floatcsep 0.1.4

CSEP Floating Experiment application

CSEP Floating Experiments

An application to deploy reproducible and prospective experiments of earthquake forecasting

• Set up a testing experiment for your earthquake forecasts using authoritative data sources and benchmarks.
• Encapsulate the complete experiment's definition and rules in a couple of lines.
• Reproduce, reuse, and share forecasting experiments from you, other researchers and institutions.

Table of Contents

• Installation
• Documentation
• Run and explore
• Useful Links
• Roadmap/Issues
• Contributing
• License

Installing floatCSEP

The core of floatCSEP is built around the pyCSEP package (https://github.com/sceccode/pycsep), which itself contains the core dependencies.

The simplest way to install floatCSEP, is by creating a conda environment (https://conda.io - checkout Anaconda or Miniconda) and install pyCSEP from conda-forge

conda env create -n $NAME
conda activate $NAME
conda install -c conda-forge pycsep

Clone and install the floatCSEP source code using pip

git clone https://github.com/cseptesting/floatcsep
cd floatcsep
pip install .

Please read the Installation documentation for detailed instructions and additional installation methods.

Run an Experiment

Using the command terminal, navigate to an example experiment in floatcsep/examples/ and type

floatcsep run config.yml

A runtime directory will be created in a results folder. The experiment results can be visualized in results/report.md. Check out the experiment, models and tests definition in the examples!

Important Links

• Documentation
• CSEP Website
• pyCSEP Github
• pyCSEP Documentation

Roadmap and Known Issues

• Add functionality to compare original results and reproduced results
• Add registry to filetrees (e.g. hash/byte count) for a proper experiment re-instantiation
• Add report customization
• Fix the hooks properly (user code) to be inserted into plotting/reporting functionalities.
• Add multiple logging/levels
• Create tool to check a TD model's interface with fecsep
• Define a dependency strategy to ensure experiments' reproducibility.
• Publish in conda

Contributing

We encourage all types of contributions, from reporting bugs, suggesting enhancements, adding new features and more. Please refer to the Contribution Guidelines and the Code of Conduct for more information

License

The floatCSEP (as well as pyCSEP) software is distributed under the BSD 3-Clause open-source license. Please see the license file for more information.