1) First, look at the FG20 download in UVB_data. The README describes the various data formats that are included. 
The one we want is *_spec_nu.dat, which is the full UVB spectrum J in units of 10^-21 ergs/s/cm^2/Hz/sr, 
as a function of frequency in units of Ry.
The first line is the redshifts from 0 - 10; subsequent lines have the frequency as the first entry, and then the value of the spectrum at each redshift
The directory also contains fg20_cloudy.ascii, which is in a format such that you can replace the hm12 file in Cloudy.

2) Britton's CIAOLoop Cloudy tools use the UVB as a series of spectrum files, one file per redshift. To get the FG20 data in this format, run analysis/fg19_britton_format.py. This will populate a FG20_UVB directory with z_*.out files.

3) Look at cloudy_cooling_config. This contains ib_fg2020_hr.par, which is the parameter file for Britton's CIAOLoop tools. Check that this file points to the FG20_UVB directory, that all paths are correct in general, and that the redshifts are correct. Don't do any redshift > than what is in the FG20 dataset, otherwise there will be missing runs.

4) To run CIAOLoop:
	cd main_directory
	cloudy_cooling_tools/CIAOLoop cloudy_cooling_config/ib_fg2020_hr.par -np 16

5) Run Britton's tool for collating the results into h5 format:
	python hdf5_ion_balance_tools/convert.py <runfile> <output_file> <elements>

5b) If this fails with missing runs, use tasks/get_missed_runs.py to identify which runs failed. Previously, this has been redshifts that aren't included in the FG20 dataset. In this case, we can rename the files with the run numbers they would have had without this redshift. For this, use analysis/correct_run_numbers.py

Where the runfile is ion_balance_tables/ion_fraction_FG20/ion_fraction_FG20.run

6) Make Oppenheimer style tables using analysis/make_oppenheimer_tables.py