Replace difference plot in fitobserver with the true metric used for minimization
|Status:||Backlog||Start date:||18 Oct 2018|
The objective function that is minimized is basically a scalar valued function. In most cases, it decomposes as a sum over loss values per detector pixel, which could be displayed in the difference plot. This is not necessarily true in full generality however. My suggestion is to replace the chi2 plot with the true objective function instead.