Calculate cumulative sum in stan

Modeling
1

model {
alpha ~ normal(0,10);
sigma ~ cauchy(0,10);
b_0 ~ cauchy(0,10);
b_titer ~ cauchy(0,10);

for (n in 1:N){
vector[N] theta;
theta[n] = inv_logit(b_0 + b_titer*low[n]);

for (i in 1:100){
vector[100] theta_sep;
theta_sep[i] = inv_logit(b_0 + b_titer*(low[n]+i*(diff[n]/100)));

}

theta[n] = theta[n] + sum(theta_sep);

target += bernoulli_lpmf(infection[n] | theta[n]);

}

always error, than there is no thets_sep…

2

Hi! It seems your theta_sep variable gets reassigned every iteration during the for loop over i while you only define theta_sep[i] in every iteration. This would mean that theta_sep[-i] is not defined and therefore you cannot sum this. Can you try the following?

model {
alpha ~ normal(0,10);
sigma ~ cauchy(0,10);
b_0 ~ cauchy(0,10);
b_titer ~ cauchy(0,10);

for (n in 1:N){
  vector[N] theta;
  vector[100] theta_sep;
  theta[n] = inv_logit(b_0 + b_titer*low[n]);

  for (i in 1:100){
    theta_sep[i] = inv_logit(b_0 + b_titer*(low[n]+i*(diff[n]/100)));
  }

  theta[n] = theta[n] + sum(theta_sep);
  target += bernoulli_lpmf(infection[n] | theta[n]);
}

Also, if this doesn’t work, can you copy and paste the exact error message?

3

You probably also want to move theta out of the n loop. For the same reason.

model {
vector[N] theta;
alpha ~ normal(0,10);
sigma ~ cauchy(0,10);
b_0 ~ cauchy(0,10);
b_titer ~ cauchy(0,10);

for (n in 1:N){
  vector[100] theta_sep;
  theta[n] = inv_logit(b_0 + b_titer*low[n]);

  for (i in 1:100){
    theta_sep[i] = inv_logit(b_0 + b_titer*(low[n]+i*(diff[n]/100)));
  }

  theta[n] = theta[n] + sum(theta_sep);
  target += bernoulli_lpmf(infection[n] | theta[n]);
}
}
1 Like
4

Yes, good catch!

5

yeah, I changed, works now, thanks all :p

1 Like