trap improper posteriors

[bob-carpenter]

We want to be able to trap improper posteriors. Daniel just demoed this model

data {
  int<lower=0> N;
  vector[N] x;
  vector[N] y;
}
parameters {
  real a;
  real b;
  real<lower=0> sigma;
}
model {
  y ~ normal(a + b * x, sigma);
}

and it samples and gives garbage (low n_eff for a and b, but high n_eff for err_sd).

[betanalpha]

Isn’t this proper for any N > 1?

On Jul 19, 2015, at 4:08 PM, Bob Carpenter notifications@github.com wrote:

We want to be able to trap improper posteriors. Daniel just demoed this model

data {
int<lower=0> N;
vector[N] x;
vector[N] y;
}
parameters {
real a;
real b;
real<lower=0> sigma;
}
model {
y ~ normal(a + b * x, sigma);
}
and it samples and gives garbage (low n_eff for a and b, but high n_eff for err_sd).

—
Reply to this email directly or view it on GitHub.

[syclik]

it was for N = 0.

data <- list(N = 0, x = numeric(0), y = numeric(0))

On Sun, Jul 19, 2015 at 11:11 AM, Michael Betancourt <
notifications@github.com> wrote:

Isn’t this proper for any N > 1?

On Jul 19, 2015, at 4:08 PM, Bob Carpenter notifications@github.com
wrote:

We want to be able to trap improper posteriors. Daniel just demoed this
model

data {
int<lower=0> N;
vector[N] x;
vector[N] y;
}
parameters {
real a;
real b;
real<lower=0> sigma;
}
model {
y ~ normal(a + b * x, sigma);
}
and it samples and gives garbage (low n_eff for a and b, but high n_eff
for err_sd).

—
Reply to this email directly or view it on GitHub.

—
Reply to this email directly or view it on GitHub
#1555 (comment).