Number of components study: on simulated data

[GregoryAshton]

I'm going to collect here what I do just playing around with varying the number of components (hope you don't mind me using the issues Colm).

Here is just a quick run using the SplineLorentizian (with no Lorentzians) on simulated data and 11 spline points.

[GregoryAshton]

So here is a plot showing the increasing, then thresholding of the evidence as you increase the number of splines

For reference, the 3-component model check looks like this

while the 9-component looks like

[GregoryAshton]

Increasing the number of components, we can start to see the turn over

[ColmTalbot]

I hadn't seen this, apparently, I don't get emails from this. Cool!