Positive lp__ value

Hi there,

May I ask a silly question? Does lp__ value have to be negative? I was trying to use “mcmc_nuts_divergence” to diagnostics the model, the plots are just as expected, but the lp__ values are all positive, is it normal to have positive lp__? As I understand it is the log density, then it should not be restricted as negative. When density is larger than 1 log should be positive though, but seems total density should not exceed 1, but we don’t normalize posterior in HMC then it could exceed 1… I am confused, could you please help to explain or give some hints or material to read? By the way, as I remember most of the vignettes I read so far seem to have negative values.

Thanks,
Jean-Chen

It can be either positive or negative since it is the logarithm of a posterior PDF possibly ignoring constants.

2 Likes

Thanks for the clarification.