deletions | additions       diff --git a/untitled.tex b/untitled.tex  index 984a6fc..f8b4bf0 100644  --- a/untitled.tex  +++ b/untitled.tex 
...
\subsection{moment fitting}  We record a number of moments of the observed variables, e.g.$$,  $ $, \langle V \rangle$, $ \langle V^2 \rangle$,  etc. We then fit analytical moments of $P(V|\theta)$ to the observed moments. The advantage of this is that you don't need to save every data point, just the running averages. The downside is obviously that you're only fitting a few parameters. This might be good if you have too many observations to realistically record them all. \subsection{Bayesian updates}