How to use the known position of peaks for peak fitting?

[nicocopez]

I saw that I can use .find_peak_multipeak() on my spectrum to find some peak candidates and then use the result as candidates when I use .fit_peak_multimodel().
However, if I know the (aproximate) position of the peaks that should be in the spectrum, how could I use this knowledge to do peak fitting with .fit_peak_multimodel()? Is there a way?

Thank you in advance for your time,
nico

[georgievgeorgi]

.fit_peak_multimodel() require a parameter candidates of type ListPeakCandidateMultiModel in order to build the model and to calculate the initial guess. An instance of ListPeakCandidateMultiModel can be created manually, but it is much more practical to use .find_peak_multipeak() create it automatically.
There is this example: examples/fit_spe_composite_model.ipynb that describes how it works.
If there is a problem please let me know.

[nicocopez]

Thank you for the quick answer Georgi!
I already tried to use .find_peak_multipeak() (actually following your example), but I am unable to properly catch all the peaks that I need. I will keep on playing with the paremeters. In any case, if it is not too complicated to explain, could you tell me how can I create manually my own ListPeakCandidateMultiModel ? Alternatively, it would also be very useful if one can choose the range in which the peak should be found.

[georgievgeorgi]

This example plots manually created candidates.

from ramanchada2.misc.types import ListPeakCandidateMultiModel
cand = ListPeakCandidateMultiModel.validate([
    {'base_slope': -.1, 'base_intersept': 100, 'boundaries': [0, 300], 'peaks': [
        {'amplitude': 100, 'position': 50, 'sigma': 30, 'skew': 0},
        {'amplitude': 10, 'position': 150, 'sigma': 30, 'skew': .5},
        {'amplitude': 200, 'position': 250, 'sigma': 30, 'skew': -.5},
    ], },
    {'base_slope': .1, 'base_intersept': -50, 'boundaries': [500, 800], 'peaks': [
        {'amplitude': 100, 'position':550, 'sigma': 30, 'skew': 0},
        {'amplitude': 10, 'position': 650, 'sigma': 30, 'skew': .5},
        {'amplitude': 200, 'position': 700, 'sigma': 30, 'skew': -.5},
    ], },
])
cand.plot()

If you are interested in the candidates only in a specified range, you could do something like:

cand = spe.trim_axes('x-axis', (580, 600)).find_peak_multipeak()
spe.fit_peak_multimodel(candidates=cand)

[georgievgeorgi]

If needed we can have sum operation in context of candidates. For example:

cand_auto = spe.find_peak_multipeak(...)
cand_manual = ListPekCandidateMultiModel.validate(...)
cand_all = cand_auto + cand_manual
fit_res = spe.fit_peak_multimodel(candidates=cand_all, ...)