Some of the spectral libraries (e.g. bosz/lr) have information of stellar parameters (usually Teff and log(g)) hidden somewhere in the comments of the index.yaml file. It would be super useful to make this information actually useable, given it's already there! Some libraries might also have [Fe/H] or similar for different metallicities (e.g. Pickles). This would also help the planned KDTree implementation, as that could use this (more accurate) information instead of just the spectral types.
TBD is whether this information should stay in the index.yaml file, as it's a bit too tabular for the YAML format. Moving it to a separate tabular format (e.g. escv) would make it a lot easier to load into a table and subsequently KDTree. Other metadata would be fine to stay in YAML format though. In terms of units, it might be worth considering storing everything in log, i.e. log(Teff), because log(g) and [Fe/H] etc. are already given in logarithmic (or dex) units. The KDTree could then also just work in log space (without having to know).
Some of the spectral libraries (e.g. bosz/lr) have information of stellar parameters (usually Teff and log(g)) hidden somewhere in the comments of the index.yaml file. It would be super useful to make this information actually useable, given it's already there! Some libraries might also have [Fe/H] or similar for different metallicities (e.g. Pickles). This would also help the planned KDTree implementation, as that could use this (more accurate) information instead of just the spectral types.
TBD is whether this information should stay in the index.yaml file, as it's a bit too tabular for the YAML format. Moving it to a separate tabular format (e.g. escv) would make it a lot easier to load into a table and subsequently KDTree. Other metadata would be fine to stay in YAML format though. In terms of units, it might be worth considering storing everything in log, i.e. log(Teff), because log(g) and [Fe/H] etc. are already given in logarithmic (or dex) units. The KDTree could then also just work in log space (without having to know).