This repository is designed to estimate the number of times a timestep length should be halved in order for a given solver to attain convergence. For solvers with adaptive timestepping functionality, this allows the user to predict the timestep length that can be used at each step in the timestepping scheme. If this prediction is accurate then it avoids the need for trial-and-error approaches, whereby successively halved timestep lengths are tried until the solver converges.
We strongly advise that users of mlstep create a Python virtual environment
before installing it. Doing so avoids polluting the system Python environment.
See https://docs.python.org/3/library/venv.html for details on how to do this.
For a basic install the mlstep module, activate your virtual environment,
clone the repository, and then run
cd mlstep
pip install -e .For a development install, some further steps are recommended:
cd mlstep
# Install optional dev dependencies
pip install -e .[dev]
# Configure pre-commit hooks
pre-commit install