] z_alpha[C]; //untransformed intercepts
//hierarchical parameters
corr_matrix[K] Omega[C]; //country level correlation matrix
vector[K] tau_loc; //mean for variance scaling factor
vector<lower=0>[K] tau_scale; //scale for tau
matrix[K, K] bhat_location; //mean for prior on beta
matrix<lower = 0>[K, K] bhat_scale; //scale for prior on beta
vector[K] ahat_location; //means for prior on intercepts
vector<lower = 0>[K] ahat_scale; //variance for intercept prior
}
transformed parameters{
matrix[K, K] beta[C]; //VAR(1) coefficients, country specific
vector[K] alpha[C]; //country specific intercepts
for(c in 1:C){
//recentering random effects
alpha[c] = ahat_location + ahat_scale .*z_alpha[c];
beta[c] = bhat_location + bhat_scale*z_beta[c];
}
}
model{
int grainsize = 1;
//hyperpriors
tau_loc ~ cauchy(0,1);The pa
tau_scale ~ cauchy(0,1);
ahat_location ~ normal(0,1);
ahat_scale ~ cauchy(0, 1);
to_vector(bhat_location) ~ normal(0, 0.5);
to_vector(bhat_scale) ~ cauchy(0, 0.5);
//hierarchical priors
for(c in 1:C){
//non-centered paramaterization to avoid convergence issues
z_alpha[c] ~ normal(0, 1);
to_vector(z_beta[c]) ~ normal(0, 1);
tau[c] ~ normal(tau_loc, tau_scale);
Omega[c] ~ lkj_corr(5);
}
target+= reduce_sum(partial_sum, Y, grainsize, L,
alpha, beta, Omega, tau,
country);
}
This fails to compile with the following error:
Ill-typed arguments supplied to function 'reduce_sum'. Available arguments:
(T[], int, int, ...) => real, T[], int, ...
(T[,], int, int, ...) => real, T[,], int, ...
(T[,,], int, int, ...) => real, T[,,], int, ...
(T[,,,], int, int, ...) => real, T[,,,], int, ...
(T[,,,,], int, int, ...) => real, T[,,,,], int, ...
(T[,,,,,], int, int, ...) => real, T[,,,,,], int, ...
(T[,,,,,,], int, int, ...) => real, T[,,,,,,], int, ...
Where T is any one of int, real, vector, row_vector or matrix.
Instead supplied arguments of incompatible type: (matrix, int, int, matrix, vector[], matrix[], matrix[], vector[], int[]) => real, matrix, int, matrix, vector[], matrix[], matrix[], vector[], int[]
Error: An error occured during compilation! See the message above for more information.
I don’t know exactly what this error is trying to tell me, specifically which term is incompatible. This code compiles and runs using only partial_sum in the model block, so I’m guessing it’s some issue with how I’m passing or defining the arguments and reduce_sum, but I’m in the dark as to what the issue is.