ctools

In ctbutterfly::save(), you can write the energies, fluxes, and errors of your spectral fit to a file. But, in a python pipeline with ctools.ctbutterfly, there’s no other way to access the energies/fluxes/errors without saving the results to a file, then reading them back in and parsing them again.

So, I propose 4 new methods for ctbutterfly:

vector<float> ctbutterfly::energies() {
  return m_energies ;
}
vector<float> ctbutterfly::fluxes() {
  return m_fluxes ;
}
vector<float> ctbutterfly::errors() {
  return m_errors ;
}
??? ctbutterfly::covariance_matrix() {
  return m_covariance ; // ???
}

Each return a vector of floats for the (logarithmically center) energies, fluxes, and errors. Then, someone can do:

obs = GObservations()
# load observations into obs

emin = gammalib.GEnergy( 100, 'GeV' )
emax = gammalib.GEnergy( 100, 'TeV' )

spec = ctools.ctbutterfly( obs )
spec['srcname'].string( 'Crab' )
spec['emin'   ].real( emin.TeV() )
spec['emax'   ].real( emax.TeV() )
spec['enumbins'].integer(100)
spec.run()

# yay!
en   = spec.energies()  
flux = spec.fluxes()
errs = spec.errors()
covar = spec.covariance_matrix()

# carry on with storage and plotting

In addition, letting the covariance matrix also be readable in python would be nice too. I’m not certain a GMatrixSparse gets converted to in python. Maybe a full matrix, like a list of lists?