Q: What is a good way to compute the plot bounds for reasonable visualisation?
I've considered alternatives, eg:
- set the smallest $(\mu_i - 3\sigma_i)$ and largest $(\mu_i - 3\sigma_ i)$ for the left and right bound, respectively. (Here, $i$ indexes the priors.)
- set the bounds using a small threshold for the derivative of each $pdf_i$.
The difficulty is caused by possibly disparate priors: flat vs sharp. There are UI solutions (e.g. user-defined plot bounds) but they're laborious, and ignore the initial bounds choice for display initialisation.