跳转至

evotoolkit.solve()

evotoolkit.solve 

solve(
    interface: BaseMethodInterface,
    output_path: str = "./results",
    **kwargs,
) -> Any

Factory method to create and run an evolutionary optimization workflow.

This is the main entry point for using evotool with an explicit, unambiguous API. Users must explicitly create task and interface instances before calling solve().

Parameters:

Name Type Description Default
interface BaseMethodInterface

Interface instance (e.g., EoHPythonInterface, EvoEngineerPythonInterface). Must be explicitly created with a task instance. The algorithm is automatically inferred from the interface type.

 required
output_path str

Path to save results.

 './results'
**kwargs Any

Additional parameters passed to algorithm config (e.g., max_generations, max_sample_nums, pop_size, running_llm).

 {}

Returns:

Name Type Description
Any Any

Best solution found during the evolutionary optimization run.
Example

Create task instance explicitly

task = FuncApproxTask(x_data, y_noisy, y_true)

Create interface instance explicitly

interface = EoHPythonInterface(task)

Call solve with explicit interface

result = evotool.solve( interface=interface, output_path='./results', running_llm=llm_api, max_generations=5, max_sample_nums=10 )

Source code in src/evotoolkit/__init__.py 
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
def solve(
    interface: BaseMethodInterface, output_path: str = "./results", **kwargs
) -> Any:
    """
    Factory method to create and run an evolutionary optimization workflow.

    This is the main entry point for using evotool with an explicit, unambiguous API.
    Users must explicitly create task and interface instances before calling solve().

    Args:
        interface (BaseMethodInterface): Interface instance (e.g.,
            `EoHPythonInterface`, `EvoEngineerPythonInterface`). Must be
            explicitly created with a task instance. The algorithm is
            automatically inferred from the interface type.
        output_path (str): Path to save results.
        **kwargs (Any): Additional parameters passed to algorithm config
            (e.g., `max_generations`, `max_sample_nums`, `pop_size`,
            `running_llm`).

    Returns:
        Any: Best solution found during the evolutionary optimization run.

    Example:
        # Create task instance explicitly
        task = FuncApproxTask(x_data, y_noisy, y_true)

        # Create interface instance explicitly
        interface = EoHPythonInterface(task)

        # Call solve with explicit interface
        result = evotool.solve(
            interface=interface,
            output_path='./results',
            running_llm=llm_api,
            max_generations=5,
            max_sample_nums=10
        )
    """
    # Step 1: Infer algorithm from interface
    algorithm_name = infer_algorithm_from_interface(interface)

    # Step 2: Get algorithm info from registry
    algo_info = get_algorithm_info(algorithm_name)
    algorithm_class = algo_info["class"]
    config_class = algo_info["config"]

    # Step 3: Create config with all parameters
    # Note: task is accessed via interface.task
    config = config_class(interface=interface, output_path=output_path, **kwargs)

    # Step 4: Create and run algorithm
    algorithm_instance = algorithm_class(config)
    algorithm_instance.run()

    # Step 5: Get the best solution from the run
    best_solution = algorithm_instance._get_best_sol(
        algorithm_instance.run_state_dict.sol_history
    )

    return best_solution

示例

import evotoolkit
from evotoolkit.task.python_task.scientific_regression import ScientificRegressionTask
from evotoolkit.task.python_task import EvoEngineerPythonInterface
from evotoolkit.tools import HttpsApi

# 创建任务
task = ScientificRegressionTask(dataset_name="bactgrow")

# 创建接口
interface = EvoEngineerPythonInterface(task)

# 配置 LLM
llm_api = HttpsApi(
    api_url="https://api.openai.com/v1/chat/completions",
    key="your-api-key",
    model="gpt-4o"
)

# 求解
result = evotoolkit.solve(
    interface=interface,
    output_path='./results',
    running_llm=llm_api,
    max_generations=5,
    max_sample_nums=10,
    pop_size=5
)

print(f"最佳得分: {result.evaluation_res.score}")