Methods API

Evolutionary methods are the core algorithms that drive the optimization process.

---

Available Algorithms

EvoToolkit provides three main evolutionary algorithms:

Algorithm	Best For	Characteristics
EvoEngineer	General optimization	Versatile, robust, good default choice
FunSearch	Function discovery	Specialized for function approximation
EoH	Heuristic optimization	Fast, efficient for simple problems

---

EvoEngineer

See the dedicated page: EvoEngineer.

---

FunSearch

See the dedicated page: FunSearch.

---

EoH (Evolution of Heuristics)

See the dedicated page: EoH.

---

Algorithm Comparison

When to Use Each Algorithm

Use EvoEngineer when: - You have a general optimization problem - You want a robust, well-tested algorithm - You need good default behavior

Use FunSearch when: - You're specifically looking for novel functions - Function discovery is the primary goal - You want to explore a diverse function space

Use EoH when: - You need fast iterations - Your problem has simple heuristics - You want efficient resource usage

Performance Characteristics

Algorithm	Speed	Exploration	Exploitation	Best Fitness
EvoEngineer	Medium	High	High	⭐⭐⭐⭐⭐
FunSearch	Slow	Very High	Medium	⭐⭐⭐⭐
EoH	Fast	Medium	High	⭐⭐⭐

---

Advanced: Custom Algorithms

You can implement custom evolutionary algorithms by extending BaseMethod:

from evotoolkit.core import BaseMethod, BaseConfig

class MyCustomAlgorithm(BaseMethod):
    def run(self):
        for generation in range(self.config.max_generations):
            # 1. Generate solutions
            solutions = self.generate_solutions()

            # 2. Evaluate solutions
            for solution in solutions:
                eval_res = self.config.interface.task.evaluate_code(solution.sol_string)
                solution.evaluation_res = eval_res

            # 3. Select best solutions
            self.select_and_update_population(solutions)

    def generate_solutions(self):
        # Your custom generation logic
        pass

    def select_and_update_population(self, solutions):
        # Your custom selection logic
        pass

See Advanced Usage Tutorial for details.

---

Next Steps

• Try different algorithms with the Tutorials
• Learn about Interfaces for connecting tasks to algorithms
• Explore Core API for the high-level solve() function