Closed kabeleh closed 3 years ago
I realised that the results for the matter power specturm P(k) and cl(lensed) are either very high, 0.00000 or -nan. CLASS itself did not complain about this, but MontePython aborted. So it is rather an issue with my own code and I apologise for opening this thread. I didn't realise this earlier, because I wasn't aware of the fact that MontePython automatically askes for 'output': 'mPk tCl lCl pCl '
, even if it is not specified by the user.
I can now happily close this issue. It was not actually an issue with my code per se. But the perturbed equations of motion were in a way that weren't handled well by the implicit solver. Setting evolver=0
in the *.ini-file solved my problem. The spectra look good now and MontePython is running without further hiccups (so far).
I am trying to run a modified version of CLASS 2.9.4 together with MontePython. Just calling CLASS works fine, but MontePython fails. Now I am wondering, if there might be some changes required to the python wrapper? I've seen similar issues already around (#390 , MP#183, MP#198) yet no viable solution. I am on python 3.9.3 and I am using CLASS 2.9.4 with modified input.c, background.c, and perturbations.c (with corresponding changes in background.h and perturbations.h, as well as classy.c, classy.pyx, and cclassy.pyxd). All changes are related to the scalar field or dark matter. Everything runs fine with the default perturbations, but not with the modified perturbations. CLASS itself runs with the modified perturbations, but when I try using it in MontePython, it throws between 3 and 10 times the following error message (with different
interval: [X:Y]
values)before it aborts with
I played around with the parameter ranges in MontePython and I also tried fixing the scalar field parameters in MontePython to values that work with CLASS when just calling CLASS through an *.ini.file, i.e. setting those to constant ones in the MontePython parameter file. I also tried including
data.cosmo_arguments['lensing'] = 'yes'
anddata.cosmo_arguments['output'] = 'tCl,pCl,lCl,mPk'
in MontePython without success.Is there anything else I could try or something I should check? Do I need to make some specific changes to the python wrapper, when changing something in perturbations.c?
Thank you very much in advance. Any help is highly appreciated.