"trying deprecated constructor" error for observation error model

Modeling
#1

I am having trouble with an observation error model that won’t compile correctly in rstan despite being syntactically correct. The error I keep getting is:

trying deprecated constructor; please alert package maintainer
Error in new_CppObject_xp(fields$.module, fields$.pointer, …) :
no valid constructor available for the argument list
failed to create the sampler; sampling not done

It doesn’t even create a .RDS file

Here is the associated stan file (also attached):
data {
//provided data for a sample
int<lower = 1> N; //number of individuals
int<lower = 1> Sites; // number of sites
int<lower = 1> N_Msres; //number of measurements
int<lower = 1> Msre_Types; //number of different measurement types
int<lower = 1, upper = Sites> Site[N]; //what site is the individual from?
real msre_obs[N_Msres];
real<lower = 0> msre_sd[N_Msres];
real ref_obs[Msre_Types]; //reference values for the measurements
real<lower = 0> ref_sd[Msre_Types];
int<lower = 1, upper = Msre_Types> msre_num[N_Msres];
int<lower = 1, upper = N> msre_Indiv[N_Msres];
}
parameters {
vector<lower = 0>[N_Msres] Msre;
vector<lower = 0>[Msre_Types] Ref;
vector[N] LSI;
real<lower = 0> LSI_sd; //variation of LSI values within an individual
//modeling average LSI for a site
real mu_LSI;
vector<lower = 0>[Sites] site_sigma;
}
model {
real LSI_msre;
//measurement error
msre_obs ~ normal(Msre, msre_sd);
ref_obs ~ normal(Ref, ref_sd);
LSI_sd ~ normal(0, 0.05);
for(i in 1:N_Msres) {
LSI_msre = log(Msre[i] / Ref[msre_num[i]]);
LSI_msre ~ normal(LSI[msre_Indiv[i]], LSI_sd);
}
//modeling average LSI for a site
mu_LSI ~ normal(0, 1);
site_sigma ~ normal(0, 1);
for(i in 1:N) {
LSI[i] ~ normal(mu_LSI, site_sigma[Site[i]]);
}
}

Thank you!

Here is the script and an attached R script and associated data file.LSI_observationerror.stan (1.3 KB)
R Script for Error in Observation Model.R (1.2 KB)
Bos Size Dataset.csv (137.3 KB)

#2

rstan 2.17.3 will / does say the right error message:

Exception: mismatch in number dimensions declared and found in context; processing stage=data initialization; variable name=Site; dims declared=(1504); dims found=(1504,2) (in ‘model421f2e2fd998_LSI_observationerror’ at line 7)

since

> str(LSIdata_observationerror$Site)
Classes ‘data.table’ and 'data.frame':	1504 obs. of  2 variables:
 $ V1: int  2 2 2 2 2 2 2 2 2 2 ...
 $ V1: int  1 9 10 11 12 13 30 29 2 15 ...
 - attr(*, ".internal.selfref")=<externalptr>
#3

Thank you, that is exactly what was wrong and now I feel embarrassed. Is it
safe to say that until 2.17.3 I should consider that error as a
"something’s wrong with my data initialization" or is it more of a general
"something’s wrong" message?

#4

Something’s wrong with your data initialization, and you can now install 2.17.3 to see exactly what is wrong.

#5

just to let you know that this solved my problem also: I was using a scalar as data “hoping” that it would be recycled into the vector I’d declared, but as soon as I provided a vector of the right length, everything worked.