[FlopCpp] wrote first flopcpp model. have a lot of questions
Mario Rappi
marappi at wichita.edu
Tue Dec 1 20:19:40 EST 2009
Thank you both for your help.
I linearized the objective function using Charnes/Cooper.
Then got the writeLP (which gives me what I was expecting from
constraint.display) to work and confirmed my data wasn't loading, so changed
the indexing from (Input) to (S,I) to load it right.
Then I saw the model was unbounded, reviewed it and realized that when I
earlier linearized the constraints I inadvertently flipped the sign of the
inequalities.
When the problem was unbounded the exe crashed. Is this normal?
Everything works now.
Next step is to loop to solve for each member of the set S and try to read
the data from a database.
Thanks again, FlopC++ is a great product.
Mario
On Tue, Dec 1, 2009 at 10:42 AM, Tim Hultberg <Tim.Hultberg at eumetsat.int>wrote:
> Hi Mario,
>
> here are some answers:
>
>
>
> 1. Yes, that is possible.
>
>
>
> 2. You can output the problem in MPS format or 'lp format'. See the recent
> mails on the list about how to do this.
>
>
>
> 3.
>
>
>
> 4. You cant. Flopc++ can only be used to model LPs and will not allow you
> to divide with an expression involving decision variables.
>
>
>
> 5.
>
>
>
> 6.
>
> *void* MP_constraint*::*display(string s) *const* {
>
> 89
>
> cout*<<*s*<<*endl;
>
> 90
>
> *if* (offset *>=*0) {
>
> 91
>
> *for* (*int* i*=*offset; i*<*offset*+*size(); i*++*) {
>
> 92
>
> cout*<<*i*<<*" "*<<*M*->*bl[i]*<<*" "*<<*M*->*rowActivity[i]*<<*"
> "*<<*M*->*bu[i]*<<*" "*<<*M*->*rowPrice[i]*<<*endl;
>
> 93
>
> }
>
> 94
>
> } *else* {
>
> 95
>
> cout*<<*"No solution available!"*<<*endl;
>
> 96
>
> }
>
> 97
>
> }
>
>
>
>
>
> I know I didn’t answer everthing, but I have to go now..
>
>
>
> Tim
>
>
>
> *From:* flopcpp-bounces at list.coin-or.org [mailto:
> flopcpp-bounces at list.coin-or.org] *On Behalf Of *Mario Rappi
> *Sent:* Sunday, November 29, 2009 2:47 AM
> *To:* flopcpp at list.coin-or.org
> *Subject:* [FlopCpp] wrote first flopcpp model. have a lot of questions
>
>
>
> First I include the model, then the output, and at the end the questions:
>
>
> *****model************************
> // $Id$
> #include "flopc.hpp"
> using namespace flopc;
> #include <OsiClpSolverInterface.hpp>
>
> int main() {
>
> MP_model &model = MP_model::getDefaultModel();
> model.setSolver(new OsiClpSolverInterface);
> model.verbose();
>
> enum {storeA, storeB, storeC, storeD, numS};
> enum {Input1, Input2, numI};
> enum {Output1, numO};
>
> MP_set S(numS); // Stores
> MP_set I(numI); // Inputs
> MP_set O(numO); // Outputs
>
> MP_subset<2> Input(S,I);
> Input.insert(S,I);
> Input.display("Input");
> MP_subset<2> Output(S,O);
> Output.insert(S,O);
> Output.display("Output");
>
> //this will be easier with an array??
> MP_data InputVal(Input);
> InputVal(storeA,Input1)=1; InputVal(storeA,Input2)=3;
> InputVal(storeB,Input1)=3; InputVal(storeB,Input2)=3;
> InputVal(storeC,Input1)=4; InputVal(storeC,Input2)=2;
> InputVal(storeD,Input1)=3; InputVal(storeD,Input2)=1;
>
> MP_data OutputVal(Output);
> OutputVal(storeA,Output1)=1;
> OutputVal(storeB,Output1)=1.5;
> OutputVal(storeC,Output1)=2;
> OutputVal(storeD,Output1)=1;
>
> MP_variable x(I), y(O);
> x.lowerLimit(I)=0;
> y.lowerLimit(O)=0;
>
> MP_constraint efficiency(S);
> efficiency(S) = sum(I, InputVal(S, I) * x(I)) <= sum(O, OutputVal(S, O)
> * y(O));
>
> minimize( (sum(O, OutputVal(storeB, O) * y(O))) - (sum(I,
> InputVal(storeB, I) * x(I))));
>
> efficiency.display("constraints");
> x.display("x");
> y.display("y");
>
> cout<<"Test dea passed."<<endl;
> }
> *****model end************************
>
> *****output************************
> Input
> Output
> FlopCpp: Constraint
> 0 1 0 0
> 1 1 0 0
> 2 1 0 0
> 3 1 0 0
> 0 2 0 0
> 1 2 0 0
> 2 2 0 0
> 3 2 0 0
> 0 0 -0 0
> 1 0 -0 0
> 2 0 -0 0
> 3 0 -0 0
> FlopCpp: Number of constraint blocks: 1
> FlopCpp: Number of individual constraints: 4
> FlopCpp: Number of variable blocks: 2
> FlopCpp: Number of individual variables: 3
> FlopCpp: Number of non-zeroes (including rhs): 12
> Objective
> -1 0 0
> -1 1 -0
> -1 2 -0
> FlopCpp: Generation time: -2.64274e-19
> Clp0000I Optimal - objective value 0
> Clp0032I Optimal objective 0 - 0 iterations time 0.002
> FlopCpp: Optimal obj. value = 0
> FlopCpp: Solver(m, n, nz) = 4 3 0
> constraints
> 0 -1.79769e+308 0 0 -0
> 1 -1.79769e+308 0 0 -0
> 2 -1.79769e+308 0 0 -0
> 3 -1.79769e+308 0 0 -0
> x
> 0 0
> 1 0
> y
> 0 0
> Test dea passed.
> *****output end************************
>
> *****Questions:******************
> I would like to define a set A that includes all members of O and I and
> then derive O and I from A.
> This is what I am thinking in GAMS terms:
> A / Input1, Input2, Output1 /
> O(A) / Input1, Input2 /
> I(A) / Output1 /
> Is this possible?. I thought it would allow me to write the model more
> easily (but may be not)
> *******************
> I tried to display the constraints but I didn't get what I was expecting
> (the actual mathematical expression). Is it possible to see the constraints
> in any other way?
> *******************
> When I compile I get the following two warnings. Should I be concerned? How
> do I fix them?
>
> c:\program files\coin-flopcpp\flopcpp\src\mp_set.hpp(259) : warning C4267:
> 'return' : conversion from 'size_t' to 'int', possible loss of data
> c:\program files\coin-flopcpp\flopcpp\src\mp_set.hpp(258) : while
> compiling class template member function 'int
> flopc::MP_subset<nbr>::size(void) const'
> with
> [
> nbr=2
> ]
> c:\program files\coin-flopcpp\flopcpp\examples\dea.cpp(21) : see
> reference to class template instantiation 'flopc::MP_subset<nbr>' being
> compiled
> with
> [
> nbr=2
> ]
> c:\program files\coin-flopcpp\flopcpp\src\mp_set.hpp(243) : warning C4267:
> 'initializing' : conversion from 'size_t' to 'const int', possible loss of
> data
> c:\program files\coin-flopcpp\flopcpp\src\mp_set.hpp(232) : while
> compiling class template member function 'void
> flopc::MP_subset<nbr>::insert(const std::vector<_Ty> &)'
> with
> [
> nbr=2,
> _Ty=int
> ]
>
> *******************
> The objective function that I really want to use is
> maximize( (sum(O, OutputVal(storeB, O) * y(O))) / (sum(I, InputVal(storeB,
> I) * x(I))));
> But it doesn't compile, I get this error:
>
> c:\program files\coin-flopcpp\flopcpp\examples\dea.cpp(51) : error C2678:
> binary '/' : no operator found which takes a left-hand operand of type
> 'flopc::MP_expression' (or there is no acceptable conversion)
> c:\program files\coin-flopcpp\flopcpp\src\mp_constant.hpp(141):
> could be 'flopc::Constant flopc::operator /(const flopc::Constant &,const
> flopc::Constant &)'
> c:\program
> files\coin-flopcpp\coinutils\src\coinpackedvector.hpp(483): or
> 'CoinPackedVector operator /(const CoinPackedVectorBase &,const
> CoinPackedVectorBase &)'
> c:\program
> files\coin-flopcpp\coinutils\src\coinpackedvector.hpp(583): or
> 'CoinPackedVector operator /(const CoinPackedVectorBase &,double)'
> c:\program
> files\coin-flopcpp\coinutils\src\coinpackedvector.hpp(625): or
> 'CoinPackedVector operator /(double,const CoinPackedVectorBase &)'
> while trying to match the argument list '(flopc::MP_expression,
> flopc::MP_expression)'
>
> I am using MSVCPP, I read of some problems specific to VCPP so I tried the
> same in Linux(I think it is build-essential in Ubuntu) but I get the same
> errors.
> How can I fix this problem with the division operation?
> *******************
> Is my formulation of the objective function OK in how I sum only for storeB
> or do I have to use such_that or any other type of control?
> How do I display the objective function?
> *******************
> What are these numbers in the output?
> constraints
> 0 -1.79769e+308 0 0 -0
> 1 -1.79769e+308 0 0 -0
> 2 -1.79769e+308 0 0 -0
> 3 -1.79769e+308 0 0 -0
> My four constraints are supposed to have only three terms (two on the left
> and one on the right)
> ********************
> What is this?
> Objective
> -1 0 0
> -1 1 -0
> -1 2 -0
> ********************
>
> If it helps at all: the model is a simple DEA formulation (primal).
> I understand some of my questions may be more C++ than flopcpp questions
> but I can't tell.
> I will appreciate help on any of these questions and on any other comments
> you may have on the model formulation.
> Thanks in advance.
> Mario
>
> _______________________________________________
> FlopCpp mailing list
> FlopCpp at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/flopcpp
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://list.coin-or.org/pipermail/flopcpp/attachments/20091201/9be41cac/attachment-0001.html
More information about the FlopCpp
mailing list