pyem

pyem is an open-source Python library designed to support rapid prototyping and execution of empirical models and small‑scale experiments. The library provides a compact set of abstractions for defining models, organizing experiments, and collecting results, with an emphasis on clarity and ease of use. It can be used for tasks such as exploring empirical relationships, validating hypotheses, or prototyping algorithms before integrating them into larger systems.

At its core, pyem offers a Model abstraction to describe inputs, parameters, and outputs; an Experiment container to orchestrate multiple runs; and a lightweight runner to advance time or iterations. The design favors a minimal API surface and straightforward integration with NumPy, SciPy, and pandas for data processing and analysis. It supports deterministic execution as well as simple stochastic variations, enabling researchers to compare scenarios quickly.

Users typically define a model, configure an experiment with a set of runs or seeds, execute the workflow, and then analyze the aggregated results through built-in summary statistics or external plotting tools. The project includes utilities for parameter sweeps, result logging, and export to common data formats. It is designed to be easily extended with custom model components and data adapters, allowing it to fit a variety of lightweight research workflows.

pyem is maintained as an open-source project with documentation and community contributions. It is distributed via standard Python packaging channels and is compatible with major Python versions. The project aims to remain lightweight enough for quick experimentation while providing enough structure to support reproducible research.