This utility creates source code to calculate the Jacobian matrix analytically for a chemical reaction mechanism.
pyJac can be run either as a module or an executable. To run as an executable, from the command line change to the proper directory, change the file mode to executable, and run:
chmod +x pyJac.py
./pyJac.py [options]
To run it as a module, change to the appropriate directory and run:
python -m pyjac [options]
The generated source code is placed within the out (by default) directory, which is created if it doesn't exist initially.
In the above, [options] indicates where command line options should be specified. The options available can be seen using -h or --help, or below:
-h, --help show this help message and exit
-l {c,cuda,fortran,matlab}, --lang {c,cuda,fortran,matlab}
Programming language for output source files.
-i INPUT, --input INPUT
Input mechanism filename (e.g., mech.dat).
-t THERMO, --thermo THERMO
Thermodynamic database filename (e.g., therm.dat), or
nothing if in mechanism.
-b, --build_path
The folder to generate the Jacobian and rate subroutines in
-ls SPECIES NAME, --last_species SPECIES NAME
The species to move to the end of the mechanism to force mass conservation.
Defaults to (in order) N2, Ar, He.
-ic STRING, --initial-conditions STRING
A comma separated list of initial conditions to set in the
set_same_initial_conditions method.
Expected Form: T,P,Species1=...,Species2=...,...
Temperature in K
Pressure in Atm
Species in moles
-nco, --no-cache-optimizer
Turn off the cache optimization functionality
-nosmem, --no-shared-memory
Turn off shared memory usage for CUDA
-pshare, --prefer-shared
Prefer a larger shared memory size (compared to L1 cache size) for CUDA.
(Not recommended)
-nb, --num_blocks
The target number of blocks per streaming multiprocessor for CUDA.
Default is 8
-nt, --num_threads
The number of threads per block for CUDA.
Default is 64
-mt, --multi-threaded
The number of threads to use during cache-optimization.
Default is the CPU count returned by multiprocessing
-fopt, --force-opt
Force re-running of cache optimizer.
By default cache optimizer will turn off if it detects a previous optimization
for the same mechanism
-ad, --auto-diff
Generate code for Adept autodifferentiation library (used for validation)
-sj, --skip_jac
Do not generate Jacobian subroutine
pyJac is released under the MIT license; see LICENSE for details.
If you use this package as part of a scholarly publication, please cite the following paper in addition to this resource:
- TBD
Created by Kyle Niemeyer. Email address: kyle.niemeyer@gmail.com