Closed tritemio closed 10 years ago
This is already working on branch fit_experim
. Greatly improved fitting for E and S histograms.
Branch fit_experim is stable. A PR is ready (#18). Will be merged once the notebook in https://github.com/tritemio/FRETBursts_notebooks will be updated.
Branch merged. Closing.
Currently all the fitting are performed using the standard scipy functions.
In particular, 2-Gaussiam mixture fit of the FRET histogram is currently implemented using
scipy.optimize.minimize
. The only methods supporting boundaries (constrains) are SLSQP and L-BFGS-B.Lmfit is a nice pythonic wrapper for non-linear least squares and other methods. Moreover lmfit displays nice fit summaries with several goodness-of-fit statistics.
Using lmfit instead of "home made" functions would make the fitting infrastructure more flexible and easy to use while reducing the maintenance burden.