Closed stephengreen closed 1 year ago
If it was for pesummary what would fit best with their code would be something like:
pesummary.core.file.formats.dingo.read_dingo
and
pesummary.gw.file.formats.dingo.read_dingo
This is based on how they read files here https://lscsoft.docs.ligo.org/pesummary/igwn_pinned/core/file_formats.html. This would I guess involve a fork+PR to https://github.com/pesummary/pesummary. The function would essentially be a rehash of pesummary.core.file.formats.bilby.read_bilby. This would include extra information like f_max, window type etc.
The other (out of the box) option is to store the parameter samples as a structured numpy array to directly load with hdf5. Then the pesummary object would just hold the parameter samples.
We'd need to provide more information than just the parameter samples. One other option is to use a format that pesummary is already familiar with but that contains this additional information. It would be good in any case to condense some of the metadata into something more understandable by downstream codes.
We need to set up the
Result
classes to output final samples in a format suitable for downstream tasks. What format should this be? Something suitable for PESummary? Or is there a standard format?