Testing¶
simsopt
includes unit and regression tests, and continuous integration.
Python test suite¶
The main test suite is based on the standard unittest
python module.
Source code for the python tests is located in the tests
directory.
These tests will use the installed version of the simsopt
python package,
which may differ from the code in your local repository if you did not
make an editable install (see Installation).
To run all of the tests in the test suite on one processor, you can type
./run_tests
from the command line in the repository’s home directory. Equivalently, you can run
python -m unittest
from the tests
directory.
For some of the tests involving MPI, it is useful to execute the tests for various numbers of processes to make sure the tests pass in each case. For this purpose, it is not necessary to run the entire test suite, only the tests for which MPI is involved. For convenience, you can run the script
./run_tests_mpi
in the repository’s home directory. This script runs only the tests
that have mpi
in the name, and the tests are run on 1, 2, and 3
processors.
The tests make use of data files in the tests/test_files
directory.
Modular testing¶
Often you may want to run fewer tests focusing on a module or a submodule or a single class or a method inside a class.
To run all tests within a given folder, let’s say geo
, run
python -m unittest discover -t . -s geo
from the tests
directory. For evaluating tests within a single file, for example geo/test_curve.py
, run
python -m unittest geo.test_curve
from the tests
directory. By using the dot
operator, you can run a single test suite or a single test case also.
python -m unittest geo.test_curve.Testing.test_curve_helical_xyzfourier
Parallel Testing¶
unittest
executes tests in serial, which takes a really long time especially when all the simsopt tests are checked. If you have a beefier machine with multiple cores, you can speed up testing by running pytest
in parallel mode. Install pytest
and pytest-xdist
and then run
pytest -n <NUMBER_OF_CORES>
from the tests
directory, where <NUMBER_OF_CORES>
is the number of CPU cores of the machine.
Longer examples¶
For convenience, the main test suite is designed to run in no more than a few minutes.
This means that some more complicated integrated and regression tests that require substantial time
are not included. You may wish to run some of these more complicated tests by hand during development.
A number of such examples can be found in the examples
subdirectory.
Also, simsopt
and stellopt
have been benchmarked for several problems in the
stellopt_scenarios collection,
which includes the corresponding simsopt
input files.
Continuous integration¶
The serial and MPI tests are automatically run after every commit to
the repository. This automation is handled by GitHub Actions, and
controlled by the script .github/workflows/ci.yml
.
To view the results of the continuous integration runs, you can click on the “Actions”
link from the GitHub repository page,
or you can directly visit https://github.com/hiddenSymmetries/simsopt/actions.