Examples

Start with a complete optimization task. Each example provides a runnable configuration and explains the choices that affect the result.

Solve a problem

Numeric optimization

Minimize the Rastrigin function with a genetic algorithm over real vectors. This is the shortest example and a good first test of an installation.

Traveling salesperson

Solve a TSPLIB instance with permutation crossover and mutation. The example loads berlin52 and records a quality curve.

Symbolic regression

Train a symbolic regression model from generated observations. The result is an expression tree that can be inspected and compiled.

Go further

Model your own problem

Build a problem type from scratch for a production planning task with domain data, integer decisions and capacity constraints. Read this one when the built-in problems stop matching your work.

Multiobjective optimization

Find a Pareto front with NSGA-II for a design with two conflicting goals. The result is a set of tradeoffs rather than one answer.

Connect from Python

Symbolic regression from Python

Fit an expression from Python through pythonnet. The example configures a run, observes each population and reads the best model without hiding the .NET boundary.

To define a new problem or write a reusable component, read Problems, Writing operators and Writing algorithms.