What should I do to limit the search space to [-1,1]？

[KeXue-NJU]

Here are my core-codes:

self.strategy = cma.CMAEvolutionStrategy(x_start, 0.5)

es = self.strategy

for count in range(budget):
    solutions = es.ask()
    es.tell(solutions, [cma.ff.rosen(s) for s in solutions])
    es.disp()

[nikohansen]

self.strategy = cma.CMAEvolutionStrategy(x_start, 0.5, {'bounds': [[-1], [1]]})

should do.

Check out the output of

import cma

cma.CMAOptions('boun')

{'BoundaryHandler': 'BoundTransform  # or BoundPenalty, unused when ``bounds in (None, [None, None])``',
 'bounds': '[None, None]  # lower (=bounds[0]) and upper domain boundaries, each a scalar or a list/vector'}

Furthermore,

help(cma.CMAEvolutionStrategy)

gives an example, see also here.

[KeXue-NJU]

Thanks a lot.

[unghee]

hello, can we set a different bounds for each dimension? for example, for 2dim case:

def f(x, y):  # a 1-D function
    return x**2 + y**2

x0 = [0.5, 0.8]
strategy = cma.CMAEvolutionStrategy(x0, 0.5, {'bounds': [[-1, 1], [0, 5]]})

if I do this, for me it throws an error as following :

ValueError: argument of inverse must be within the given bounds

[nikohansen]

The lower bound of the second variable, bounds[0][1], was set to 1 and the initial solution was 0.8, below the lower bound, hence the error.

[unghee]

It worked. Thank you for your help!