radis.test.io.test_hitemp module

Created on Tue Feb 2 13:51:40 2021

@author: erwan

test_calc_hitemp_CO_noneq(verbose=True, *args, **kwargs)[source]

Test proper download of HITEMP CO database.

Good to test noneq calculations with direct-download of HITEMP.

05_HITEMP2020.par.bz2 is about 14 Mb. We still download it (once per machine) because it allows to compute & test noneq spectra, which failed with direct HITEMP download in RADIS 0.9.28.

Approximate cost for the planet 🌳 : ~ 14 gr CO2 (hard to evaluate !).

test_calc_hitemp_spectrum(*args, **kwargs)[source]

Test direct loading of HDF5 files

test_fetch_hitemp_OH_pytables(verbose=True, *args, **kwargs)[source]

Test proper download of HITEMP OH database, with two engines.

Good to test fetch_hitemp. 13_HITEMP2020.par.bz2 is only 900 kb so it can be run on online tests without burning the planet 🌳

⚠️ if using the default chunksize=100000, it uncompresses in one pass (only 57k lines for OH) and we cannot test that appending to the same HDF5 works. So here we use a smaller chunksize of 20,000. .

test_fetch_hitemp_OH_vaex(verbose=True, *args, **kwargs)[source]

Test proper download of HITEMP OH database, with two engines.

Good to test fetch_hitemp. 13_HITEMP2020.par.bz2 is only 900 kb so it can be run on online tests without burning the planet 🌳

⚠️ if using the default chunksize=100000, it uncompresses in one pass (only 57k lines for OH) and we cannot test that appending to the same HDF5 works. So here we use a smaller chunksize of 20,000. .

test_fetch_hitemp_all_molecules(molecule, verbose=True, *args, **kwargs)[source]

Test fetch HITEMP for all molecules whose download URL is available.

..warning::

this downloads gigabytes of data. It is unselected by default by Pytest (see radis/setup.cfg)

The bz2 compression factor gives about 17 MB / million lines :

  • OH (57 k lines) is only 900 kb

  • CO (0.7 M lines) is only ~14 Mb

  • CH4 (114 M lines) is 435 MB

If it fails, check the databases downloaded in ~/.radisdb

Notes

Performance tests of chunksize tested on CO :
  • chunksize=1000000 > 22s , 1 iteration ~ 22s

  • chunksize=100000 > 18s, , 1 iteration ~ 4s

  • chunksize=50000 > 19s ,1 iteration ~ 2s

  • chunksize=1000 –> 90s , 1 iteration << 1s

test_fetch_hitemp_partial_download_CO2(verbose=True, *args, **kwargs)[source]

Test partial download of HITEMP CO2 database.

Good to test fetch_hitemp. CO2-02_02500-03000_HITEMP2010.h5 is only 20 MB so it can be run on online tests without burning the planet too much 🌳

test_no_redownload(*args, **kwargs)[source]

Test that modification time for 2 successive hitemp fetch is the same, i.e., there is no redownload

test_parse_hitemp_missing_labels_issue280(*args, **kwargs)[source]

Test dtype problems resulting from missing labels

Issue 280 https://github.com/radis/radis/issues/280

test_partial_loading(*args, **kwargs)[source]

Assert that using partial loading of the database works

Also check ‘vaex’ engine and radis.config["AUTO_UPDATE_DATABASE"]

test_relevant_files_filter()[source]