Evolve algebraic expressions that rediscover hidden equations from data
Physics—and, really, all the natural sciences—often start with measurements whose underlying relationship is unknown.
Symbolic regression powered by Genetic Programming (GP) lets you search the space of possible formulas directly, producing human-readable equations instead of black-box coefficients.
In this repo you will:
- Generate fully synthetic “experiments.”
- Evolve expression trees using GP (via the DEAP library).
- Rediscover the true equation (or an equally good surrogate) and quantify its accuracy.