FAQ

How should negative optical depth be interpreted?

As discussed by [vanderTak07], for certain parameters (“especially low density and/or strong radiation field”, to quote [vanderTak07]), negative optical depth can arise due to a level population inversion. This requires non-local treatement of the radiative transfer, which is not possible with pyhonradex. The results should still be valid, though less accurate, for optical depths only slightly negative (\(\tau_\nu\gtrsim-0.1\)). However, the results for more strongly negative optical depths should be ignored ([vanderTak07]). pyhonradex provides an option to print a warning if the solution contains negative optical depths using the warn_negative_tau parameter of the Source class.

I get an “Underlying object has vanished” error

This probably indicates a problem with the caching of the numba-compiled functions used by pyhonradex. Deleting the __pycache__ folder should solve the issue.