Originally posted by: belengarciamar

Hello all,

I am having some problems with my model. It's taking more than 9 hours to be solved, so I am trying to find out how to use the Warm Start in Cplex. I can only find the user manual but with no mathematical examples, only the code. This code uses arrays but my model doesn't have any array.

Can someone help me?

 

Thank you so much.

-Belén

#CPLEXOptimizers
#DecisionOptimization

If you are using one of the programming APIs (you did not say which, if any), just create a vector of variables (IloNumVar[] in Java, for instance) with dimension big enough to hold all your variables, then use loops to stick all your variables into the vector (in any order you like). Create a double precision vector of the same length for the values, and load that with the variable values for your warm start (in the same order that you used when loading the variable vector).

#CPLEXOptimizers
#DecisionOptimization

Originally posted by: belengarciamar

So, lets say this is my model with my data:

{string} MP=...;
{string} PT=...;
{string} PT0= PT union {"0"};
{string} MP0= MP union {"0"};

// parameters

int Composition[MP,PT]=...;
int N=...;
int M=...;

//N and M same number of MP and PT

// decission variables

dvar boolean alpha[PT0,PT0];

dvar boolean beta [MP0,MP0,PT0,PT0];

dvar int u[PT0];
dvar int v[MP0,PT0];

// objective function

minimize sum (i in MP, j in MP: j!=i, p in PT, t in PT) beta[i,j,p,t];

// constraints

subject to{
    c01:
    forall (t in PT)
      sum (p in PT0: p!=t) (alpha[p,t])<=1;
      
    c02:
    forall (p in PT, t in PT: t!=p)
      sum (i in MP:Composition[i,p]==1 , j in MP: Composition[j,t]==1) beta[i,j,p,t]==alpha[p,t];
      
    c03:
    forall (t in PT, j in MP: Composition[j,t]==1)
      beta["0",j,"0",t] + (sum (i in MP: Composition[i,t]==1) beta[i,j,t,t])+ (sum (p in PT: p!=t, i in MP: Composition[i,p]==1) beta[i,j,p,t])==1;
    
    c05:
    sum(p in PT)alpha["0",p]==1;
    
    c06:
    sum(t in PT, j in MP: Composition[j,t]==1) beta ["0",j,"0",t]==1;
    
    c07:
    forall ( p in PT, i in MP: Composition [i,p]==1)
     sum(t in PT, j in MP: Composition [j,t]==1) beta[i,j,p,t]<=1;
    
    c08:
    forall (t in PT)
      sum (j in MP: Composition[j,t]==1) beta ["0",j,"0",t] + sum (p in PT: p!=t, i in MP: Composition[i,p]==1, j in MP: Composition[j,t]==1) beta[i,j,p,t]==1;
      
    c09:
    forall (p in PT)
     sum ( i in MP: Composition [i,p]==1, t in PT: t!=p, j in MP: Composition [j,t]==1) beta [i,j,p,t]<=1;

    c10:
    forall (p in PT, t in PT, i in MP: Composition[i,p]==1, j in MP: Composition [j,t]==1)
      beta[i,j,p,t]+ beta [j,i,t,p]<=1;
      
    c11:
    u["0"]==0;
    v["0","0"]==0;
    
    c12:
    forall(t in PT)
      u[t]<=N;
     forall (j in MP, t in PT)
      v[j,t]<=M;
      
      
    c13:
    
    forall(p in PT0, t in PT: p!=t)
        u[t]>=u[p]+1-N*(1-alpha[p,t]);
      
    forall (p in PT0, t in PT,i in MP: Composition[i,p]==1, j in MP: Composition[j,t]==1)
      v[j,t]>=v[i,p]+1-M*(1-beta[i,j,p,t]);

// data file

MP={"A"    "B""E""M" "N" "O" "P" };
PT={"1"    "2""3" "4" "5" "6" "7" "8" };
N=15;
M=8;

Composition=[
            [0    0    0    0    1    1    1    1],
            [1    1    1    1    1    1    1    1],
            [0    0    0    0    1    1    1    1],
            [0    0    0    0    1    1    1    1],
            [1    1    1    1    1    1    1    1],
            [1    1    1    1    1    1    1    1],
            [1    1    1    1    0    0    0    0],
             ];
           

What do I have to do now? I am using CPLEX for 2 months and this is really high level for me... I really need to solve my model because it's a projetc forthe university...

 

Thank you so much.

 

Regards,

 

Belén

#CPLEXOptimizers
#DecisionOptimization

There is an example that uses warmstart in OPL: opl/examples/opl/warmstart/warmstart.mod.

Can you please take a look at this and see if it answers your questions? If not, please come back here.

#CPLEXOptimizers
#DecisionOptimization