***********
Convolution
***********

A convolution model requires both the evaluation grid *and* the
data to convolve. Examples include using a point-spread function
(:term:`PSF`) model to modify a two-dimensional model to account
for blurring due to the instrument (or other sources, such as
the atmosphere for ground-based Astronomical data sets),
or the redistribution of the counts as a function of energy
as modelled by the :term:`RMF` when analyzing astronomical
X-ray spectra.

.. note::

   The convolution kernel may require grid points that are not within
   the requested range. The two-dimensional PSF models support this
   when using the Session layer, via the
   :py:func:`sherpa.astro.ui.set_psf` function. It is not yet obvious
   to me how this is handled at this level. The PHA case is handled
   by the :py:class:`sherpa.astro.instrument.RSPModelPHA` class
   (and related classes), but it is still not clear how to document
   and use.