Bootstrapping doesn't seem to work when I have a defined model for fitting

[MagnaHyperion]

Hello!

I was recently fitting some data and ran into an issue when I tried to run fitting with bootstrapping when I have a defined model, specifically, I was using the model dd_gauss2. Here is the code:

`import numpy as np
import matplotlib.pyplot as plt
import deerlab as dl

file =

tmin = 0.14
tau1 = 0.12
tau2 = 3.000

t,Vexp = dl.deerload(file)
t = t[:-2]
Vexp = Vexp[:-2]
Vexp = dl.correctphase(Vexp)
Vexp = Vexp/max(Vexp)
t = t- t[0] + tmin

r = np.arange(2,6,0.05)

Pmodel= dl.dd_gauss2
Vmodel = dl.dipolarmodel(t, r, Pmodel, experiment=dl.ex_4pdeer(tau1, tau2, pathways=[1], pulselength=0.04))
compactness = dl.dipolarpenalty(Pmodel, r=r, type='compactness')

results = dl.fit(Vmodel,Vexp,penalties=compactness,bootstrap=1000,bootcores=3)

print(results)`

Running this code gives the following error after a few minutes:

However, when I simply do 'Pmodel = None', the script works fine. Is there some reason I'm missing why you can't do bootstrapping with a defined Pmodel?

[HKaras]

Hi,

My suspicion that this caused by a poorly defined starting parameter and bounds in the dl.dd_guass model. What you could do is do an initial fit and then use this as starting parameters for a fit with bootstrapped averages. You might also want to restrict the boundaries.

FYI: we fixed an issue with time-domain bootstrapped uncertainties which have not been included in a PyPi release yet. So you might want to consider install DeerLab from source to benefit from the bug fix. https://jeschkelab.github.io/DeerLab/installation.html#installing-from-source

[MagnaHyperion]

Thanks for the swift reply! Will try these fixes out now.