Cannot provide an empty int-valued list

[riddell-stan]

Models which use data which contains an empty int-valued list cannot be used.

This does not have a trivial solution since we use JSON for serializing data and JSON cannot distinguish between an empty list of floats and an empty list of ints.

A completely satisfactory solution will probably involve asking Stan or stanc3 to tell us the type of data variables.

[ahartikainen]

Stanc3 should have this option, because pystan 2 has the same problem.

[riddell-stan]

I think cmdstan might have solved this. If this is true, we could copy their solution.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[ahartikainen]

Now with Stan 2.27 stanc3 has --info which returns a dictionary with needed information.

stanc --info path/to/stan_file.stan

(Here is an example with bernoulli example in CmdStan examples.)

{ "inputs": { "N": { "type": "int", "dimensions": 0},
              "y": { "type": "int", "dimensions": 1} },
  "parameters": { "theta": { "type": "real", "dimensions": 0} },
  "transformed parameters": {  },
  "generated quantities": {  },
  "functions": [  ],
  "distributions": [ "bernoulli_lupmf",
                     "beta_lupdf" ] }

And in httpstan solution (copy from httpstan/compile) with example from 8-schools (pystan docs)

import httpstan
import stan
import importlib.resources
import tempfile
from pathlib import Path
from typing import List, Union, Tuple
import os
import subprocess
import json

program_code = """
data {
  int<lower=0> J;         // number of schools
  real y[J];              // estimated treatment effects
  real<lower=0> sigma[J]; // standard error of effect estimates
}
parameters {
  real mu;                // population treatment effect
  real<lower=0> tau;      // standard deviation in treatment effects
  vector[J] eta;          // unscaled deviation from mu by school
}
transformed parameters {
  vector[J] theta = mu + tau * eta;        // school treatment effects
}
model {
  target += normal_lpdf(eta | 0, 1);       // prior log-density
  target += normal_lpdf(y | theta, sigma); // log-likelihood
}
"""

schools_data = {"J": 8,
                "y": [28,  8, -3,  7, -1,  1, 18, 12],
                "sigma": [15, 10, 16, 11,  9, 11, 10, 18]}

posterior = stan.build(program_code, data=schools_data)

def model_info(program_code: str, stan_model_name: str) -> Tuple[str, str]:
    with importlib.resources.path(httpstan.__package__, "stanc") as stanc_binary:
        with tempfile.TemporaryDirectory(prefix="httpstan_") as tmpdir:
            filepath = Path(tmpdir) / f"{stan_model_name}.stan"
            with filepath.open("w") as fh:
                fh.write(program_code)
            run_args: List[Union[os.PathLike, str]] = [
                stanc_binary,
                "--info",
                str(filepath),
            ]
            completed_process = subprocess.run(run_args, capture_output=True, timeout=1)
    stderr = completed_process.stderr.decode().strip()
    if completed_process.returncode != 0:
        raise ValueError(stderr)
    return completed_process.stdout.decode().strip(), stderr

model_name = httpstan.models.calculate_model_name(program_code)
stan_model_name = f"model_{model_name.split('/')[1]}"
stdout, stderr = model_info(program_code, stan_model_name)
model_input_info = json.loads(stdout)
model_input_info

{'inputs': {'J': {'type': 'int', 'dimensions': 0},
  'y': {'type': 'real', 'dimensions': 1},
  'sigma': {'type': 'real', 'dimensions': 1}},
 'parameters': {'mu': {'type': 'real', 'dimensions': 0},
  'tau': {'type': 'real', 'dimensions': 0},
  'eta': {'type': 'real', 'dimensions': 1}},
 'transformed parameters': {'theta': {'type': 'real', 'dimensions': 1}},
 'generated quantities': {},
 'functions': [],
 'distributions': ['normal_lpdf']}

edit. Fix code