Succeeds if the value of objective function ObjF can be constrained to be a local maximum, i.e, it's "slope" is 0 in every dimension; otherwise fails. This requires that a partial derivative of ObjF, and any subexpression in Constraints, exists for each variable. local_maxima should be executed prior to a call to clpBNR:global_maximum using the same objective function, e.g.,
?- [X1,X2]::real,OF=X1**4*exp(-0.01*(X1*X2)**2),
local_maxima(OF,{2*X1**2+X2**2==10}), global_maximum(OF,Z),solve([X1,X2]).
OF = X1**4*exp(-0.01*(X1*X2)**2),
X1:: -2.23606797749979...,
X2:: 0.0000000000000000...,
Z:: 25.0000000000000... ;
OF = X1**4*exp(-0.01*(X1*X2)**2),
X1:: 2.23606797749979...,
X2:: 0.0000000000000000...,
Z:: 25.0000000000000... .
- See also
- -
clpBNR:local_maxima/1