The Principia Machina/Tools/AIFeynman

AI Feynman: Physics-Inspired Symmetry & Dimensional Symbolic Discovery

Overview & Architectural Fit

Developed by Udrescu & Tegmark (MIT), AI Feynman is a symbolic regression framework designed specifically around the fundamental structural properties of theoretical physics: dimensional homogeneity, translational/rotational symmetries, separable sub-functions, and polynomial invariants.

Mathematical Evaluation for TPM & SVD

• Noether Symmetry Exploitation (9.0/10): Unlike blind genetic algorithms, AI Feynman tests datasets for fundamental physical symmetries (e.g., $f(x,y)=g(x+y)$ or $f(x,y)=g(\sqrt{x^2+y^2})$). This directly mirrors Noether's First Theorem: identified symmetries correspond to underlying continuous conservation laws.
• Dimensional Analysis (Buckingham $\Pi$ Theorem): Enforces strict physical units, reducing the dimensional search space prior to expression generation.
• NN-Driven Recursive Graph Decomposition: Uses trained neural networks as surrogates to inspect partial derivatives and split complex multi-variable field dynamics into simpler, separable lower-dimensional algebraic trees.

Strategic Role in SVD

AI Feynman serves as the primary automated tool within TPM for verifying whether empirical field data exhibits the precise Maxwellian symmetries required by Structured Vacuum Dynamics (SVD).

References

Added on 2026-10-06 by the Research Toolkit: the tool's citations, each checked against the registry or the project's own page that day. The text above is the survey's own.

• AIF20 – Udrescu, S.-M., Tegmark, M. (2020). AI Feynman: A physics-inspired method for symbolic regression. Science Advances 6, eaay2631. doi:10.1126/sciadv.aay2631; arXiv:1905.11481.
• AIF2-20 – Udrescu, S.-M., Tan, A., Feng, J., Neto, O., Wu, T., Tegmark, M. (2020). AI Feynman 2.0: Pareto-optimal symbolic regression exploiting graph modularity. Advances in Neural Information Processing Systems 33 (NeurIPS 2020), 4860–4871. arXiv:2006.10782.