.. DO NOT EDIT.
.. THIS FILE WAS AUTOMATICALLY GENERATED BY SPHINX-GALLERY.
.. TO MAKE CHANGES, EDIT THE SOURCE PYTHON FILE:
.. "examples/plot_custom_fitter.py"
.. LINE NUMBERS ARE GIVEN BELOW.

.. only:: html

    .. note::
        :class: sphx-glr-download-link-note

        :ref:`Go to the end <sphx_glr_download_examples_plot_custom_fitter.py>`
        to download the full example code.

.. rst-class:: sphx-glr-example-title

.. _sphx_glr_examples_plot_custom_fitter.py:


================================
Using a custom fitter or sampler
================================

How to use your own minimizer or MCMC sampler for fitting light curves.

SNCosmo has three functions for model parameter estimation based on
photometric data: `sncosmo.fit_lc`, `sncosmo.mcmc_lc` and
`sncosmo.nest_lc`. These are wrappers around external minimizers or
samplers (respectively: iminuit, emcee and nestle). However, one may
wish to experiment with a custom fitting or sampling method.

Here, we give a minimal example of using the L-BFGS-B minimizer from scipy.

.. GENERATED FROM PYTHON SOURCE LINES 16-50

.. code-block:: Python


    import numpy as np
    from scipy.optimize import fmin_l_bfgs_b
    import sncosmo

    model = sncosmo.Model(source='salt2')
    data = sncosmo.load_example_data()

    # Define an objective function that we will pass to the minimizer.
    # The function arguments must comply with the expectations of the specfic
    # minimizer you are using.
    def objective(parameters):
        model.parameters[:] = parameters  # set model parameters

        # evaluate model fluxes at times/bandpasses of data
        model_flux = model.bandflux(data['band'], data['time'],
                                    zp=data['zp'], zpsys=data['zpsys'])

        # calculate and return chi^2
        return np.sum(((data['flux'] - model_flux) / data['fluxerr'])**2)

    # starting parameter values in same order as `model.param_names`:
    start_parameters = [0.4, 55098., 1e-5, 0., 0.]  # z, t0, x0, x1, c

    # parameter bounds in same order as `model.param_names`:
    bounds = [(0.3, 0.7), (55080., 55120.), (None, None), (None, None),
              (None, None)]

    parameters, val, info = fmin_l_bfgs_b(objective, start_parameters,
                                          bounds=bounds, approx_grad=True)

    print(parameters)






.. rst-class:: sphx-glr-script-out

 .. code-block:: none

    [ 5.03132699e-01  5.50980000e+04  1.19775631e-05 -2.59687074e-03
      2.22018151e-01]




.. GENERATED FROM PYTHON SOURCE LINES 51-57

The built-in parameter estimation functions in sncosmo take care of
setting up the likelihood function in the way that the underlying
fitter or sampler expects. Additionally, they set guesses and bounds
and package results up in a way that is as consistent as
possible. For users wishing use a custom minimizer or sampler, it
can be instructive to look at the source code for these functions.


.. rst-class:: sphx-glr-timing

   **Total running time of the script:** (0 minutes 0.521 seconds)


.. _sphx_glr_download_examples_plot_custom_fitter.py:

.. only:: html

  .. container:: sphx-glr-footer sphx-glr-footer-example

    .. container:: sphx-glr-download sphx-glr-download-jupyter

      :download:`Download Jupyter notebook: plot_custom_fitter.ipynb <plot_custom_fitter.ipynb>`

    .. container:: sphx-glr-download sphx-glr-download-python

      :download:`Download Python source code: plot_custom_fitter.py <plot_custom_fitter.py>`

    .. container:: sphx-glr-download sphx-glr-download-zip

      :download:`Download zipped: plot_custom_fitter.zip <plot_custom_fitter.zip>`


.. only:: html

 .. rst-class:: sphx-glr-signature

    `Gallery generated by Sphinx-Gallery <https://sphinx-gallery.github.io>`_