repack

Repack and Compress Line-transition Data for Radiative-tranfer Calculations

This code identifies the strong lines that dominate the spectrum from the large-majority of weaker lines. The code returns a binary line-by-line (LBL) file with the strong lines info (wavenumber, Elow, gf, and isotope ID), and an ascii file with the combined contribution of the weaker lines compressed into a continuum extinction coefficient (in cm-1 amagat-1) as function of wavenumber and temperature.

Currently available databases:

• ExoMol (http://www.exomol.com/)
• HITRAN (https://www.cfa.harvard.edu/hitran/)
• Kurucz's TiO (http://kurucz.harvard.edu/molecules/tio)

Team Members

• Patricio Cubillos (IWF) patricio.cubillos@oeaw.ac.at

Install

repack has been tested to work on Python 3.6 and 3.7; and runs (at least) in both Linux and OSX. You can install repack from the terminal with pip:

# Note that on PyPI ``repack``is indexed as ``lbl-repack``:
pip install lbl-repack

Getting Started

The following example compresses the Exomol HCN line-transition data. First, download the ExoMol HCN dataset (there is no need to unzip the files):

# Download ExoMol HCN data:
wget http://exomol.com/db/HCN/1H-12C-14N/Harris/1H-12C-14N__Harris.states.bz2
wget http://exomol.com/db/HCN/1H-12C-14N/Harris/1H-12C-14N__Harris.trans.bz2
wget http://exomol.com/db/HCN/1H-12C-14N/Harris/1H-12C-14N__Harris.pf
wget http://exomol.com/db/HCN/1H-13C-14N/Larner/1H-13C-14N__Larner.states.bz2
wget http://exomol.com/db/HCN/1H-13C-14N/Larner/1H-13C-14N__Larner.trans.bz2
wget http://exomol.com/db/HCN/1H-13C-14N/Larner/1H-13C-14N__Larner.pf

Then create a repack configuration file ('repack_HCN.cfg') like this below:

[REPACK]

# Line-transition files:
lblfiles = 1H-12C-14N__Harris.trans.bz2
           1H-13C-14N__Larner.trans.bz2

# Database type [exomol, hitran, or kurucz]:
dbtype = exomol

# Output file name (without file extension):
outfile = HCN_exomol_harris-larner_0.3-33um_100-3000K_sthresh_0.01

# Wavenumber boundaries and sampling rate (in cm-1):
wnmin =   303.0
wnmax = 33334.0
dwn   =     1.0

# Temperature sampling:
tmin  =  100.0
tmax  = 3000.0
dtemp =  100.0

# Line-intensity threshold for strong/weak lines:
sthresh = 0.01

# Maximum chunk size of lines to handle at a time:
chunksize = 5000000
ncpu = 5

And run repack:

repack repack_H2O.cfg

which will produce the following screen output:

# Call the repack command-line executable for the HCN demo config file:
repack repack_HCN.cfg

::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
  repack: line-transition data compression.
  Version 1.4.1.
  Copyright (c) 2017-2020 Patricio Cubillos.
  repack is open-source software under the MIT license.
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::


Starting: Fri Apr  3 14:45:25 2020
Unzipping: '1H-12C-14N__Harris.trans.bz2'.
Unzipping: '1H-13C-14N__Larner.trans.bz2'.
Reading: '1H-12C-14N__Harris.trans.bz2'.
Reading: '1H-13C-14N__Larner.trans.bz2'.
  Flagging lines at  100 K (chunk 1/14):
  Compression rate:       96.82%,    148,115/ 4,662,663 lines.
  Flagging lines at 3000 K:
  Compression rate:       86.89%,    611,256/ 4,662,663 lines.
  Total compression rate: 84.60%,    717,921/ 4,662,663 lines.

...

  Flagging lines at  100 K (chunk 14/14):
  Compression rate:       95.47%,    209,217/ 4,619,175 lines.
  Flagging lines at 3000 K:
  Compression rate:       75.13%,  1,148,804/ 4,619,175 lines.
  Total compression rate: 73.22%,  1,237,122/ 4,619,175 lines.

With a threshold strength factor of 0.01,
kept a total of 7,553,671 line transitions out of 65,586,274 lines.

Successfully rewriten exomol line-transition info into:
  'HCN_exomol_harris-larner_0.3-33um_100-3000K_sthresh_0.01_lbl.dat' and
  'HCN_exomol_harris-larner_0.3-33um_100-3000K_sthresh_0.01_continuum.dat'.
End: Fri Apr  3 14:51:06 2020

The output binary file 'HCN_exomol_harris-larner_0.3-33um_100-3000K_sthresh_0.01_lbl.dat' contains the line-by-line opacity information for HCN, which represent most of the opacity contribution into the spectrum. The information is encoded as a sequence of three doubles and an integer containing the wavenumber (in cm-1), lower-state energy (in cm-1 units), gf value, and isotope index, respectively, for each transition. This info can be easily read with the following python script:

import repack.utils as u
wn, elow, gf, iiso = u.read_lbl('HCN_exomol_harris-larner_0.3-33um_100-3000K_sthresh_0.01_lbl.dat')

The output ascii file 'HCN_exomol_harris-larner_0.3-33um_100-3000K_sthresh_0.01_continuum.dat' contains the remaining opacity contribution of the weak lines (in cm-1 amagat-1 units) as function of wavenumber and temperature. This is a minor contribution compared to that of the LBL output file.

Re-sorting MARVELized files

Since some ExoMol .states files have been MARVELized (refined energy levels), the .trans files are no longer sorted by wavenumber. This is a problem for repack since its binaary searches rely on a sorted wavenumber files. To solve this, the user should sort the files before repacking:

# First sort the .trans files (use same config file as a repack file):
repack -sort repack_H2O.cfg

# Now run repack as usual:
repack repack_H2O.cfg

Be Kind

Please, be kind and acknowledge the effort of the authors by citing the article asociated to this project:

Cubillos (2017): An Algorithm to Compress Line-transition Data for Radiative-transfer Calculations, ApJ 850, 32.

License

Copyright (c) 2017-2020 Patricio Cubillos. repack is open-source software under the MIT license (see LICENSE).