simul_mle_fit

nbi_stat.simul_mle_fit(regions, p0, *args, **kwargs)[source]

Perform simulatinous MLE fit over several regions

This will fit a combined function to data in several regions.

Each region has it’s own data and it’s own model function. The kind of MLE to do in each region can also be customized.

Parameters:

  • regions (sequence of containers) –

    A sequence of regions. Each region is specified as

    dataarray-like, (array-like,array-like)

    Data for the region (either observations, or a binned data)

    funccallable

    Function to model the data in the region. Note, all functions receive all parameters (except extended overall scaling). It is up to the user to extract the needed parameters for a given region

    kwdict (optional)

    Additional keyword arguments to pass to the logarithmic likelihood function (either binned_llh or llh). These update the general keywords passed to simul_mle_fit

  • p0 (array-like) –

    Initial parameters. This must be _all_ parameters used in the fit. Extended scale parameters must come first in the container.

    Note, all functions in all regions receive _all_ parameters (except the extended scale parameters), and it is up to the user to filter out hte relevant parameters for a given region.

  • *args (tuple) – Additional arguments

  • **kwargs (dict) –

    Keyword arguments

    extendedbool

    Perform an extended MLE

    logpdfbool

    If the functions are logarithmic PDFs pass True for this

    normalizedbool

    If we’re doing extended fits, and the PDFs are not normalised pass False for this. Has no effect for Poisson binned fits.

    xtracallable

    Extra stuff to add to logarithmic PDF

    densitybool, int, float

    For binned likelihood fits.

    cdfbool

    For binned likelihood fits

    poissonbool

    For binned likelihood fits.

    raw_narray-like

    Cached calculation of raw count equivalent

    log_Gamma_Nnfloat

    Cached calculation of binned corrections

    Other arguments are passed to scipy.optimize.minimize

Returns:

  • p (array-like) – Found parameter values (including possibly extended normalisations)

  • cov (array-like) – Covariance of parameters

  • opt (OptimizeResult (optional)) – If full_output=True is passed, also get full result from minimize.

See also

mle_fit, llh, binned_llh, plot_fit, plot_nsigma_contour, fit