# We set the maximum allowed difference in the number of fitted components between individual neighboring spectra to 2
sp.max_jump_comps = 2
# We will flag and try to refit all spectra which show jumps in the number of components of more than 2 to at least two direct neighbors
sp.n_max_jump_comps = 1
# Maximum difference in offset positions of fit components for grouping.
sp.mean_separation = 2.
# Maximum difference in FWHM values of fit components for grouping.
sp.fwhm_separation = 4.
# Start phase 1 of the spatially coherent refitting
sp.spatial_fitting()
# (Optional) Plot maps of the reduced chi-square values and the number of fitted components
# Initialize the 'GaussPyDecompose' class and read in the parameter settings from 'gausspy+.ini'.
decompose = GaussPyDecompose(config_file='gausspy+.ini')
# Filepath to pickled dictionary of the prepared data.
decompose.path_to_pickle_file = sp.path_to_pickle_file
# Filepath to the pickled dictionary with the decomposition results
path_to_decomp_pickle = os.path.join(
'decomposition_grs', 'gpy_decomposed',
'grs-test_field_g+_fit_fin_sf-p1.pickle')
# Load the decomposition results
decompose.load_final_results(path_to_decomp_pickle)
# Produce a FITS image showing the number of fitted components
decompose.produce_component_map()
# Produce a FITS image showing the reduced chi-square values
decompose.produce_rchi2_map()
# (Optional) Plot some of the spectra and the decomposition results
filenames = glob.glob(
"decomposition/gpy_decomposed/smc_HI_cube_askap_sub_*_g+_fit_fin.pickle")
fileprefs = [f.split("_g+_fit")[0] for f in filenames]
fileprefs = [f.split("sub_")[1] for f in fileprefs]
donenames = glob.glob(
"decomposition/gpy_maps/smc_HI_cube_askap_sub_*_g+_component_map_MW.fits")
doneprefs = [f.split("_g+_fit")[0] for f in donenames]
doneprefs = [f.split("sub_")[1] for f in doneprefs]
prefs = [f for f in fileprefs if f not in doneprefs]
for i, filestr in enumerate(prefs):
# Initialize the 'GaussPyDecompose' class and read in the parameter settings from 'gausspy+.ini'.
decompose = GaussPyDecompose(config_file="gausspy+.ini")
# Filepath to pickled dictionary of the prepared data.
decompose.path_to_pickle_file = os.path.join(
'decomposition', 'gpy_prepared',
"smc_HI_cube_askap_sub_" + filestr + '.pickle')
#
# # # Filepath to pickled dictionary of the prepared data.
path_to_pickled_file = decompose.path_to_pickle_file
# Filepath to pickled dictionary with the decomposition results
path_to_decomp_pickle = os.path.join(
"decomposition",
"gpy_decomposed",
"smc_HI_cube_askap_sub_" + filestr + "_g+_fit_fin.pickle",
)
print(path_to_decomp_pickle)
if (filename !=
'decomposition/gpy_prepared/smc_HI_cube_askap_sub_40.pickle'
) and (
filename !=
'decomposition/gpy_prepared/smc_HI_cube_askap_sub_16.pickle') and (
filename !=
'decomposition/gpy_prepared/smc_HI_cube_askap_sub_24.pickle'
) and (
filename !=
'decomposition/gpy_prepared/smc_HI_cube_askap_sub_32.pickle'):
filestr = filename.split(".pickle")[0]
filestr = filestr.split("sub_")[1]
# Filepath to pickled dictionary of the prepared data.
# Initialize the 'GaussPyDecompose' class and read in the parameter settings from 'gausspy+.ini'.
decompose = GaussPyDecompose(config_file="gausspy+.ini")
decompose.path_to_pickle_file = filename
# First smoothing parameter
decompose.alpha1 = 1.13
# Second smoothing parameter
decompose.alpha2 = 2.06
# We set the upper limit for the reduced chi-square deliberately to a low value to enforce the best fitting results for each individual spectrum.
decompose.rchi2_limit = 2.0
# Suffix for the filename of the pickled dictionary with the decomposition results.
decompose.suffix = "_g+"
# Start the decomposition.
decompose.decompose()
# (Optional) Produce a FITS image showing the number of fitted components
decompose.produce_component_map()
def main():
# Initialize the 'GaussPyDecompose' class and read in the parameter settings from 'gausspy+.ini'.
decompose = GaussPyDecompose(config_file='gausspy+.ini')
# The following lines will override the corresponding parameter settings defined in 'gausspy+.ini'.
# Filepath to pickled dictionary of the prepared data.
decompose.path_to_pickle_file = os.path.join('decomposition_grs',
'gpy_prepared',
'grs-test_field.pickle')
# First smoothing parameter
decompose.alpha1 = 2.58
# Second smoothing parameter
decompose.alpha2 = 5.14
# Suffix for the filename of the pickled dictionary with the decomposition results.
decompose.suffix = '_g+'
# Start the decomposition.
decompose.decompose()
# (Optional) Produce a FITS image showing the number of fitted components
decompose.produce_component_map()
# (Optional) Produce a FITS image showing the reduced chi-square values
decompose.produce_rchi2_map()
# (Optional) Plot some of the spectra and the decomposition results
# Filepath to pickled dictionary of the prepared data.
path_to_pickled_file = decompose.path_to_pickle_file
# Filepath to pickled dictionary with the decomposition results
path_to_decomp_pickle = os.path.join('decomposition_grs', 'gpy_decomposed',
'grs-test_field_g+_fit_fin.pickle')
# Directory in which the plots are saved.
path_to_plots = os.path.join('decomposition_grs', 'gpy_plots')
# Here we select a subregion of the data cube, whose spectra we want to plot.
pixel_range = {'x': [30, 34], 'y': [25, 29]}
plot_spectra(path_to_pickled_file,
path_to_plots=path_to_plots,
path_to_decomp_pickle=path_to_decomp_pickle,
signal_ranges=True,
pixel_range=pixel_range)
# @Author: riener
# @Date: 2019-04-02T17:42:46+02:00
# @Filename: decompose--grs.py
# @Last modified by: riener
# @Last modified time: 31-05-2019
import os
from gausspyplus.decompose import GaussPyDecompose
from gausspyplus.plotting import plot_spectra
# Initialize the 'GaussPyDecompose' class and read in the parameter settings from 'gausspy+.ini'.
decompose = GaussPyDecompose(config_file='gausspy+.ini')
# The following lines will override the corresponding parameter settings defined in 'gausspy+.ini'.
# Filepath to pickled dictionary of the prepared data.
decompose.path_to_pickle_file = os.path.join('decomposition_grs',
'gpy_prepared',
'grs-test_field.pickle')
# First smoothing parameter
decompose.alpha1 = 2.58
# Second smoothing parameter
decompose.alpha2 = 5.14
# Suffix for the filename of the pickled dictionary with the decomposition results.
decompose.suffix = '_g+'
# Start the decomposition.
decompose.decompose()
# (Optional) Produce a FITS image showing the number of fitted components
decompose.produce_component_map()
def main():
# Initialize the 'SpatialFitting' class and read in the parameter settings from 'gausspy+.ini'.
sp = SpatialFitting(config_file='gausspy+.ini')
# The following lines will override the corresponding parameter settings defined in 'gausspy+.ini'.
# filepath to the pickled dictionary of the prepared data
sp.path_to_pickle_file = os.path.join('decomposition_grs', 'gpy_prepared',
'grs-test_field.pickle')
# Filepath to the pickled dictionary of the decomposition results
sp.path_to_decomp_file = os.path.join('decomposition_grs',
'gpy_decomposed',
'grs-test_field_g+_fit_fin.pickle')
# Try to refit blended fit components
sp.refit_blended = True
# Try to refit spectra with negative residual features
sp.refit_neg_res_peak = True
# Try to refit broad fit components
sp.refit_broad = True
# Flag spectra with non-Gaussian distributed residuals
sp.flag_residual = True
# Do not try to refit spectra with non-Gaussian distributed residuals
sp.refit_residual = False
# Try to refit spectra for which the number of fit components is incompatible with its direct neighbors
sp.refit_ncomps = True
# We set the maximum allowed difference in the number of fitted components compared to the weighted median of all immediate neighbors to 1
sp.max_diff_comps = 1
# We set the maximum allowed difference in the number of fitted components between individual neighboring spectra to 2
sp.max_jump_comps = 2
# We will flag and try to refit all spectra which show jumps in the number of components of more than 2 to at least two direct neighbors
sp.n_max_jump_comps = 1
# Maximum difference in offset positions of fit components for grouping.
sp.mean_separation = 2.
# Maximum difference in FWHM values of fit components for grouping.
sp.fwhm_separation = 4.
# Start phase 1 of the spatially coherent refitting
sp.spatial_fitting()
# (Optional) Plot maps of the reduced chi-square values and the number of fitted components
# Initialize the 'GaussPyDecompose' class and read in the parameter settings from 'gausspy+.ini'.
decompose = GaussPyDecompose(config_file='gausspy+.ini')
# Filepath to pickled dictionary of the prepared data.
decompose.path_to_pickle_file = sp.path_to_pickle_file
# Filepath to the pickled dictionary with the decomposition results
path_to_decomp_pickle = os.path.join(
'decomposition_grs', 'gpy_decomposed',
'grs-test_field_g+_fit_fin_sf-p1.pickle')
# Load the decomposition results
decompose.load_final_results(path_to_decomp_pickle)
# Produce a FITS image showing the number of fitted components
decompose.produce_component_map()
# Produce a FITS image showing the reduced chi-square values
decompose.produce_rchi2_map()
# (Optional) Plot some of the spectra and the decomposition results
# Filepath to pickled dictionary of the prepared data.
path_to_pickled_file = sp.path_to_pickle_file
# Directory in which the plots are saved.
path_to_plots = os.path.join('decomposition_grs', 'gpy_plots')
# Here we select a subregion of the data cube, whose spectra we want to plot.
pixel_range = {'x': [30, 34], 'y': [25, 29]}
plot_spectra(path_to_pickled_file,
path_to_plots=path_to_plots,
path_to_decomp_pickle=path_to_decomp_pickle,
signal_ranges=True,
pixel_range=pixel_range)
def gpp_parameters(self):
# This code creates a temporary fits file of the data called "gpp-temp.fits" that GaussPy+ will use to decompose.
self.y_data = np.reshape(self.y_data, (self.y_data.shape[0], 1, 1))
self.y_data.shape
hdu = fits.PrimaryHDU(self.y_data)
CRVAL3 = self.x_data[0]
CDELT3 = self.x_data[1] - self.x_data[0]
hdu1 = fits.HDUList([hdu])
hdu1.writeto("gpp-temp.fits", overwrite=True)
# This is taken directly from the GaussPy+ documentation, with the signal-to-noise ratio, minimum FWHM, and
# the gpp-temp.fits file used.
prepare = GaussPyPrepare()
prepare.path_to_file = os.path.abspath("gpp-temp.fits")
prepare.p_limit = 0.02
prepare.pad_channels = 2
prepare.signal_mask = True
prepare.min_channels = 100
prepare.mask_out_ranges = []
prepare.snr = self.sig
prepare.significance = 5.0
prepare.snr_noise_spike = self.sig
data_location = (0, 0)
prepared_spectrum = prepare.return_single_prepared_spectrum(
data_location)
decompose = GaussPyDecompose()
decompose.two_phase_decomposition = True
decompose.alpha1 = 2.58
decompose.alpha2 = 5.14
decompose.improve_fitting = True
decompose.exclude_mean_outside_channel_range = True
decompose.min_fwhm = self.min
decompose.max_fwhm = 64.
decompose.snr = self.sig
decompose.snr_fit = None
decompose.significance = 3.0
decompose.snr_negative = None
decompose.min_pvalue = 0.01
decompose.max_amp_factor = 1.1
decompose.refit_neg_res_peak = True
decompose.refit_broad = True
decompose.refit_blended = True
decompose.separation_factor = 0.8493218
decompose.fwhm_factor = 2.
decompose.single_prepared_spectrum = prepared_spectrum
decomposed_test = decompose.decompose()
# This stores the parameters of each gaussian component into a single array, in hindsight
# I should have made an array of arrays (each one containing the parameters of each component).
self.params = np.concatenate(
(np.array(decomposed_test["amplitudes_fit"][0]),
(np.array(decomposed_test["fwhms_fit"][0]) * CDELT3),
((np.array(decomposed_test["means_fit"][0]) * CDELT3) + CRVAL3)))
# This is the number of components that the spectrum has been decomposed into.
self.components = int(len(self.params) / 3.0)