Previously, I assumed that the number of spline points in the scaled proton excitation rates in the psplups files depended on the type of B&T scaling that was being applied. However, this is not the case (see Young et al., 2003). This led to the extraneous empty entries in the rates being filled with NaN which led the proton rates to be zero at all temperatures. This fixes that issue by excluding the empty entries before passing them to an astropy table which prevents them from being converted to NaN.
This PR also fixes a bug in the DEM file parsing where, when converting to HDF5, there was a naming conflict between the top level files and the files in the version3 subdirectory. This is because at least one of these files has the same name and thus, when flattening all to the same level, we cannot have conflicting group names. I'm now prepending "version3" to the group name of those files that were originally in that subdirectory.
Previously, I assumed that the number of spline points in the scaled proton excitation rates in the
psplups
files depended on the type of B&T scaling that was being applied. However, this is not the case (see Young et al., 2003). This led to the extraneous empty entries in the rates being filled with NaN which led the proton rates to be zero at all temperatures. This fixes that issue by excluding the empty entries before passing them to an astropy table which prevents them from being converted to NaN.This PR also fixes a bug in the DEM file parsing where, when converting to HDF5, there was a naming conflict between the top level files and the files in the
version3
subdirectory. This is because at least one of these files has the same name and thus, when flattening all to the same level, we cannot have conflicting group names. I'm now prepending "version3" to the group name of those files that were originally in that subdirectory.