rft1d

One-Dimensional Random Field Theory in Python.

rft1d is a Python package for exploring and validating Random Field Theory (RFT) expectations regarding upcrossings in univariate and multivariate 1D continua. These expectations can be used to make statistical inferences regarding signals observed in experimentally measured 1D continua including scalar and vector time series.

Please cite:

Pataky TC (2016) RFT1D: Smooth One-Dimensional Random Field Upcrossing Probabilities in Python. Journal of Statistical Software 71(7): 1-22. https://doi.org/10.18637/jss.v071.i07

Documentation is available at: www.spm1d.org/rft1d

Installation

pip install rft1d

rft1d requires Python 3.9 or later, along with numpy, scipy and matplotlib.

rft1d has been tested on Python 3.9 through 3.13.

Running examples:

The scripts in ./examples call import rft1d so they need the package importable. Either install rft1d, or add the ./src directory to your PYTHONPATH.

The scripts in ./examples/paper reproduce the figures and code examples from the Journal of Statistical Software paper. Run ./examples/paper/all_results.py to run generate everything in one pass.

Development

The package uses a src layout, and the test suite adds ./src to sys.path itself (see tests/conftest.py), so no installation step is needed to work on the source:

git clone https://github.com/0todd0000/rft1d.git
cd rft1d
pip install pytest numpy scipy matplotlib

Running tests:

pytest                      # the whole suite
pytest -m "not slow"        # skip the 100,000-case validation (about 20 s)
pytest tests/test_p_RF.py   # a single file

There are two test types:

• test_p_RF.py and test_uc_RF.py check rft1d's probabilities and critical thresholds against reference values computed with SPM12b, which are stored in tests/data-spm12b. These establish that rft1d's probability calculations are correct.
• test_characterization.py replays about 1,400 probes of the public API -- including an inventory of every public name and signature -- against values recorded in tests/data-characterization/golden.npz. These establish that behaviour has not changed.

If you deliberately change public behavior, re-record the golden values and review the resulting diff:

python tests/gen_characterization_golden.py

Issues

Please report software bugs or other problems by searching existing issues or creating a new issue here.