def test_compute_fluxcalibration(self): """ Test compute_fluxcalibration interface """ #get frame data frame = get_frame_data() #get model data modelwave, modelflux = get_models() # pick std star fibers stdfibers = np.random.choice(9, 3, replace=False) # take 3 std stars fibers frame.fibermap['DESI_TARGET'][stdfibers] = desi_mask.STD_FAINT input_model_wave = modelwave input_model_flux = modelflux[ 0: 3] # assuming the first three to be best models,3 is exclusive here fluxCalib = compute_flux_calibration(frame, input_model_wave, input_model_flux, input_model_fibers=stdfibers, nsig_clipping=4.) # assert the output self.assertTrue(np.array_equal(fluxCalib.wave, frame.wave)) self.assertEqual(fluxCalib.calib.shape, frame.flux.shape) #- nothing should be masked for this test case self.assertFalse(np.any(fluxCalib.mask))
def test_outliers(self): '''Test fluxcalib when input starts with large outliers''' frame = get_frame_data() modelwave, modelflux = get_models() nstd = 5 frame.fibermap['OBJTYPE'][0:nstd] = 'STD' nstd = np.count_nonzero(frame.fibermap['OBJTYPE'] == 'STD') frame.flux[0] = np.mean(frame.flux[0]) fluxCalib = compute_flux_calibration(frame, modelwave, modelflux[0:nstd],input_model_fibers=np.arange(nstd))
def test_outliers(self): '''Test fluxcalib when input starts with large outliers''' frame = get_frame_data() modelwave, modelflux = get_models() nstd = 5 frame.fibermap['OBJTYPE'][0:nstd] = 'STD' nstd = np.count_nonzero(frame.fibermap['OBJTYPE'] == 'STD') frame.flux[0] = np.mean(frame.flux[0]) fluxCalib = compute_flux_calibration(frame, modelwave, modelflux[0:nstd],input_model_fibers=np.arange(nstd))
def test_masked_data(self): """Test compute_fluxcalibration with some ivar=0 data """ frame = get_frame_data() modelwave, modelflux = get_models() nstd = 1 frame.fibermap['OBJTYPE'][2:2+nstd] = 'STD' frame.ivar[2:2+nstd, 20:22] = 0 fluxCalib = compute_flux_calibration(frame, modelwave, modelflux[2:2+nstd], input_model_fibers=np.arange(2,2+nstd), debug=True) self.assertTrue(np.array_equal(fluxCalib.wave, frame.wave)) self.assertEqual(fluxCalib.calib.shape,frame.flux.shape)
def test_masked_data(self): """Test compute_fluxcalibration with some ivar=0 data """ frame = get_frame_data() modelwave, modelflux = get_models() nstd = 1 frame.fibermap['OBJTYPE'][2:2+nstd] = 'STD' frame.ivar[2:2+nstd, 20:22] = 0 fluxCalib = compute_flux_calibration(frame, modelwave, modelflux[2:2+nstd], input_model_fibers=np.arange(2,2+nstd), debug=True) self.assertTrue(np.array_equal(fluxCalib.wave, frame.wave)) self.assertEqual(fluxCalib.calib.shape,frame.flux.shape)
def _write_models(self): wav, mods = get_models(wavemin=2900, wavemax=11000) nmod = len(mods) tid, logg, feh, teff = [np.arange(nmod) for _ in range(4)] tab = Table({ 'TEMPLATEID': tid, 'LOGG': logg, 'FEH': feh, 'TEFF': teff }) fits.HDUList( [fits.PrimaryHDU(mods), fits.BinTableHDU(tab), fits.ImageHDU(wav)]).writeto(self.modelfile, overwrite=True)
def test_match_templates(self): """ Test with simple interface check for matching best templates with the std star flux """ from desispec.fluxcalibration import match_templates frame = get_frame_data() # first define dictionaries flux = {"b": frame.flux, "r": frame.flux * 1.1, "z": frame.flux * 1.2} wave = {"b": frame.wave, "r": frame.wave + 10, "z": frame.wave + 20} ivar = {"b": frame.ivar, "r": frame.ivar / 1.1, "z": frame.ivar / 1.2} # resol_data={"b":np.mean(frame.resolution_data,axis=0),"r":np.mean(frame.resolution_data,axis=0),"z":np.mean(frame.resolution_data,axis=0)} resol_data = { "b": frame.resolution_data, "r": frame.resolution_data, "z": frame.resolution_data } #model nmodels = 10 modelwave, modelflux = get_models(nmodels) teff = np.random.uniform(5000, 7000, nmodels) logg = np.random.uniform(4.0, 5.0, nmodels) feh = np.random.uniform(-2.5, -0.5, nmodels) # say there are 3 stdstars stdfibers = np.random.choice(9, 3, replace=False) frame.fibermap['OBJTYPE'][stdfibers] = 'STD' #pick fluxes etc for each stdstars find the best match bestid = -np.ones(len(stdfibers)) bestwave = np.zeros((bestid.shape[0], modelflux.shape[1])) bestflux = np.zeros((bestid.shape[0], modelflux.shape[1])) red_chisq = np.zeros(len(stdfibers)) for i in range(len(stdfibers)): stdflux = {"b": flux["b"][i], "r": flux["r"][i], "z": flux["z"][i]} stdivar = {"b": ivar["b"][i], "r": ivar["r"][i], "z": ivar["z"][i]} stdresol_data = { "b": resol_data["b"][i], "r": resol_data["r"][i], "z": resol_data["z"][i] } bestid, redshift, chi2 = \ match_templates(wave, stdflux, stdivar, stdresol_data, modelwave, modelflux, teff, logg, feh)
def test_compute_fluxcalibration(self): """ Test compute_fluxcalibration interface """ #get frame data frame=get_frame_data() #get model data modelwave,modelflux=get_models() # pick std star fibers stdfibers=np.random.choice(9,3,replace=False) # take 3 std stars fibers frame.fibermap['DESI_TARGET'][stdfibers] = desi_mask.STD_FAINT input_model_wave=modelwave input_model_flux=modelflux[0:3] # assuming the first three to be best models,3 is exclusive here fluxCalib =compute_flux_calibration(frame, input_model_wave,input_model_flux,input_model_fibers=stdfibers,nsig_clipping=4.) # assert the output self.assertTrue(np.array_equal(fluxCalib.wave, frame.wave)) self.assertEqual(fluxCalib.calib.shape,frame.flux.shape) #- nothing should be masked for this test case self.assertFalse(np.any(fluxCalib.mask))
def test_match_templates(self): """ Test with simple interface check for matching best templates with the std star flux """ from desispec.fluxcalibration import match_templates frame=get_frame_data() # first define dictionaries flux={"b":frame.flux,"r":frame.flux*1.1,"z":frame.flux*1.2} wave={"b":frame.wave,"r":frame.wave+10,"z":frame.wave+20} ivar={"b":frame.ivar,"r":frame.ivar/1.1,"z":frame.ivar/1.2} # resol_data={"b":np.mean(frame.resolution_data,axis=0),"r":np.mean(frame.resolution_data,axis=0),"z":np.mean(frame.resolution_data,axis=0)} resol_data={"b":frame.resolution_data,"r":frame.resolution_data,"z":frame.resolution_data} #model nmodels = 10 modelwave,modelflux=get_models(nmodels) teff = np.random.uniform(5000, 7000, nmodels) logg = np.random.uniform(4.0, 5.0, nmodels) feh = np.random.uniform(-2.5, -0.5, nmodels) # say there are 3 stdstars stdfibers=np.random.choice(9,3,replace=False) frame.fibermap['OBJTYPE'][stdfibers] = 'STD' #pick fluxes etc for each stdstars find the best match bestid=-np.ones(len(stdfibers)) bestwave=np.zeros((bestid.shape[0],modelflux.shape[1])) bestflux=np.zeros((bestid.shape[0],modelflux.shape[1])) red_chisq=np.zeros(len(stdfibers)) for i in range(len(stdfibers)): stdflux={"b":flux["b"][i],"r":flux["r"][i],"z":flux["z"][i]} stdivar={"b":ivar["b"][i],"r":ivar["r"][i],"z":ivar["z"][i]} stdresol_data={"b":resol_data["b"][i],"r":resol_data["r"][i],"z":resol_data["z"][i]} bestid, redshift, chi2 = \ match_templates(wave, stdflux, stdivar, stdresol_data, modelwave, modelflux, teff, logg, feh)