The Sphere function minimization problem with an arbitrary number of variables and objectives
506
This problem calculates a sphere function with an arbitrary number of variables and objectives.
$ echo 2 | ./sphere.py
{"objective": 4}
The location of optima can be specified via SPHERE_OPTIMA environmental variable, which must be a matrix.
If you need a 1-variable, 1-objective problem with the optimum at the origin (default), then set:
SPHERE_OPTIMA=[[0]]
If you need a 3-variable, 2-objective problem with argmin f1=[1, 2, 3] and argmin f2=[4, 5, 6], then set:
SPHERE_OPTIMA="[[1, 2, 3], [4, 5, 6]]"
Content type
Image
Digest
Size
591.8 MB
Last updated
almost 6 years ago
docker pull opthub/sphere