problem with IC integrator in python

[Daniele Gaggero]

Hi Andrej, I have a problem with the IC integrator, only in the python interface

I am reading leptons from a FITS file like this:
dragon2D_lepton = cosmicrays.Dragon2D(filename, [Electron])

and then, when I initialize the integrator like this:
integratorIC_cmb = InverseComptonIntegrator(dragon2D_lepton, CMB_photons, kn_crosssection)

I get the error

TypeError: init(): incompatible constructor arguments. The following argument types are supported:
1. pyhermes.InverseComptonIntegrator(arg0: hermes::cosmicrays::CosmicRayDensity, arg1: hermes::photonfields::PhotonField, arg2: hermes::interactions::DifferentialCrossSection)

Invoked with: <pyhermes.cosmicrays.Dragon2D object at 0x15ae3b9c70>, ('cmb', <pyhermes.photonfields.CMB object at 0x15ae3e36f8>), <pyhermes.interactions.KleinNishina object at 0x15ae3e32d0>

The problem is not present in the pi0 integrator.
Also, it is not present in the C++ version.

We noticed that, in the file "integrators.cpp" enclosed in the "python/" folder,

the pi0 integrator has this initialization:

py::init<
const std::vector<std::shared_ptrcosmicrays::CosmicRayDensity>,

while the IC does not have the vector initialization. Can it be part of the problem?
Thanks
Daniele

[Andrej Dundovic]

Hi Daniele,

thanks for the detailed report. It is true that the IC integrator, unlike Pi0, does not accept more than one CR object (provided as a python list), but it should accept one just fine (as you wrote). I tested the following python script and it runs:

https://github.com/cosmicrays/hermes-examples/blob/master/python/e3-ic-dragon2d.py

Hence, your example should also work if all input objects are the same. Can you compare your script with it?

Regarding passing many CR objects to IC - it can be accomplished by implementing additional "init" with vector (like in Pi0), but I remember vaguely that we said it is not required in case of leptons... If there is a case in which we put different CRs as a listo to the IC integrator, we'll easily extend it.

[Carmelo Evoli]

Yes, if that works it should suffices the scope.

[Daniele Gaggero]

Hi Andrej, the example works.

I found out the problem: I have to remember to set

name1, CMB_photons = ('cmb', photonfields.CMB())
name2, ISRF_photons = ('isrf', photonfields.ISRF())

before passing CMB_photons to the integrator, since ('cmb', photonfields.CMB()) returns a pair of objects

[Andrej Dundovic]

Hi Daniele,

yes, but that is a matter of taste. I did it to keep photon field names for later - to put them in file names for this specific example... but you can also just write without them:

CMB_photons = photonfields.CMB()
ISRF_photons = photonfields.ISRF()

[Daniele Gaggero]

sure, yes this is cleaner indeed :)