[FlopCpp] such_that functionality...

[Vishy Jeet]

Thanks Tim, that worked.

For the benefit of those who are reading this thread in future: you must run
the following three commands all over again anytime you make changes in
source code:

>> make
>> make test
>> make install

this would reinstall the affected libraries...

Once again Tim, this was really great help.

regards,
vishy


On 12/20/06, Tim Hultberg <Tim.Hultberg at eumetsat.int> wrote:
>
> Hi Vishy,
>
> a) you probably want to use: such_that(N < N.size()-1)  (since N
> implicitly counts from zero)
>
> b) the bad news is that this wont work either with the current version.
> (it used to though) To correct it go to MP_constraint.cpp
> and find the lines:
>        left->generate(
> S1(I1)*S2(I2)*S3(I3)*S4(I4)*S5(I5).such_that(B),v,f,1.0);
>        right->generate(
> S1(I1)*S2(I2)*S3(I3)*S4(I4)*S5(I5).such_that(B),v,f,-1.0);
> and change them to:
>        left->generate(
> (S1(I1)*S2(I2)*S3(I3)*S4(I4)*S5(I5)).such_that(B),v,f,1.0);
>        right->generate(
> (S1(I1)*S2(I2)*S3(I3)*S4(I4)*S5(I5)).such_that(B),v,f,-1.0);
> I will correct it in the next release. Sorry for the inconvenience.
>
> Let me know if you get any more problems with this.
>
> Cheesr, Tim
>
>
> Tim Hultberg
>
> >>> "Vishy Jeet" <vishv.jeet at gmail.com> 12/19/06 8:53 PM >>>
> I have written the following small prograg to solve GAP on size (3, 10)
>
>      1 // $Id$
>      2 #include "flopc.hpp"
>      3 using namespace flopc;
>      4 #include "OsiCbcSolverInterface.hpp"
>      5
>      6 /* Generalized Assignment Problem */
>      7
>      8 class Gap : public MP_model {
>      9 public:
>     10   MP_set M,N;
>     11   MP_data c;
>     12   MP_data a;
>     13   MP_data b;
>     14   MP_variable x;
>     15   MP_constraint cap, feas;
>     16
>     17   Gap(int numM, int numN) : MP_model(new OsiCbcSolverInterface),
>     18   M(numM), N(numN),
>     19   c(M,N), a(M,N), b(M),
>     20   x(M,N),
>     21   cap(M), feas(N)
>     22   {
>     23     x.binary();
>     24     cap(M) = sum(N, a(M,N)*x(M,N)) <= b(M);
>     25     feas(N).*such_that(N < N.size())* = sum(M, x(M,N)) == 1;
>     26     add(cap); add(feas);
>     27     setObjective( sum(M*N, c(M,N)*x(M,N)));
>     28   }
>     29 };
>     30
>     31 int main() {
>     32     double cost[3][11]  = {{47,16,27,21,25,12,55,31,40,40,1000},
>     33                           {45,18,56,33,36,46,11,17,31,50,1000},
>     34                           {50,23,22,59,31,35,40,41,40,11,1000}};
>     35
>     36     double req[3][11] = {{24,12,8,12,16,14,13,15,23,5,100},
>     37                           {7,17,8,5,6,19,20,7,22,9,100},
>     38                           {23,18,14,8,18,23,19,11,15,11,100}};
>     39
>     40     double capc[3] =  {37,32,42};
>     41
>     42     Gap instance(3,11);
>     43
>     44     instance.c.value(&cost[0][0]);
>     45     instance.a.value(&req[0][0]);
>     46     instance.b.value(&capc[0]);
>     47
>     48     instance.minimize();
>     49
>     50     cout<<"numRow:="<<instance->getNumRows()<<"\n";
>     51     cout<<"numCols:="<<instance->getNumCols()<<"\n";
>     52     cout<<"numEle:="<<instance->getNumElements()<<"\n";
>     53     cout<<"obj:="<<instance->getObjValue()<<"\n";
>     54     instance.x.display("Solution of first instance (x)");
>     55 }
>
> To learn how the such_that statement works, I introduced the 11th job
> with
> very high cost of assignment and infeasible
> requirement on any of the machines...to prevent infeasibility I write
> the
> "feas" constraints using such_that statement so
> that 11th job doesn't have to be assigned....but I am not able to see
> whats
> wrong with the code that it doesn't work.
> Any help on this is much appreciated..
>
> regards,
> vishy
>
>


-- 
--vishy
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://list.coin-or.org/pipermail/flopcpp/attachments/20061220/7b201de2/attachment.html