# Idea In Stan's output, `lp__` is the **log density up to a constant**. Stan implicitly removes terms in the likelihood that are independent of any model parameters. These terms are called scale or scaling factors. They are constants, so removing them won't affect parameter estimation. However, because the constants are missing from `lp__`, it is not the true log density. To obtain the true log density, we must include the constants. # References - https://www.jax.org/news-and-insights/jax-blog/2015/october/lp-in-stan-output# ![[lp__.pdf]]