Originally posted by: emmyl

I know someone has a similar post before. I checked, but it didn't seem to apply to my case.

I am trying to a mixed integer quadratic programming problem with linear objective function and quadratic constraints. Here is how I construct the cplex class. I do not post the part on creating each individual matrix. 

cplex = Cplex('saa_mip_2d');

cplex.Model.sense = 'maximize';

 

A=[Aineq;Aeq];

cplex.Model.A=A;

lhs=[repmat(-inf,size(Aineq,1),1);beq];

cplex.Model.lhs=lhs;

rhs=[bineq;beq];

cplex.Model.rhs=rhs;

 

cplex.Model.obj=f';

ctype=blanks(ncol);

ctype(1:2)='C';

ctype(3:ncol)='B';

cplex.Model.ctype=ctype;

 

%adding quadratic constraints

Q=blkdiag(n*pvec(1)*pvec(2)*ones(2,2),zeros(ncol-2));

pprod=2*sqrt(n*pvec(1)*pvec(2))*sqrt(pvec);

amat=repmat(pprod,m,1).*epsilon;

for j=1:m

    a=zeros(1,ncol);

    a(4+m+j)=M3;

    a(1:2)=amat(j,:);

    cplex.Model.qc(j).a=a';

    cplex.Model.qc(j).Q=Q;

end

 

cplex.solve

 

 

I have checked the size of each matrix. I have 304 variables.

size of A  is 508 by 304

size of rhs and lhs is 508 by 1

size of f' is 304 by 1

ctype is a string of length 304

size of a' is 304 by 1

size of Q is 304 by 304.

 

When I run cplex.solve, I keep getting the same message "CPLEX encountered arrays with inconsistent lengths." 

Anybody has a hint of what goes wrong here?

 

 

 

#CPLEXOptimizers
#DecisionOptimization

I don't see you setting cplex.Model.qc(j).sense or cplex.Model.qc(j).rhs anywhere. Could this be the problem?

To narrow the problem down you could try if everything goes right if you do not add any quadratic constraints.

#CPLEXOptimizers
#DecisionOptimization