Bugs: delta2d

Bugs

The delta2d model parameters remain unchanged in a fit to 2D non-integrated data

If the delta2d model is used to fit a non-integrated data set, with an optimization method other than Nelder-Mead, the xpos and ypos model parameters will not vary in the fit. Users can work around this issue by fitting the model to 2D integrated data sets, e.g.:

sherpa> load_data("img.ccd7.0.3-7_box_0.2bin.fits", dstype=DataIMGInt)

# instead of

sherpa> load_data("img.ccd7.0.3-7_box_0.2bin.fits")

The dataspace2d, load_data, unpack_data, unpack_image, and load_image commands accept this 'dstype' specification to determine the data set type.

A 2D integrated data set can be specified using dataspace2d with:

sherpa> dataspace2d([100,100], dstype=DataIMGInt)

# or

sherpa> dataspace2d([100,100], dstype=Data2DInt)

A work-around to force the delta2d model to be integrated from a non-integrated grid is provided in the example Python script below.

from sherpa.astro.ui import *

from sherpa.models import Delta2D

class MyDelta2D(Delta2D):
    """
    Force the calculation of Delta2D to be integrated from
    a non-integrated grid
    """
    def calc(self, p, x0, x1, *args, **kwargs):
        # NOTE: image coordinates only!
        return Delta2D.calc(self, p, x0-0.5, x1-0.5, x0+0.5, x1+0.5,
                            *args, **kwargs)


dataspace2d([100,100])

set_source(const2d.c2)
fake()

set_method("neldermead")
set_stat("cash")

load_psf("psf2d",gauss2d.g2)
set_psf(psf2d)
g2.xpos=30.
g2.ypos=70.
freeze(g2)

add_model(MyDelta2D)

set_source(mydelta2d.sig+const2d.bkg)
sig.xpos=50.
sig.ypos=50.
sig.ampl=1000.
fake()
freeze(bkg)
fit()

# Here the g2 model is also being fit, while it is a definition of the
# psf model and it should be kept frozen.

fit()
sig.xpos=49.
sig.ypos=51.

fit()