Keplearn is a Rust tool for recovering equations from tabular data

Keplearn searches measurement tables for compact mathematical relationships and returns candidate equations as JSON. Its documentation reports strong results on some symbolic-regression benchmarks, while noting that it performs poorly on datasets without closed-form solutions.

• Keplearn reads CSV or TSV data and searches for equations relating the input columns to a selected target; its documentation gives force = m1 × m2 / r² as an example.· 1 source
• The program uses a deterministic search over mathematical transformations and monomial forms, then scores the printed equation against the supplied data.· 1 source
• In reported SRBench results, Keplearn recovered ground-truth equations for 64.7% of noise-free Feynman datasets; its median test R² on black-box datasets was about -0.003.· 1 source