Skip to content

LALInference to save command line to headers and fhigh to PP fixed output params

Salvatore Vitale requested to merge salvatore-vitale/lalsuite:better_lalinference_logging into master

Save the actual command line used in the dagman in the header of each output hdf5 files, for both nest and MCMC.

These appear in the header sections of the PP pages:

Nest: https://ldas-jobs.ligo.caltech.edu/~salvatore.vitale/test/O3/PE_Rota/S190425z/C01/IMRPhenomD_NRT_lowspin/1240215503.02-330564/V1L1/posplots.html MCMC: https://ldas-jobs.ligo.caltech.edu/~salvatore.vitale/test/O3/PE_Rota/S190425z/C01/IMRPhenomD_NRT_lowspin/MCMC/1240215503.02-330564/V1L1/posplots.html

Also add fhigh to the summary table, together with flow and f_ref, that where already there.

Notes:

  1. I am sure that what I have done doesn't have any side effects for Nest. An MCMC expert (@vivien, @bfarr) should check the same is true for MCMC. I believe that my hack will make MCMC save the CL also in the resume files, since the same function seems to be used for intermediate checkpointing and for the final writing, but hopefully that is not a problem.
  2. A command line is printed for each parallel chain, if one runs with 40 parallel chains, as we sometimes do for very expensive approximants, that might get very long. Here an example with two parallel runs. We might want to find a way to only print the first one, I have not done that in this patch.
  3. As already the case for flow, only one value of fhigh is printed, not one for each ifo. As such, I am not really sure which one would be printed if the user picked different flow and fhigh for different IFOs. This should be fixable quite easily, but since we don't do that for flow, I have not done it for fhigh for this patch. Might want to discuss this in a PE call.

Merge request reports