Investigating the DT SED computation I realized that this part is quite problematic. If you compute DT SED for nu that are not sorted (e.g. if you generate lambdas and convert them to nu) the value will be wrong (I got negative). Also if the calculations are done only at course frequencies (or not covering the whole emission) this renormalization can screw up the integral emission. BB emission can be integrated analitically, so it would be better to use such analytical integration instead.
Investigating the DT SED computation I realized that this part is quite problematic. If you compute DT SED for nu that are not sorted (e.g. if you generate lambdas and convert them to nu) the value will be wrong (I got negative). Also if the calculations are done only at course frequencies (or not covering the whole emission) this renormalization can screw up the integral emission. BB emission can be integrated analitically, so it would be better to use such analytical integration instead.
https://github.com/cosimoNigro/agnpy/blob/c7ede0c2d75fc66276c0487f6955c8cec204f504/agnpy/targets/targets.py#L596