Neural networks produce compact scalar and graviton scattering expressions

What was already known

The amplitudes can be calculated with established methods, but their resulting expressions can have hundreds of terms. Rediscovering Parke–Taylor is a reproduction; the new compact representations and simplification method are the advance.

What changed and why it matters

A reusable symbolic-calculation method with new compact amplitude representations; the contribution concerns expression structure rather than a new physical law.

The AI contribution

A trained transformer proposes simplified expressions; a contrastive-learning network groups terms for sequential simplification. Human researchers designed the training and checking procedure.

Limitations

The pipeline validates reductions numerically at sample momenta; that is not a formal proof of every identity. Generalization and success rates depend on the tested expression families. AIProofWatch has not run the models or audited the identities.

Verification evidence

Evidence labels show what supports the claim. Importance rates its impact if correct.

Correction and publication history