Only chi2 regularization works

[ljlamarche]

Currently, only the chi2 regularization method is actually functional although there are several other regularization functions coded into Fit.py. Alternative methods for determining the regularization parameter, such as GCV, should be more thoroughly investigated, and either added as functional option or removed from the source code.

[ljlamarche]

There's also some evidence that chi2 regularization doesn't even work that well. If different points are filtered out of the same input density measurements, there are noticeable differences in the interpolation. In the example below, all times use the same input synthetic data set, but the chi2 array hasn't been overwritten so different points are removed as "bad data" from each time. Because a relatively low number of points are expected to be filtered out this way, we don't expect to see differences in the pattern in each column.