Apologies if this topic is in the wrong place. I’ve been using a simple toy model to experiment with the rstan and loo packages (most current verison of each, as far as I’m aware), and I noticed what appears to be a discrepancy between the relative effective sample sizes I get from the summary meth…

Effective sample size discrepancy between loo and rstan packages

Max_Mantei December 29, 2018, 10:05am 3

Did you have a look at this thread?

Topic		Replies	Views
Getting warning with `loo` package regarding r_eff Modeling loo	5	2747	April 21, 2018
Error in loo::relative_eff( ) - Error in n_eff_vec/S : non-numeric argument to binary operator Modeling rstan , fitting-issues , loo	7	1234	May 13, 2020
Errors when calculating PSIS-LOO CV in loo version 2.2.0 Modeling fitting-issues , loo	7	1247	January 2, 2020
Problems running LOO on >1 core General loo	29	3610	July 29, 2020
Target += normal_lpdf(eta \| 0, 1); vs. eta ~ normal(0,1) Modeling rstan	9	1937	July 22, 2021