Genetic Algorithm: undefined function or method for input arguments of type' double '.

i just trained a neural network and I want to test it with new data and save them

namefunction = net;
 save namefunction

clear all
clc 
load namefunction

my program ga :

fitness= @namefunction ;
nvars=2;% Number of variables
LB=[8 5];%Lb Lower bound on x
UB=[30 10]; %Ub Upper bound on x
[x,y] = ga(@(x)-fitness(x),nvars,[],[],[],[],LB,UB)

i have error !!

??? Error using ==> makeState at 51
GA cannot continue because user supplied fitness function failed with the following error:
Undefined function or method 'namefunction' for input arguments of type 'double'.

Error in ==> galincon at 18
state = makeState(GenomeLength,FitnessFcn,Iterate,output.problemtype,options);

Error in ==> ga at 306
        [x,fval,exitFlag,output,population,scores] = galincon(FitnessFcn,nvars, ...

Error in ==> Untitled3 at 5
[x,y] = ga(@(x)-fitness(x(1),x(2)),nvars,[],[],[],[],LB,UB)

Answer 1

[Star Strider]

The fitness function needs to accept arguments, and (in general) needs to return a scalar result. That will likely require that ‘namefunction’ be specified with the argument list it accepts, for example:

[x,y] = ga(@(x)-namefunction(x(1),x(2)),nvars,[],[],[],[],LB,UB)

Note that since ‘namefunction’ is not provided, we cannot test it with a ga call to see what the problems could be.

See the ga documentation section on fun.

[ad aymen]

Unfortunately,the same error, the results were there before, but when i saved the results neural network again, an error has occurred

[Star Strider]

I have no idea what is ‘over the horizon’ — such as whatever ‘namefunction’ is and what its arguments and outputs might be — and so out-of-sight with respect to what you posted.