About
Moon Walk is an autonomous AI-driven physics research project exploring whether Einstein's field equations and the cosmological constant can be derived entirely from quantum information theory — with zero free parameters.
The conjecture
The spacetime metric is uniquely determined by the information-theoretic timing capacity of quantum field detectors, and Einstein's field equations emerge from the Clausius inequality when heat flow is measured by this capacity.
This is tested through three theorems: the Slope Theorem (capacity encodes temperature), Metric Recovery (metric from capacity optimization), and Field Equation Selection (Einstein's equations from Clausius + capacity).
The prediction
The project predicts the cosmological constant from the logarithmic correction to entanglement entropy. Summing all Standard Model species (4 scalars, 45 Weyl fermions, 12 vectors), the trace anomaly δSM = −11.06 and lattice-measured area-law coefficient αSM = 2.8 give Λpredicted / Λobserved ≈ 1.2 — within 20% of observation, with zero free parameters. A single scalar field alone gives Λ / Λobs ≈ 0.7.
Methodology
Each experiment tests one specific assumption with no circular reasoning. Every result is backed by reproducible numerical computations with explicit error bounds. Multiple extraction methods are computed in parallel to verify robustness.
The agent
An autonomous AI agent (Claude) designs experiments, writes code, runs computations, analyzes results, and plans next steps — pushing each new experiment to the repository automatically. It reads previous reports, identifies gaps, designs the next test, and iterates until assumptions are confirmed or falsified.
Phases
Non-circularity
No experiment assumes General Relativity or its consequences as input. Every step is built from quantum field theory, information theory, and thermodynamics. This is verified through explicit non-circularity audits in each experiment report.
Inspiration
This ad inspired our design process of our autonomous agent to be built around the scientific process.