Python package to calculate the population of molecules in particular quantum states using a master equation approach. Designed for (and currently only usable) for excitation of the hydroxyl radical (OH) to one excited vibrational state and one excited electronic state, with two different lasers, and up to four vibronic states being tracked.
This represents a slow accumulation of calculations I've needed to do for a research project, and would need some work to become more generalized.
- Extract absorption feature information (upper/lower states, energy gap, degeneracies, Einstein coefficients, ...) from a HITRAN-type file using
loadhitran. (Vibrational lines of OH only, with limited parsing of H2O.)
- Calculate shape of absorption feature from Doppler and pressure broadening (
- Automatically define fast modulation of narrow laser linewidth over broadened absorption feature (
- Solve system of ODEs to calculate population in each state over time. Processes included in the ODEs are stimulated absorption/emission, spontaneous emission, and lambda doublet/rotational/vibrational/electronic relaxation (
matplotlibfigures of populations or laser frequency over time; create figure of infrared absorption feature (
- convenience unit conversion functions related to atmospheric science (
- constants and functions related to OH spectroscopy (
- Simulate absorption cross-section spectrum (
popmodel.simspecmodule -- needs to be imported separately from
The core of
popmodel is the
KineticsRun object. Each
KineticsRun instance requires dictionaries of parameters describing rates of spectroscopic transitions, lasers, detection cell, transition lines, and ODE integration. The expected dictionary format is designed for extraction from a YAML file and compatible with command line use.
Required input files
Infrared line parameters are extracted from the 140-character-format HITRAN 2012 file for OH (default filename
13_hit12.par), which can be accessed at https://www.cfa.harvard.edu/HITRAN/. Some low-level functions within
loadhitran module can also read other molecules' HITRAN files, but trying to go through the full workflow called by
loadhitran.processhitran() used in setting up a
KineticsRun will not work due to the need to parse strings describing molecule-specific term descriptions. See the HITRAN website for more documentation related to the record format.
An 200-line excerpt from the OH HITRAN file is included at
src/popmodel/data/hitran_sample.par for use by the test module. To extract the path to
from pkg_resources import resource_filename hpath = resource_filename('popmodel','data/hitran_sample.par')
YAML parameter file
Parameters for setting up a
KineticsRun instance are organized in dictionaries corresponding to a YAML parameter file. A template for the format that the YAML file must follow can be found at
To extract the path to
from pkg_resources import resource_filename yamlpath = resource_filename('popmodel','data/parameters_template.yaml')
pip creates command-line command
popmodel. Format of command line arguments:
HITFILE PARAMETERS [-l] LOGFILE [-c] CSVOUTPUT [-i] IMAGE [-v]
$ popmodel 13_hit12.par parameters.yaml -l output.log -c output.csv -i output.png
import popmodel as pm pm.add_streamhandler() # optional, print logging.INFO to screen pm.add_filehandler("path/to.log") # optional, write logging.INFO to file par = pm.importyaml("path_to/yaml/parameters.yaml") hpar = pm.loadhitran.processhitran("path_to/13_hit12.par") k = pm.KineticsRun(hpar,**par) k.solveode() k.popsfigure()
Install from PyPI:
$ pip install popmodel
Install most recent commit on Github (less stable):
$ pip install git+https://github.com/awbirdsall/popmodel
Tested for Python 2.7 and 3.5.
matplotlib>=1.5 (automatically handled if using
pip to install). I recommend using
conda to install the Scipy stack on a Windows machine if
pip is having issues.
Developed in a Windows environment. Travis-CI tests performed on Linux virtual machine.
Tests written using
pytest with the
pytest-mpl plugin to check matplotlib image output and the
pytest-cov plugin to assess coverage. To create baseline images that will be compared against (from root project folder):
$ py.test --mpl-generate-path=tests/baseline
Then to run full test suite with matplotlib and coverage plugins:
$ py.test --mpl --cov=popmodel
Travis-CI tests do not run the
--mpl image comparison.
- fix error in oh.MASS (was 106 too large)
- fix absolute scaling of cross-sections in
simspec(missing factor of 3E10)
simspecto make cross-section functions accessible without using hitran line
- add support for Python 3 (Python 3.5 tested)