Decision Optimization

Using your script, I have successfully exported most of tuple data into csv files, but I see two errors. I imagined a smooth transitioning from OPL environment to WML, but it is not. Sad..

1. I am using array-tuple in OPL. I manually exported (since your script does support this format) and fed WML, but I got an following error.

[2021-03-10T23:03:38Z, INFO] Element 'DataK' provided in CSV data source cannot be mapped to OPL type: 'ARRAY' as this type is not supported for input data on the Cloud.

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataK[Mchs];

2. I am also using a following tuple format that is not supported by your script.

tuple t_Mx {
key int job;
{int} mch;
}; {t_Mx}

[2021-03-10T23:03:38Z, SEVERE] expecting a collection and found an integer/float

Can you please suggest any solution?
Thanks,
Andy

    range r=1..2;        int a1[r]=[1,2];    int a2[r]=[1,4];    {int} set1={5,6};    {int} set2={9,10};    tuple t    {    string firstname;    int number[r];    {int} intset;    }    {t} s={<"Nicolas",a1,set1>,<"Alexander",a2,set2>};        tuple t2    {      string firstname;      int keynumber;      int number;      int intset;    }        {t2} s2;        execute    {      for(var i in s)        for(var j in r)           for(var k in i.intset)             s2.add(i.firstname,j,i.number[j],k)      for(var l in s2 )writeln(l);    }  

which gives

 <"Nicolas" 1 1 5> <"Nicolas" 1 1 6> <"Nicolas" 2 2 5> <"Nicolas" 2 2 6> <"Alexander" 1 1 9> <"Alexander" 1 1 10> <"Alexander" 2 4 9> <"Alexander" 2 4 10>

More scripting tricks in https://www.linkedin.com/pulse/javascript-within-opl-cplex-alex-fleischer/

Using your script, I have successfully exported most of tuple data into csv files, but I see two errors. I imagined a smooth transitioning from OPL environment to WML, but it is not. Sad..

1. I am using array-tuple in OPL. I manually exported (since your script does support this format) and fed WML, but I got an following error.

[2021-03-10T23:03:38Z, INFO] Element 'DataK' provided in CSV data source cannot be mapped to OPL type: 'ARRAY' as this type is not supported for input data on the Cloud.

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataK[Mchs];

2. I am also using a following tuple format that is not supported by your script.

tuple t_Mx {
key int job;
{int} mch;
}; {t_Mx}

[2021-03-10T23:03:38Z, SEVERE] expecting a collection and found an integer/float

Can you please suggest any solution?
Thanks,
Andy

    range r=1..2;        int a1[r]=[1,2];    int a2[r]=[1,4];    {int} set1={5,6};    {int} set2={9,10};    tuple t    {    string firstname;    int number[r];    {int} intset;    }    {t} s={<"Nicolas",a1,set1>,<"Alexander",a2,set2>};        tuple t2    {      string firstname;      int keynumber;      int number;      int intset;    }        {t2} s2;        execute    {      for(var i in s)        for(var j in r)           for(var k in i.intset)             s2.add(i.firstname,j,i.number[j],k)      for(var l in s2 )writeln(l);    }  

which gives

 <"Nicolas" 1 1 5> <"Nicolas" 1 1 6> <"Nicolas" 2 2 5> <"Nicolas" 2 2 6> <"Alexander" 1 1 9> <"Alexander" 1 1 10> <"Alexander" 2 4 9> <"Alexander" 2 4 10>

More scripting tricks in https://www.linkedin.com/pulse/javascript-within-opl-cplex-alex-fleischer/

Using your script, I have successfully exported most of tuple data into csv files, but I see two errors. I imagined a smooth transitioning from OPL environment to WML, but it is not. Sad..

1. I am using array-tuple in OPL. I manually exported (since your script does support this format) and fed WML, but I got an following error.

[2021-03-10T23:03:38Z, INFO] Element 'DataK' provided in CSV data source cannot be mapped to OPL type: 'ARRAY' as this type is not supported for input data on the Cloud.

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataK[Mchs];

2. I am also using a following tuple format that is not supported by your script.

tuple t_Mx {
key int job;
{int} mch;
}; {t_Mx}

[2021-03-10T23:03:38Z, SEVERE] expecting a collection and found an integer/float

Can you please suggest any solution?
Thanks,
Andy

I flatten the array tuple by adding machine ID into b1(1000*MchID+b1) and b2 (1000*MchID+b2) fields and assign "types" of (mch.id*1000+jid) to a sequence. It works ok with a small problem, but it got really slow when the size gets bigger. How to avoid this performance issue coming from flattening tuples?

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataK[Mchs];

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataFlatten; //

    range r=1..2;        int a1[r]=[1,2];    int a2[r]=[1,4];    {int} set1={5,6};    {int} set2={9,10};    tuple t    {    string firstname;    int number[r];    {int} intset;    }    {t} s={<"Nicolas",a1,set1>,<"Alexander",a2,set2>};        tuple t2    {      string firstname;      int keynumber;      int number;      int intset;    }        {t2} s2;        execute    {      for(var i in s)        for(var j in r)           for(var k in i.intset)             s2.add(i.firstname,j,i.number[j],k)      for(var l in s2 )writeln(l);    }  

which gives

 <"Nicolas" 1 1 5> <"Nicolas" 1 1 6> <"Nicolas" 2 2 5> <"Nicolas" 2 2 6> <"Alexander" 1 1 9> <"Alexander" 1 1 10> <"Alexander" 2 4 9> <"Alexander" 2 4 10>

More scripting tricks in https://www.linkedin.com/pulse/javascript-within-opl-cplex-alex-fleischer/

Using your script, I have successfully exported most of tuple data into csv files, but I see two errors. I imagined a smooth transitioning from OPL environment to WML, but it is not. Sad..

1. I am using array-tuple in OPL. I manually exported (since your script does support this format) and fed WML, but I got an following error.

[2021-03-10T23:03:38Z, INFO] Element 'DataK' provided in CSV data source cannot be mapped to OPL type: 'ARRAY' as this type is not supported for input data on the Cloud.

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataK[Mchs];

2. I am also using a following tuple format that is not supported by your script.

tuple t_Mx {
key int job;
{int} mch;
}; {t_Mx}

[2021-03-10T23:03:38Z, SEVERE] expecting a collection and found an integer/float

Can you please suggest any solution?
Thanks,
Andy

/*here we ll see how to turn a non flat tuple set with arrays, sub tuple and sets insideinto a flat tuple set that can be tuen into a CSV that can be used withinCloudpak for Data Watson Studio*/    execute    {            function isTupleSetFlat(tupleSet)      {        var nbFields=tupleSet.getNFields();        var fistLine=Opl.first(tupleSet);        var res=1; // True unless proved false                for(var j=0;j<nbFields;j++)        {          var value=fistLine[tupleSet.getFieldName(j)];          var typ=typeof(value);          writeln(tupleSet.getFieldName(j), " is ",typ);          if ((typ!="number") && (typ!="string"))  res=0;        }           if (res==1) writeln("This tuple set is flat");        else  writeln("This tuple set is not flat");         return res;    }  }                range r=1..2;        int a1[r]=[1,2];    int a2[r]=[1,4];    {int} set1={5,6};    {int} set2={9,10};        tuple subt    {      int x;      int y;    }        tuple t    {    string firstname;    int number[r];    {int} intset;    subt sub;    }        // not flat tuple set    {t} s={<"Nicolas",a1,set1,<1,2>>,<"Alexander",a2,set2,<6,7>>};        tuple t2    {      string firstname;      int keynumber;      int number;      int intset;      int sub1;      int sub2;    }        {t2} s2;        // turn s that is not flat into s2 that is a flat tuple set    execute    {      writeln();      writeln("is tuple set s flat ? ");      writeln(isTupleSetFlat(s));      writeln();            for(var i in s)        for(var j in r)           for(var k in i.intset)             s2.add(i.firstname,j,i.number[j],k,i.sub.x,i.sub.y)                              writeln();      writeln("is tuple set s2 flat ? ");      writeln(isTupleSetFlat(s2));      writeln();    }        // the same but through OPL modeling and not scripting    {t2} s2bis=union (i in s) union(j in r) union(k in i.intset) {<i.firstname,j,i.number[j],k,i.sub.x,i.sub.y>};        execute display_s2    {      writeln("s=");      writeln(s);      writeln();      writeln("s2=");      for(var l in s2 )writeln(l);      writeln();      writeln("s2bis=");      for(var l in s2bis )writeln(l);    }      /*    gives    is tuple set s flat ? firstname is stringnumber is IloMapintset is IloDiscreteDataCollectionsub is IloTupleThis tuple set is not flat0is tuple set s2 flat ? firstname is stringkeynumber is numbernumber is numberintset is numbersub1 is numbersub2 is numberThis tuple set is flat1s= {<"Nicolas" [1 2] {5 6} <1 2>>     <"Alexander" [1 4] {9 10} <6 7>>}s2= <"Nicolas" 1 1 5 1 2> <"Nicolas" 1 1 6 1 2> <"Nicolas" 2 2 5 1 2> <"Nicolas" 2 2 6 1 2> <"Alexander" 1 1 9 6 7> <"Alexander" 1 1 10 6 7> <"Alexander" 2 4 9 6 7> <"Alexander" 2 4 10 6 7>  s2bis= <"Nicolas" 1 1 5 1 2> <"Nicolas" 1 1 6 1 2> <"Nicolas" 2 2 5 1 2> <"Nicolas" 2 2 6 1 2> <"Alexander" 1 1 9 6 7> <"Alexander" 1 1 10 6 7> <"Alexander" 2 4 9 6 7> <"Alexander" 2 4 10 6 7>  */      ​

I flatten the array tuple by adding machine ID into b1(1000*MchID+b1) and b2 (1000*MchID+b2) fields and assign "types" of (mch.id*1000+jid) to a sequence. It works ok with a small problem, but it got really slow when the size gets bigger. How to avoid this performance issue coming from flattening tuples?

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataK[Mchs];

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataFlatten; //

    range r=1..2;        int a1[r]=[1,2];    int a2[r]=[1,4];    {int} set1={5,6};    {int} set2={9,10};    tuple t    {    string firstname;    int number[r];    {int} intset;    }    {t} s={<"Nicolas",a1,set1>,<"Alexander",a2,set2>};        tuple t2    {      string firstname;      int keynumber;      int number;      int intset;    }        {t2} s2;        execute    {      for(var i in s)        for(var j in r)           for(var k in i.intset)             s2.add(i.firstname,j,i.number[j],k)      for(var l in s2 )writeln(l);    }  

which gives

 <"Nicolas" 1 1 5> <"Nicolas" 1 1 6> <"Nicolas" 2 2 5> <"Nicolas" 2 2 6> <"Alexander" 1 1 9> <"Alexander" 1 1 10> <"Alexander" 2 4 9> <"Alexander" 2 4 10>

More scripting tricks in https://www.linkedin.com/pulse/javascript-within-opl-cplex-alex-fleischer/

Using your script, I have successfully exported most of tuple data into csv files, but I see two errors. I imagined a smooth transitioning from OPL environment to WML, but it is not. Sad..

1. I am using array-tuple in OPL. I manually exported (since your script does support this format) and fed WML, but I got an following error.

[2021-03-10T23:03:38Z, INFO] Element 'DataK' provided in CSV data source cannot be mapped to OPL type: 'ARRAY' as this type is not supported for input data on the Cloud.

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataK[Mchs];

2. I am also using a following tuple format that is not supported by your script.

tuple t_Mx {
key int job;
{int} mch;
}; {t_Mx}

[2021-03-10T23:03:38Z, SEVERE] expecting a collection and found an integer/float

Can you please suggest any solution?
Thanks,
Andy

/*here we ll see how to turn a non flat tuple set with arrays, sub tuple and sets insideinto a flat tuple set that can be tuen into a CSV that can be used withinCloudpak for Data Watson Studio*/    execute    {            function isTupleSetFlat(tupleSet)      {        var nbFields=tupleSet.getNFields();        var fistLine=Opl.first(tupleSet);        var res=1; // True unless proved false                for(var j=0;j<nbFields;j++)        {          var value=fistLine[tupleSet.getFieldName(j)];          var typ=typeof(value);          writeln(tupleSet.getFieldName(j), " is ",typ);          if ((typ!="number") && (typ!="string"))  res=0;        }           if (res==1) writeln("This tuple set is flat");        else  writeln("This tuple set is not flat");         return res;    }  }                range r=1..2;        int a1[r]=[1,2];    int a2[r]=[1,4];    {int} set1={5,6};    {int} set2={9,10};        tuple subt    {      int x;      int y;    }        tuple t    {    string firstname;    int number[r];    {int} intset;    subt sub;    }        // not flat tuple set    {t} s={<"Nicolas",a1,set1,<1,2>>,<"Alexander",a2,set2,<6,7>>};        tuple t2    {      string firstname;      int keynumber;      int number;      int intset;      int sub1;      int sub2;    }        {t2} s2;        // turn s that is not flat into s2 that is a flat tuple set    execute    {      writeln();      writeln("is tuple set s flat ? ");      writeln(isTupleSetFlat(s));      writeln();            for(var i in s)        for(var j in r)           for(var k in i.intset)             s2.add(i.firstname,j,i.number[j],k,i.sub.x,i.sub.y)                              writeln();      writeln("is tuple set s2 flat ? ");      writeln(isTupleSetFlat(s2));      writeln();    }        // the same but through OPL modeling and not scripting    {t2} s2bis=union (i in s) union(j in r) union(k in i.intset) {<i.firstname,j,i.number[j],k,i.sub.x,i.sub.y>};        execute display_s2    {      writeln("s=");      writeln(s);      writeln();      writeln("s2=");      for(var l in s2 )writeln(l);      writeln();      writeln("s2bis=");      for(var l in s2bis )writeln(l);    }      /*    gives    is tuple set s flat ? firstname is stringnumber is IloMapintset is IloDiscreteDataCollectionsub is IloTupleThis tuple set is not flat0is tuple set s2 flat ? firstname is stringkeynumber is numbernumber is numberintset is numbersub1 is numbersub2 is numberThis tuple set is flat1s= {<"Nicolas" [1 2] {5 6} <1 2>>     <"Alexander" [1 4] {9 10} <6 7>>}s2= <"Nicolas" 1 1 5 1 2> <"Nicolas" 1 1 6 1 2> <"Nicolas" 2 2 5 1 2> <"Nicolas" 2 2 6 1 2> <"Alexander" 1 1 9 6 7> <"Alexander" 1 1 10 6 7> <"Alexander" 2 4 9 6 7> <"Alexander" 2 4 10 6 7>  s2bis= <"Nicolas" 1 1 5 1 2> <"Nicolas" 1 1 6 1 2> <"Nicolas" 2 2 5 1 2> <"Nicolas" 2 2 6 1 2> <"Alexander" 1 1 9 6 7> <"Alexander" 1 1 10 6 7> <"Alexander" 2 4 9 6 7> <"Alexander" 2 4 10 6 7>  */      ​

I flatten the array tuple by adding machine ID into b1(1000*MchID+b1) and b2 (1000*MchID+b2) fields and assign "types" of (mch.id*1000+jid) to a sequence. It works ok with a small problem, but it got really slow when the size gets bigger. How to avoid this performance issue coming from flattening tuples?

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataK[Mchs];

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataFlatten; //

    range r=1..2;        int a1[r]=[1,2];    int a2[r]=[1,4];    {int} set1={5,6};    {int} set2={9,10};    tuple t    {    string firstname;    int number[r];    {int} intset;    }    {t} s={<"Nicolas",a1,set1>,<"Alexander",a2,set2>};        tuple t2    {      string firstname;      int keynumber;      int number;      int intset;    }        {t2} s2;        execute    {      for(var i in s)        for(var j in r)           for(var k in i.intset)             s2.add(i.firstname,j,i.number[j],k)      for(var l in s2 )writeln(l);    }  

which gives

 <"Nicolas" 1 1 5> <"Nicolas" 1 1 6> <"Nicolas" 2 2 5> <"Nicolas" 2 2 6> <"Alexander" 1 1 9> <"Alexander" 1 1 10> <"Alexander" 2 4 9> <"Alexander" 2 4 10>

More scripting tricks in https://www.linkedin.com/pulse/javascript-within-opl-cplex-alex-fleischer/

Using your script, I have successfully exported most of tuple data into csv files, but I see two errors. I imagined a smooth transitioning from OPL environment to WML, but it is not. Sad..

1. I am using array-tuple in OPL. I manually exported (since your script does support this format) and fed WML, but I got an following error.

[2021-03-10T23:03:38Z, INFO] Element 'DataK' provided in CSV data source cannot be mapped to OPL type: 'ARRAY' as this type is not supported for input data on the Cloud.

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataK[Mchs];

2. I am also using a following tuple format that is not supported by your script.

tuple t_Mx {
key int job;
{int} mch;
}; {t_Mx}

[2021-03-10T23:03:38Z, SEVERE] expecting a collection and found an integer/float

Can you please suggest any solution?
Thanks,
Andy

I have successfully tested a same code on both WML (w/ basic configuration) and local OPL. I don't believe this result. WML cloud delivered a slower computation compared to my own laptop. I will try WML with full configuration in coming weeks, but I cannot understand.

Maximization problem staticLex (A, B)

WML Cloud
[2021-03-12T19:53:51Z, INFO] ! Time = 600.02s, Average fail depth = 245, Memory usage = 729.2 MB
[2021-03-12T19:53:51Z, INFO] ! Current objective is 594; 17

Laptop with i9 and 32 G memory on OPL 20.1
New objective is 604; 11
! Time = 595.67s, Average fail depth = 254, Memory usage = 7.8 GB

/*here we ll see how to turn a non flat tuple set with arrays, sub tuple and sets insideinto a flat tuple set that can be tuen into a CSV that can be used withinCloudpak for Data Watson Studio*/    execute    {            function isTupleSetFlat(tupleSet)      {        var nbFields=tupleSet.getNFields();        var fistLine=Opl.first(tupleSet);        var res=1; // True unless proved false                for(var j=0;j<nbFields;j++)        {          var value=fistLine[tupleSet.getFieldName(j)];          var typ=typeof(value);          writeln(tupleSet.getFieldName(j), " is ",typ);          if ((typ!="number") && (typ!="string"))  res=0;        }           if (res==1) writeln("This tuple set is flat");        else  writeln("This tuple set is not flat");         return res;    }  }                range r=1..2;        int a1[r]=[1,2];    int a2[r]=[1,4];    {int} set1={5,6};    {int} set2={9,10};        tuple subt    {      int x;      int y;    }        tuple t    {    string firstname;    int number[r];    {int} intset;    subt sub;    }        // not flat tuple set    {t} s={<"Nicolas",a1,set1,<1,2>>,<"Alexander",a2,set2,<6,7>>};        tuple t2    {      string firstname;      int keynumber;      int number;      int intset;      int sub1;      int sub2;    }        {t2} s2;        // turn s that is not flat into s2 that is a flat tuple set    execute    {      writeln();      writeln("is tuple set s flat ? ");      writeln(isTupleSetFlat(s));      writeln();            for(var i in s)        for(var j in r)           for(var k in i.intset)             s2.add(i.firstname,j,i.number[j],k,i.sub.x,i.sub.y)                              writeln();      writeln("is tuple set s2 flat ? ");      writeln(isTupleSetFlat(s2));      writeln();    }        // the same but through OPL modeling and not scripting    {t2} s2bis=union (i in s) union(j in r) union(k in i.intset) {<i.firstname,j,i.number[j],k,i.sub.x,i.sub.y>};        execute display_s2    {      writeln("s=");      writeln(s);      writeln();      writeln("s2=");      for(var l in s2 )writeln(l);      writeln();      writeln("s2bis=");      for(var l in s2bis )writeln(l);    }      /*    gives    is tuple set s flat ? firstname is stringnumber is IloMapintset is IloDiscreteDataCollectionsub is IloTupleThis tuple set is not flat0is tuple set s2 flat ? firstname is stringkeynumber is numbernumber is numberintset is numbersub1 is numbersub2 is numberThis tuple set is flat1s= {<"Nicolas" [1 2] {5 6} <1 2>>     <"Alexander" [1 4] {9 10} <6 7>>}s2= <"Nicolas" 1 1 5 1 2> <"Nicolas" 1 1 6 1 2> <"Nicolas" 2 2 5 1 2> <"Nicolas" 2 2 6 1 2> <"Alexander" 1 1 9 6 7> <"Alexander" 1 1 10 6 7> <"Alexander" 2 4 9 6 7> <"Alexander" 2 4 10 6 7>  s2bis= <"Nicolas" 1 1 5 1 2> <"Nicolas" 1 1 6 1 2> <"Nicolas" 2 2 5 1 2> <"Nicolas" 2 2 6 1 2> <"Alexander" 1 1 9 6 7> <"Alexander" 1 1 10 6 7> <"Alexander" 2 4 9 6 7> <"Alexander" 2 4 10 6 7>  */      ​

I flatten the array tuple by adding machine ID into b1(1000*MchID+b1) and b2 (1000*MchID+b2) fields and assign "types" of (mch.id*1000+jid) to a sequence. It works ok with a small problem, but it got really slow when the size gets bigger. How to avoid this performance issue coming from flattening tuples?

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataK[Mchs];

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataFlatten; //

    range r=1..2;        int a1[r]=[1,2];    int a2[r]=[1,4];    {int} set1={5,6};    {int} set2={9,10};    tuple t    {    string firstname;    int number[r];    {int} intset;    }    {t} s={<"Nicolas",a1,set1>,<"Alexander",a2,set2>};        tuple t2    {      string firstname;      int keynumber;      int number;      int intset;    }        {t2} s2;        execute    {      for(var i in s)        for(var j in r)           for(var k in i.intset)             s2.add(i.firstname,j,i.number[j],k)      for(var l in s2 )writeln(l);    }  

which gives

 <"Nicolas" 1 1 5> <"Nicolas" 1 1 6> <"Nicolas" 2 2 5> <"Nicolas" 2 2 6> <"Alexander" 1 1 9> <"Alexander" 1 1 10> <"Alexander" 2 4 9> <"Alexander" 2 4 10>

More scripting tricks in https://www.linkedin.com/pulse/javascript-within-opl-cplex-alex-fleischer/

Using your script, I have successfully exported most of tuple data into csv files, but I see two errors. I imagined a smooth transitioning from OPL environment to WML, but it is not. Sad..

1. I am using array-tuple in OPL. I manually exported (since your script does support this format) and fed WML, but I got an following error.

[2021-03-10T23:03:38Z, INFO] Element 'DataK' provided in CSV data source cannot be mapped to OPL type: 'ARRAY' as this type is not supported for input data on the Cloud.

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataK[Mchs];

2. I am also using a following tuple format that is not supported by your script.

tuple t_Mx {
key int job;
{int} mch;
}; {t_Mx}

[2021-03-10T23:03:38Z, SEVERE] expecting a collection and found an integer/float

Can you please suggest any solution?
Thanks,
Andy

I have successfully tested a same code on both WML (w/ basic configuration) and local OPL. I don't believe this result. WML cloud delivered a slower computation compared to my own laptop. I will try WML with full configuration in coming weeks, but I cannot understand.

Maximization problem staticLex (A, B)

WML Cloud
[2021-03-12T19:53:51Z, INFO] ! Time = 600.02s, Average fail depth = 245, Memory usage = 729.2 MB
[2021-03-12T19:53:51Z, INFO] ! Current objective is 594; 17

Laptop with i9 and 32 G memory on OPL 20.1
New objective is 604; 11
! Time = 595.67s, Average fail depth = 254, Memory usage = 7.8 GB

/*here we ll see how to turn a non flat tuple set with arrays, sub tuple and sets insideinto a flat tuple set that can be tuen into a CSV that can be used withinCloudpak for Data Watson Studio*/    execute    {            function isTupleSetFlat(tupleSet)      {        var nbFields=tupleSet.getNFields();        var fistLine=Opl.first(tupleSet);        var res=1; // True unless proved false                for(var j=0;j<nbFields;j++)        {          var value=fistLine[tupleSet.getFieldName(j)];          var typ=typeof(value);          writeln(tupleSet.getFieldName(j), " is ",typ);          if ((typ!="number") && (typ!="string"))  res=0;        }           if (res==1) writeln("This tuple set is flat");        else  writeln("This tuple set is not flat");         return res;    }  }                range r=1..2;        int a1[r]=[1,2];    int a2[r]=[1,4];    {int} set1={5,6};    {int} set2={9,10};        tuple subt    {      int x;      int y;    }        tuple t    {    string firstname;    int number[r];    {int} intset;    subt sub;    }        // not flat tuple set    {t} s={<"Nicolas",a1,set1,<1,2>>,<"Alexander",a2,set2,<6,7>>};        tuple t2    {      string firstname;      int keynumber;      int number;      int intset;      int sub1;      int sub2;    }        {t2} s2;        // turn s that is not flat into s2 that is a flat tuple set    execute    {      writeln();      writeln("is tuple set s flat ? ");      writeln(isTupleSetFlat(s));      writeln();            for(var i in s)        for(var j in r)           for(var k in i.intset)             s2.add(i.firstname,j,i.number[j],k,i.sub.x,i.sub.y)                              writeln();      writeln("is tuple set s2 flat ? ");      writeln(isTupleSetFlat(s2));      writeln();    }        // the same but through OPL modeling and not scripting    {t2} s2bis=union (i in s) union(j in r) union(k in i.intset) {<i.firstname,j,i.number[j],k,i.sub.x,i.sub.y>};        execute display_s2    {      writeln("s=");      writeln(s);      writeln();      writeln("s2=");      for(var l in s2 )writeln(l);      writeln();      writeln("s2bis=");      for(var l in s2bis )writeln(l);    }      /*    gives    is tuple set s flat ? firstname is stringnumber is IloMapintset is IloDiscreteDataCollectionsub is IloTupleThis tuple set is not flat0is tuple set s2 flat ? firstname is stringkeynumber is numbernumber is numberintset is numbersub1 is numbersub2 is numberThis tuple set is flat1s= {<"Nicolas" [1 2] {5 6} <1 2>>     <"Alexander" [1 4] {9 10} <6 7>>}s2= <"Nicolas" 1 1 5 1 2> <"Nicolas" 1 1 6 1 2> <"Nicolas" 2 2 5 1 2> <"Nicolas" 2 2 6 1 2> <"Alexander" 1 1 9 6 7> <"Alexander" 1 1 10 6 7> <"Alexander" 2 4 9 6 7> <"Alexander" 2 4 10 6 7>  s2bis= <"Nicolas" 1 1 5 1 2> <"Nicolas" 1 1 6 1 2> <"Nicolas" 2 2 5 1 2> <"Nicolas" 2 2 6 1 2> <"Alexander" 1 1 9 6 7> <"Alexander" 1 1 10 6 7> <"Alexander" 2 4 9 6 7> <"Alexander" 2 4 10 6 7>  */      ​

I flatten the array tuple by adding machine ID into b1(1000*MchID+b1) and b2 (1000*MchID+b2) fields and assign "types" of (mch.id*1000+jid) to a sequence. It works ok with a small problem, but it got really slow when the size gets bigger. How to avoid this performance issue coming from flattening tuples?

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataK[Mchs];

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataFlatten; //

    range r=1..2;        int a1[r]=[1,2];    int a2[r]=[1,4];    {int} set1={5,6};    {int} set2={9,10};    tuple t    {    string firstname;    int number[r];    {int} intset;    }    {t} s={<"Nicolas",a1,set1>,<"Alexander",a2,set2>};        tuple t2    {      string firstname;      int keynumber;      int number;      int intset;    }        {t2} s2;        execute    {      for(var i in s)        for(var j in r)           for(var k in i.intset)             s2.add(i.firstname,j,i.number[j],k)      for(var l in s2 )writeln(l);    }  

which gives

 <"Nicolas" 1 1 5> <"Nicolas" 1 1 6> <"Nicolas" 2 2 5> <"Nicolas" 2 2 6> <"Alexander" 1 1 9> <"Alexander" 1 1 10> <"Alexander" 2 4 9> <"Alexander" 2 4 10>

More scripting tricks in https://www.linkedin.com/pulse/javascript-within-opl-cplex-alex-fleischer/

Using your script, I have successfully exported most of tuple data into csv files, but I see two errors. I imagined a smooth transitioning from OPL environment to WML, but it is not. Sad..

1. I am using array-tuple in OPL. I manually exported (since your script does support this format) and fed WML, but I got an following error.

[2021-03-10T23:03:38Z, INFO] Element 'DataK' provided in CSV data source cannot be mapped to OPL type: 'ARRAY' as this type is not supported for input data on the Cloud.

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataK[Mchs];

2. I am also using a following tuple format that is not supported by your script.

tuple t_Mx {
key int job;
{int} mch;
}; {t_Mx}

[2021-03-10T23:03:38Z, SEVERE] expecting a collection and found an integer/float

Can you please suggest any solution?
Thanks,
Andy

I have successfully tested a same code on both WML (w/ basic configuration) and local OPL. I don't believe this result. WML cloud delivered a slower computation compared to my own laptop. I will try WML with full configuration in coming weeks, but I cannot understand.

Maximization problem staticLex (A, B)

WML Cloud
[2021-03-12T19:53:51Z, INFO] ! Time = 600.02s, Average fail depth = 245, Memory usage = 729.2 MB
[2021-03-12T19:53:51Z, INFO] ! Current objective is 594; 17

Laptop with i9 and 32 G memory on OPL 20.1
New objective is 604; 11
! Time = 595.67s, Average fail depth = 254, Memory usage = 7.8 GB

/*here we ll see how to turn a non flat tuple set with arrays, sub tuple and sets insideinto a flat tuple set that can be tuen into a CSV that can be used withinCloudpak for Data Watson Studio*/    execute    {            function isTupleSetFlat(tupleSet)      {        var nbFields=tupleSet.getNFields();        var fistLine=Opl.first(tupleSet);        var res=1; // True unless proved false                for(var j=0;j<nbFields;j++)        {          var value=fistLine[tupleSet.getFieldName(j)];          var typ=typeof(value);          writeln(tupleSet.getFieldName(j), " is ",typ);          if ((typ!="number") && (typ!="string"))  res=0;        }           if (res==1) writeln("This tuple set is flat");        else  writeln("This tuple set is not flat");         return res;    }  }                range r=1..2;        int a1[r]=[1,2];    int a2[r]=[1,4];    {int} set1={5,6};    {int} set2={9,10};        tuple subt    {      int x;      int y;    }        tuple t    {    string firstname;    int number[r];    {int} intset;    subt sub;    }        // not flat tuple set    {t} s={<"Nicolas",a1,set1,<1,2>>,<"Alexander",a2,set2,<6,7>>};        tuple t2    {      string firstname;      int keynumber;      int number;      int intset;      int sub1;      int sub2;    }        {t2} s2;        // turn s that is not flat into s2 that is a flat tuple set    execute    {      writeln();      writeln("is tuple set s flat ? ");      writeln(isTupleSetFlat(s));      writeln();            for(var i in s)        for(var j in r)           for(var k in i.intset)             s2.add(i.firstname,j,i.number[j],k,i.sub.x,i.sub.y)                              writeln();      writeln("is tuple set s2 flat ? ");      writeln(isTupleSetFlat(s2));      writeln();    }        // the same but through OPL modeling and not scripting    {t2} s2bis=union (i in s) union(j in r) union(k in i.intset) {<i.firstname,j,i.number[j],k,i.sub.x,i.sub.y>};        execute display_s2    {      writeln("s=");      writeln(s);      writeln();      writeln("s2=");      for(var l in s2 )writeln(l);      writeln();      writeln("s2bis=");      for(var l in s2bis )writeln(l);    }      /*    gives    is tuple set s flat ? firstname is stringnumber is IloMapintset is IloDiscreteDataCollectionsub is IloTupleThis tuple set is not flat0is tuple set s2 flat ? firstname is stringkeynumber is numbernumber is numberintset is numbersub1 is numbersub2 is numberThis tuple set is flat1s= {<"Nicolas" [1 2] {5 6} <1 2>>     <"Alexander" [1 4] {9 10} <6 7>>}s2= <"Nicolas" 1 1 5 1 2> <"Nicolas" 1 1 6 1 2> <"Nicolas" 2 2 5 1 2> <"Nicolas" 2 2 6 1 2> <"Alexander" 1 1 9 6 7> <"Alexander" 1 1 10 6 7> <"Alexander" 2 4 9 6 7> <"Alexander" 2 4 10 6 7>  s2bis= <"Nicolas" 1 1 5 1 2> <"Nicolas" 1 1 6 1 2> <"Nicolas" 2 2 5 1 2> <"Nicolas" 2 2 6 1 2> <"Alexander" 1 1 9 6 7> <"Alexander" 1 1 10 6 7> <"Alexander" 2 4 9 6 7> <"Alexander" 2 4 10 6 7>  */      ​

I flatten the array tuple by adding machine ID into b1(1000*MchID+b1) and b2 (1000*MchID+b2) fields and assign "types" of (mch.id*1000+jid) to a sequence. It works ok with a small problem, but it got really slow when the size gets bigger. How to avoid this performance issue coming from flattening tuples?

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataK[Mchs];

tuple t_Travel {
key int b1; key int b2; int d;
}; {t_Travel} DataFlatten; //