Skip to content

Phaethon Documentation Hub

Welcome to the official documentation for Phaethon, the End-to-End Physics-Constrained Scientific Computing & Sci-ML Stack.

Standard machine learning ecosystems treat data as mathematically blind, naked floating-point numbers. Phaethon bridges the gap between data-driven AI and fundamental physical reality. It provides a unified ecosystem that integrates native-speed tabular ingestion, rigorous dimensional tensor mechanics, and physics-constrained (informed) neural networks (PINNs) into a single, cohesive Python framework.

Whether you are cleaning chaotic sensor data or training Fourier Neural Operators for complex PDEs, Phaethon ensures absolute mathematical integrity from the ground up.

The Architecture

Phaethon is built upon four deeply integrated pillars. Select a module below to explore its API and theoretical foundations:

Dimensional Tensor Algebra

The mathematical heart of the framework. A metaclass-driven physics engine operating across 90+ physical domains. It features Isomorphic Firewalls, Semantic Domain Locks, real-time logarithmic scale evaluation, and zero-overhead NumPy array wrapping.

Hybrid Data Engineering

Declarative data pipelines designed for machine speed. Leverages a dedicated Rust backend for extreme-speed physical string parsing and C++ RapidFuzz for fuzzy ontologies. Integrates seamlessly with Pandas and Polars to execute vectorized imputation, clipping, and physical validation.

Classical Sci-ML

The Scikit-Learn bridge. Equips classical machine learning algorithms with physics-aware Meta-Estimators. Features the BuckinghamTransformer for automated, SVD-powered dimensionless feature synthesis, and evaluates models using dimensionally strict metrics.

Neural PDEs & PINNs

The deep learning frontier. Deep PyTorch integration that unlocks PTensor (Physics-Aware Autograd). Equip your models with native physical calculus (gradients, Laplacians, divergence, curl), Spectral Convolutions for FNOs, and absolute Physics-Informed Loss Tribunals.

Installation & Modularity

Phaethon is intentionally modular. It supports Python 3.11 to 3.14. You can install the base physics engine, or opt-in to specific scientific stacks to avoid bloating your environment.

# 1. Base Physics Engine (Zero extra dependencies)
pip install phaethon

# 2. Data Engineering (Rust parser + Pandas/Polars + RapidFuzz)
pip install 'phaethon[dataframe]'
pip install 'phaethon[polars]'

# 3. Classical Machine Learning (Scikit-Learn)
pip install 'phaethon[sklearn]'

# 4. Deep Learning & PINNs (PyTorch)
pip install 'phaethon[torch]'

# 5. Polyglot I/O Storage (HDF5 & PyArrow Parquet)
pip install 'phaethon[io]'

# 6. The Complete Enterprise Bundle
pip install 'phaethon[all]'

Open Source & Contributing

Phaethon is an actively evolving open-source project released under the MIT License. Built for the global scientific community, we highly encourage scientists, data engineers, and developers to contribute!

Whether you are fixing a typo, adding a new physical domain, or optimizing the Rust backend, your help is massively appreciated.