Пример #1
0
from autotune import TuningProblem
from autotune.space import *

# problem space
task_space = None

input_space = Space([Real(-15, 15, name=f'x_{i}') for i in range(10)])

output_space = Space([Real(-inf, inf, name='y')])


def myobj(point: dict):
    def ackley(x, a=20, b=0.2, c=2 * pi):
        x = np.asarray_chkfinite(x)  # ValueError if any NaN or Inf
        n = len(x)
        s1 = sum(x**2)
        s2 = sum(cos(c * x))
        return -a * exp(-b * sqrt(s1 / n)) - exp(s2 / n) + a + exp(1)

    x = np.array([point[f'x_{i}'] for i in range(len(point))])
    objective = ackley(x)

    return objective


Problem = TuningProblem(task_space=None,
                        input_space=input_space,
                        output_space=output_space,
                        objective=myobj,
                        constraints=None,
                        model=None)
Пример #2
0
from autotune import Space, TuningProblem

task_space = Space(matrix=["boyd1.mtx"])

input_space = Space(m=(10, 100), n=(10, 100))

problem = TuningProblem(task_space, input_space)

print('problem.to_json: ', problem.to_json(), end='\n')

problem_copy = TuningProblem(json_repr=problem.to_json())
print(problem_copy)