Originally posted by: Bahman

Hi Everyone,

Considering the following codes written in Cplex Studio IDE, I would ask you please advise me how I can write a min max function

 

tuple s {

  int i;

  int j;

}

{s} S=...;

{string} G=...;

{int} k=asSet(1..3);

{int} T=asSet(1..6);

float P=...;

float d[K][G]=...;

 

dvar boolean Z[S][K][T];

dexpr float PE[<i,j> in S][g in G]=sum (k in K, t in T) Z[<i,j>][K][T]*d[k][g]*P;

.............................................................................................................................

 

Now I would like to write two different codes with the following two objectives (for all <i,j> in S, k in K, t in T and g in G):

1) Minimize the maximum PE[<i,j>][g];

2) Minimize the maximum PE[<i,j>][g] - minimum PE[<i,j>][g]

 

I would appreciate your reply!

 

Regards
Bahman

#CPLEXOptimizers
#DecisionOptimization

Hi,

have you had a look at https://www.ibm.com/developerworks/community/forums/html/topic?id=a1eeb4cb-d62e-404a-9113-3395cf1f5c21&ps=25

?

regards

#CPLEXOptimizers
#DecisionOptimization

Originally posted by: Bahman

Dear Alex,

I appreciate your prompt reply! I would ask you if it matters if I make use of maxl or minl instead of max and min functions, respectively?

Regards,
Bahman

#CPLEXOptimizers
#DecisionOptimization

Hi

It does not matter:

range r=1..5;
dvar int x[i in r] in r;

//minimize max(i in r) x[i];
minimize maxl(x[1],x[2],x[3],x[4],x[5]);
subject to
{

}

regards

#CPLEXOptimizers
#DecisionOptimization

Originally posted by: Bahman

Dear Alex,

I got the difference! Thank you!

 

Regards
Bahman

#CPLEXOptimizers
#DecisionOptimization

In your case you can model this directly without using OPL's min() or max() function:

1. You can use an auxiliary variable PEmax and a constraint PEmax >= PE[<i,j>][g]. Then minimizing PEmax gives the desired result.
2. Similarly to the previous case, use auxiliary variable PEdist and constraint PEdist >= PE[<i1,j1>][g1] - PE[<i2,j2>][g2] for all combinations of i, j, g. Then again, minimizing PEmax gives the desired result.

#CPLEXOptimizers
#DecisionOptimization

Originally posted by: Bahman

Dear Daniel, 

I have written the following constraint but I am not sure if it works well:

 

Regards,
Bahman

#CPLEXOptimizers
#DecisionOptimization

Originally posted by: NALK

Dear Daniel and ALEX, 

Is there any specific OPL syntax  to find the minimum of the smallest positive number in an array (non-zero ))??  for example  (  A[j]={2,3,4,0,1})  

Thanks in advance,

 

Kind Regards,

 

#CPLEXOptimizers
#DecisionOptimization

Hi

{int}  A={2,3,4,0,1};

int minNonZeroPos=min(i in A:i>=1) i;
execute
{
writeln(minNonZeroPos);
}

gives

1

but please open new threads for new topics

regards

#CPLEXOptimizers
#DecisionOptimization

Originally posted by: NALK

Thank you Alex  , Thats good but it is still valid for such for such  expression  Min (i in 1..5 ) A_i*X_i         such that       A_i=[2,3,4,0,1]  and   X_i=[1,0,1,0,0.5]       X is a decisiion variable 

 

Thanks a lot, 

#CPLEXOptimizers
#DecisionOptimization

Hi

then you may write

range r=1..5;

int  A[r]=[2,3,4,0,1];
float v[r]=[1,0,1,0,0.5] ;

dvar float X[r];
dvar float E[r];
dvar float res;

subject to
{
forall(i in r) X[i]==v[i];

forall(i in r) (A[i]*X[i]>=0.001) => (E[i]==A[i]*X[i]);
forall(i in r) (A[i]*X[i]<=0.001) => (E[i]==maxint);

res==min (i in 1..5 ) (E[i]) ;
}

execute
{
writeln(res);
}

 

which gives 0.5

 

But again please open new threads for new topics

regards

 

https://www.linkedin.com/pulse/how-opl-alex-fleischer/

#CPLEXOptimizers
#DecisionOptimization

Originally posted by: NALK

Thanks a lot Alex  ,

#CPLEXOptimizers
#DecisionOptimization