Originally posted by: SystemAdmin

Hi,all
I use cplex12.2 to solve a QCP problem under matlab cplex toolbox.
when I write the Quadratic constraints,it says Too many input arguments.then I use 'cplexint' ,it's ok.
I just write like this:
QC(i).l=...;QC(i).r=...;QC(i).Q=...;
x, fval, exitflag, output] = cplexqcp (H,f,Aineq,bineq,[,http://],QC.l,QC.Q, QC.r, lb,ub,[,options);

I found if there is only one Quadratic constraint,it's ok;But if more than one,it's wrong.So I want to know how to write Quadratic constraints under cplex toolbox?
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: John Cui

If you want to generate 3 QCs as below:

QC(1).l = [0 0 0]'
QC(1).Q = [1 0 0;0 1 0;0 0 1]
QC(1).r = 1.0
 
QC(2).l = [1 0 0]'
QC(2).Q = [1 0 0;0 1 0;0 0 1]
QC(2).r = 2.0
 
QC(3).l = [0 1 0]'
QC(3).Q = [1 0 0;0 1 0;0 0 1]
QC(3).r = 3.0

Then you should write l, Q, r by:

l = [0 0 0;1 0 0;0 1 0]'
         Q = {[1 0 0;0 1 0;0 0 1]
              [1 0 0;0 1 0;0 0 1]
              [1 0 0;0 1 0;0 0 1]}
         r = [1.0 2.0 3.0]

and pass them to cplexqcp or cplexmiqcp.

Please give a try, and tell me the result, thanks.

John Cui
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Hi,John Cui
Thank you very much for your reply.
In fact ,I just write the constraints as you write below.But,Unfortunately, it doesn't work.
I use the matlab toolbox ,then it is wrong.I have used the another method in matlab ,just
using statement'cplex.addQCs',it can be work.

'Col,ng,nt 'is constant
%%%%%%%%%%%%%%%%%% wrong
...
...
...
for i=1:1:ng
for j=1:1:nt
QC((i-1)*nt+j).L=zeros(1,Col*ng*nt);
QC((i-1)*nt+j).r=0;
I=zeros(3,1);
J=zeros(3,1);
S=zeros(3,1);
I(1)=(i-1)*nt+j;J(1)=2*ng*nt+(i-1)*nt+j;S(1)=-0.5;
I(2)=ng*nt+(i-1)*nt+j;J(2)=ng*nt+(i-1)*nt+j;S(2)=1;
I(3)=2*ng*nt+(i-1)*nt+j;J(3)=(i-1)*nt+j;S(3)=-0.5;
QC((i-1)*nt+j).Q=sparse(I,J,S,ng*nt*Col,ng*nt*Col);
end
end
x, opt, status] = cplexmiqcp([,F,Aineq,Bineq,Aeq,Beq,QC.L,QC.Q,QC.r,],[,],lb,ub,ctype,[,options);
%%%%%%%%%%%%% right
...
...
...
for i=1:1:ng
for j=1:1:nt
a=zeros(Col*ng*nt,1);
I=zeros(3,1);J=zeros(3,1);S=zeros(3,1);
I(1)=(i-1)*nt+j;J(1)=2*ng*nt+(i-1)*nt+j;S(1)=-0.5;
I(2)=ng*nt+(i-1)*nt+j;J(2)=ng*nt+(i-1)*nt+j;S(2)=1;
I(3)=2*ng*nt+(i-1)*nt+j;J(3)=(i-1)*nt+j;S(3)=-0.5;
Q=sparse(I,J,S,ng*nt*Col,ng*nt*Col);
cplex.addQCs(a, Q, 'L' , 0);
end
end
cplex.solve();

My Cplex version is 12.2.0.0 ,and yesterday I found update in 12.2.0.2:
CPLEX RS00503 The lower triangles of the q matrices of quadratic constraints are not extracted in the Matlab connector

Maybe it can explain the problem.
Thank you!
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Hi,John Cui
I'm sorry for that TI'm carefully read your reply.Thank you very much for your reply.After I use the statement as below ,the problem is solved.The error due to my custom in cplexint.
I suggest that matlab cplex example should write more carefully.The Quadratic constraint in example only has one,in maybe mislead the user.

Thank you!
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: John Cui

Glad to hear it works and thanks for your suggestion.

BTW, could you please put your code inside \{code\} and \{code\} then your code will have better readability.

John Cui
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: Student2011

Hello,

We input the following into Matlab:

f = [1 1 1];
l = [0 0 0;1 0 0;0 1 0]';
Q = {[1 0 0;0 1 0;0 0 1]
              [1 0 0;0 1 0;0 0 1]
              [1 0 0;0 1 0;0 0 1]};
r = [1.0 2.0 3.0];
x = cplexmiqcp([],f,[],[],[],[],l,Q,r)

and get the following error:

??? Error using ==> cplexmiqcp at 584
Error: incorrect l,Q,r.

Is there something we are doing wrong?

Thanks!
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: John Cui

f should be:

f = [1 1 1]';

John Cui
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: Student2011

Thank you for your response.

By transposing f we still get the same error:

??? Error using ==> cplexmiqcp at 584
Error: incorrect l,Q,r.

#CPLEXOptimizers
#DecisionOptimization

Originally posted by: John Cui

Sorry, Q also should be:

Q = {[1 0 0;0 1 0;0 0 1], ...
     [1 0 0;0 1 0;0 0 1], ...
     [1 0 0;0 1 0;0 0 1]};

John Cui
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: Student2011

Hi. We greatly appreciate your help but since the documentation is limited we are having trouble deciphering the error messages.

Entering the code, with the changes you've suggested, as follows

f = [1 1 1]';
l = [0 0 0;1 0 0;0 1 0]';
Q = {[1 0 0;0 1 0;0 0 1], ...
     [1 0 0;0 1 0;0 0 1], ...
     [1 0 0;0 1 0;0 0 1]};
r = [1.0 2.0 3.0];
x = cplexmiqcp([],f,[],[],[],[],l,Q,r)

We now get the following error:

??? Error using ==> cplexmiqcp at 584
Error: qc not a valid  model qc.a.

Thank you.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: John Cui

Which version CPLEX are you using? The latest one is 12.2.0.2.
And your code works in this version:

>> f = [1 1 1]';
l = [0 0 0;1 0 0;0 1 0]';
Q = {[1 0 0;0 1 0;0 0 1], ...
     [1 0 0;0 1 0;0 0 1], ...
     [1 0 0;0 1 0;0 0 1]};
r = [1.0 2.0 3.0];
x = cplexmiqcp([],f,[],[],[],[],l,Q,r)
 
x =
 
   -0.5774
   -0.5774
   -0.5773

John Cui
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: Student2011

Thank you for you help. We did not realize we were missing an upgrade fix. We were on version 12.2. We have downloaded the fix and now the code works.

Thanks!
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: mena2011

Hi all

I am working with a quadratic objective function quadratic constraints and I am using opl cplex 12.6 . after run the problem cplex gives message  error 5002 Q in objective not positive my problem is

constraints

6<=x<=50

0<=y<=30

x^2+y^2<=80

how can I tackle this problem please

 

#CPLEXOptimizers
#DecisionOptimization

There are many threads on this forum that mention this error code.  I would recommend searching "This Forum" for "5002" using the search box above to see if you can find the answer to your question using those past responses.  If that doesn't help, you may get more attention if you create a new thread for your problem.

 

#CPLEXOptimizers
#DecisionOptimization