can anyone help me about Error using optim.problemdef.OptimizationProblem/solve Invalid argument at position 2. Value must be finite.

Question

binbin cao 2023-8-8

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2006337-can-anyone-help-me-about-error-using-optim-problemdef-optimizationproblem-solve-invalid-argument-at

编辑： binbin cao 2023-8-9

采纳的回答： Torsten

data.mat

在 MATLAB Online 中打开

when i use optimization toolbox, i get this errors：

Error using optim.problemdef.OptimizationProblem/solve

Invalid argument at position 2. Value must be finite.

Error in structureOpt (line 23)

[vv,sol,fval,exitflag,output] = solve(prob,x0,'Solver','ga','Options',options);

Caused by:

Failure in user-supplied fitness function evaluation.

Cannot continue.

Failure in initial user-supplied fitness function

evaluation. GA cannot continue.

I really can't find what goes wrong, The attachment is my program

thank you very much!

ub=[];
for i = 1:48
    ub=[ub,'1'];
end
vv=optimvar('vv','LowerBound',0,'UpperBound',bin2dec(ub),'Type','integer');
[jc,rrc,allrea,allima,~]=fcn2optimexpr(@problemfun,vv);
ans = 'optim.problemdef.OptimizationVariable'
ans = 'double'
prob=optimproblem;
prob.ObjectiveSense='maximize';
prob.Objective=jc;
tol=0.1;
prob.Constraints.rrc1=rrc-tol-6<=0;
prob.Constraints.rrc2=rrc-6+tol>=0;
prob.Constraints.allrea=allrea>=0;
prob.Constraints.allima=allima>=0; %'FunctionTolerance', 1e-8, ...
options=optimoptions(@ga,'PopulationSize', 150, ...
    'MaxGenerations', 400, ...
    'EliteCount', 10, ...
    'PlotFcn',@gaplotbestf);
rng default
x0.vv=randperm(bin2dec(ub),1);
problemfun(x0)
ans = 'struct'
Error using dec2bin
D must be numeric.

Error in solution>form_h (line 31)
nodestr=dec2bin(v);

Error in solution>problemfun (line 134)
hc=form_h(vv,map,text3,nloc);
[vv,sol,fval,exitflag,output] = solve(prob,x0,'Solver','ga','Options',options);
% rng default
% x0.vv=randperm(bin2dec(ub),1);
% [vv,sol,fval,exitflag,output] = solve(prob);
function h1=form_h(v,map,text3,nloc)
class(v)
nodestr=dec2bin(v);
nodenum=strfind(nodestr,'1')+6;
aa=zeros(6,9);
for j=1:6
    aa1=j;
    for i=1:5
        aa(j,i)=aa1+6*(i-1);
    end
end
for j=1:6
    aa1=30+(j-1)*(1+2*4);
    for i=1:4
        aa(j,5+i)=aa1+i*2;
    end
end
a4=[1+4*6 2+4*6 3+4*6 4+4*6 5+4*6 6+4*6];
[lia1,~]=ismember(aa(1,:),nodenum(1,1));
[lia2,~]=ismember(aa(2,:),nodenum(1,1));
[lia3,~]=ismember(aa(3,:),nodenum(1,1));
[lia4,~]=ismember(aa(4,:),nodenum(1,1));
[lia5,~]=ismember(aa(5,:),nodenum(1,1));
[lia6,~]=ismember(aa(6,:),nodenum(1,1));
[lia7,~]=ismember(a4,nodenum(1,end));
if sum(lia1)
    nodenum=[2,nodenum];
elseif sum(lia2)
    nodenum=[1,nodenum];
elseif sum(lia3)
    nodenum=[4,nodenum];
elseif sum(lia4)
    nodenum=[3,nodenum];
elseif sum(lia5)
    nodenum=[6,nodenum];
else
    nodenum=[5,nodenum];
end
if lia7==0
    [lia1,~]=ismember(aa(1,:),nodenum(1,end));
    [lia2,~]=ismember(aa(2,:),nodenum(1,end));
    [lia3,~]=ismember(aa(3,:),nodenum(1,end));
    [lia4,~]=ismember(aa(4,:),nodenum(1,end));
    [lia5,~]=ismember(aa(5,:),nodenum(1,end));
    [lia6,~]=ismember(aa(6,:),nodenum(1,end));
    if sum(lia1)
        nodenum=[nodenum,2+4*6];
    elseif sum(lia2)
        nodenum=[nodenum,1+4*6];
    elseif sum(lia3)
        nodenum=[nodenum,4+4*6];
    elseif sum(lia4)
        nodenum=[nodenum,3+4*6];
    elseif sum(lia5)
        nodenum=[nodenum,6+4*6];
    else
        nodenum=[nodenum,5+4*6];
    end
end
h=zeros(size(map,1),size(nodenum,2)-1);
a=nloc(nodenum(1,1),:)-nloc(nodenum(1,2),:);
a=a/norm(a);
b=nloc(nodenum(1,3),:)-nloc(nodenum(1,2),:);
b=b/norm(b);
c=(nodenum(1,2)-1)*6;
h(c+1:c+3,1)=(a+b)';
for i=3:size(nodenum,2)-1
    a=nloc(nodenum(1,i-1),:)-nloc(nodenum(1,i),:);
    b=nloc(nodenum(1,i+1),:)-nloc(nodenum(1,i),:);
    a=a/norm(a);
    b=b/norm(b);
    c=(nodenum(1,i)-1)*6;
    h(c+1:c+3,i-1)=(a+b)';
end
a=nloc(nodenum(1,end-1),:)-nloc(nodenum(1,end),:);
a=a/norm(a);
c=(nodenum(1,end)-1)*6;
h(c+1:c+3,end)=a';
[a,b]=sort(map(:,2));
h1=h(b,:);
end
function [jc,rrc,allrea,allima,hc]=problemfun(vv)
load('data.mat');
alp=0.01;
bet=0.01;
cc=alp*mm+bet*kk;
n=size(mm,1);
sys.n=n;
[V,D] = eig(kk,mm);
[lambda,ki]=sort(diag(D));
V=V(:,ki);
num=3;
Vc=V(:,1:num);
mc=Vc'*mm*Vc;
kc=Vc'*kk*Vc;
ccc=Vc'*cc*Vc;
hc=form_h(vv,map,text3,nloc);
A=[zeros(size(mc,1)),eye(size(mc,1));
    -1*inv(mc)*kc,-1*inv(mc)*ccc];
B=[zeros(size(mc,1),size(hc,2));
    inv(mc)*Vc'*hc];
X=lyap(A,B*B');
[~,b]=eig(X);
jc=trace(X)*geomean(diag(b))/std(diag(b));
rc=[];
for i=1:size(A,1)
    rc=[rc,A^(i-1)*B];
end
rrc=rank(rc);
[~,ba]=eig(A);
ba=diag(ba);
rea=real(ba);
ima=imag(ba);
allrea=all(rea);
allrea=min(abs(allrea))-1;
allima=ima<=0;
allima=all(allima);
allima=min(abs(allima))-1;
end

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Torsten 2023-8-8

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2006337-can-anyone-help-me-about-error-using-optim-problemdef-optimizationproblem-solve-invalid-argument-at#answer_1284982

移动：Torsten 2023-8-8

在 MATLAB Online 中打开

Above your error - written out more clearly so that you as the coder should be able to understand the problem.

In

nodestr=dec2bin(v);

v is of class "struct", but it must be of class "numeric", "logical" or "char".

And you know that "ga" expects a double scalar value to be returned from "problemfun" ? jc is taken as this value from "ga". Is this your intention ?

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

binbin cao 2023-8-9

编辑：binbin cao 2023-8-9

thanks for your reply.

jc is my objective function. and vv is optimvar. the value of vv is converted to binary.

ub=[];

for i = 1:48

ub=[ub,'1'];

end

vv=optimvar('vv','LowerBound',0,'UpperBound',bin2dec(ub),'Type','integer');

The maximum value of vv is achieved by converting the binary number consisting of 48 ‘1’ to decimal.

The optimization objective is to find an optimal arrangement of 0 and 1 for a 48-bit binary number, such that the value of jc is maximized.

The program is able to run when the maximum value of vv is obtained by converting the binary number consisting of 24 ‘1’ to decimal.

Later, I made some modifications to form_h.m, and it led to an increased range of required vv values.

Changing the maximum value of vv to be obtained by converting the binary number consisting of 48 ones to decimal, the program becomes infeasible to run.

thank you very much.

请先登录，再进行评论。