BUG: photometry fails when several sources are in the same group

I’m running into a bug where the photometry routine tries to access a model attribute that does not exist. As far as I can tell, that’s happening when more than one source is found in the same group, but the naming scheme of the parameters of the combined model is confusing to me, so I don’t really know what’s going on.

This happens with the current checkout of the master branch and astropy 3.1.2 as well as the current astropy master. A minimal example is below:

(photutils)melkor:/melkor/d1/guenther/soft/python> ipython
Python 3.7.3 (default, Mar 27 2019, 22:11:17) 
Type 'copyright', 'credits' or 'license' for more information
IPython 7.5.0 -- An enhanced Interactive Python. Type '?' for help.

In [1]: import numpy as np 
   ...: from astropy.table import Table 
   ...: from astropy.modeling import models 
   ...: from photutils.psf import prepare_psf_model 
   ...: from photutils import BasicPSFPhotometry, DAOGroup 
   ...:  
   ...: psffunc = models.Moffat2D(amplitude=1, gamma=2, alpha=2.8, x_0=0, y_0=0)
   ...:  
   ...: psffunc.gamma.frozen = True 
   ...: psffunc.alpha.frozen = True 
   ...: psffunc.amplitude.frozen = True  # Fix to one and add scaling factor to 
   ...: avoid confusion between amplitude and flux 
   ...:      
   ...: psfmodel = prepare_psf_model(psffunc, xname='x_0', yname='y_0', fluxname
   ...: =None) 
   ...: photometry = BasicPSFPhotometry(group_maker=DAOGroup(5), bkg_estimator=N
   ...: one,  
   ...:                                 psf_model=psfmodel, fitshape=(11,11),  
   ...:                                 aperture_radius=5) 
   ...: photometry.do_photometry(np.zeros((100, 100)), init_guesses=Table({'x_0'
   ...: : [10, 11, 12], 'y_0': [10, 11, 12]}))                                  
---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-1-c99a4ea837ff> in <module>
     14                                 psf_model=psfmodel, fitshape=(11,11),
     15                                 aperture_radius=5)
---> 16 photometry.do_photometry(np.zeros((100, 100)), init_guesses=Table({'x_0': [10, 11, 12], 'y_0': [10, 11, 12]}))

/nfs/melkor/d1/guenther/soft/anaconda/envs/photutils/lib/python3.7/site-packages/photutils-0.7.dev3439-py3.7-linux-x86_64.egg/photutils/psf/photometry.py in do_photometry(self, image, init_guesses)
    289 
    290         star_groups = self.group_maker(init_guesses)
--> 291         output_tab, self._residual_image = self.nstar(image, star_groups)
    292 
    293         star_groups = star_groups.group_by('group_id')

/nfs/melkor/d1/guenther/soft/anaconda/envs/photutils/lib/python3.7/site-packages/photutils-0.7.dev3439-py3.7-linux-x86_64.egg/photutils/psf/photometry.py in nstar(self, image, star_groups)
    351                                     image[usepixel])
    352             param_table = self._model_params2table(fit_model,
--> 353                                                    len(star_groups.groups[n]))
    354             result_tab = vstack([result_tab, param_table])
    355 

/nfs/melkor/d1/guenther/soft/anaconda/envs/photutils/lib/python3.7/site-packages/photutils-0.7.dev3439-py3.7-linux-x86_64.egg/photutils/psf/photometry.py in _model_params2table(self, fit_model, star_group_size)
    466                     param_tab[param_tab_name][i] = getattr(fit_model,
    467                                                            param_name +
--> 468                                                            '_' + str(i)).value
    469         else:
    470             for param_tab_name, param_name in self._pars_to_output.items():

/nfs/melkor/d1/guenther/soft/anaconda/envs/photutils/lib/python3.7/site-packages/astropy/modeling/core.py in __getattr__(self, attr)
   2881         # This workaround allows descriptors to work correctly when they are
   2882         # not initially found in the class __dict__
-> 2883         value = getattr(self.__class__, attr)
   2884         if hasattr(value, '__get__'):
   2885             # Object is a descriptor, so we should really return the result of

/nfs/melkor/d1/guenther/soft/anaconda/envs/photutils/lib/python3.7/site-packages/astropy/modeling/core.py in __getattr__(cls, attr)
   2273             return getattr(cls, attr)
   2274 
-> 2275         raise AttributeError(attr)
   2276 
   2277     def __repr__(cls):

AttributeError: x_0_2_0

[hamogu]

Mmmm. There must be a better way to map (compound) parameters in the psf model to the parameters we want to fit. Looking a the string of the parameter name is obviously brittle. However, that’s more than a little fix and I don’t have time to work on that now, unfortunately.

Anyway, I looked the #558 and did not realize that this is a dublicate, but with your links I now realize that, so I’ll close this issue.

[larrybradley]

Fixed in #1135.