Originally posted by: Y17P_Karthikeswar_Ivaturi

Hello all,

I am graduate research student from UMASS. I am working on virtual network Mapping algorithms for my thesis. I had to work on a code that previously existed. I came across a .mod file where the objective function is located. I need to add a small part to it, but I could not figure out how an objective function in mathematical form is written in as a computer code. I am pasting the code below, if anybody can help me understand the syntax and also help me add my part of the equation it would be great and immensely appreciated.

HUB.mod

---

file

---

set SN;
/*substrate nodes*/

set VN;
/*virtual nodes*/

set AUG:=SN union VN;

param numHubs;
param distance{i in AUG, j in AUG};
/*distance between i and j*/

param W{i in VN, j in VN};
/*demand between i and j*/

param BW{i in AUG, j in AUG};
/*Augmented graph edges BandWidth*/

param cost{i in AUG, j in AUG};

param beta;
/*unit costs of collection (origin to hub)*/
param alpha;
/*unit costs of transfer (hub to hub)*/
param delta;
/*unit costs of distribution (hub to destination*/

param CPU{i in AUG};
param O{i in VN};

param SEL{i in AUG, j in AUG};

var Z{i in AUG, k in AUG} >=0;
/*flow from origin i to hub k*/

var Y{i in VN, k in SN, l in SN} >=0;
/*flow from hub k to hub l for flow originating at node i*/

var X{i in VN, l in SN, j in VN} >=0;

var H{i in SN} binary;

minimize OBJECTIVE: sum{i in VN} ( sum{k in SN}((CPU[i]/(CPU[k]+1E-06))*costi,k*beta*distancei,k*Zi,k) + (sum{k in SN} sum{l in SN}(1/(BWk,l+1E-06)) *(alpha*distancek,l*Yi,k,l)) + sum{l in SN} (CPU[i]/(CPU[l]+1E-06)) * sum{j in VN} ( costj,ldelta distancel,j*Xi,l,j));

subject to AllocConstraint: sum{k in SN} H[k] = numHubs;
subject to AllocConstraint1{i in VN}: sum{k in SN} Zi,k =O[i];

subject to demandConstraint{i in VN, j in VN}: sum{l in SN} Xi,l,j = Wi,j;

subject to FlowConservationConstraint{i in VN, k in SN}: Zi,k+sum{l in SN} Yi,l,k = sum{l in SN} Yi,k,l + sum{j in VN} Xi,k,j ;
subject to CapConstraint{k in SN, l in SN}: sum{i in VN} (Yi,k,l) <=BWk,l;
subject to CapConstraint1{k in SN, l in SN}: sum{i in VN} (Yi,l,k) <=BWl,k;
subject to lessThanConstraint{i in VN, j in VN, l in SN}: Xi,l,j <= Wi,j * H[l];

subject to lessThanConstraint1{i in VN, k in SN}:Zi,k <= O[i]*H[k];

end;

---

I am also attaching the corresponding objective function in mathematical form.

I hope somebody would reply.

Thank you,
Karthik
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

This is an AMPL model file: see http://www.ampl.com. You might also want to post your question at http://groups.google.com/group/ampl (or send it to ampl@googlegroups.com).

Based on a quick look, I would say that your mathematical objective function was changed quite a bit in the process of being implemented in this model. If the mathematical form had been implemented as written, it would be easy to read in AMPL.

Bob Fourer
4er@ampl.com
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: Y17P_Karthikeswar_Ivaturi

Thank you Bob for the prompt reply. I appreciate it greatly. Finally I got some start.
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

Hi

I'm attaching an OPL model file that may help you with the syntax.
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: Y17P_Karthikeswar_Ivaturi

Hi,

Thank you so much. That really helped me understanding the syntax. So, if we have a mod file, a dat file and a .lp file. How do we establish a connection between the 3 files.

Dat file contains the input information for the mod file. So can you please tell me how is the procedure done?

Thank you.
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

The .mod is the formulation of your mathematical model, the .dat will contain the data or instructions to fetch data from a database or Microsoft Spreadsheet or SPSS Stream.
In your OPL project, you create a Run Configuration that will include your .mod file and your .dat file(s), then you execute that run configuration.
Or if you don't work with the IDE, you can use the command line:

oplrun yourModel.mod yourData1.dat ... yourDataN.dat

The .lp is another representation of your model that can be used with interactive CPLEX, not the IDE.
If you want to work with interactive CPLEX, you can easily generate the .lp file from the OPL model.

The Getting Started in the documentation may be useful to you.
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: Y17P_Karthikeswar_Ivaturi

Thank you so much for the Getting Started guide that you provided, that was a lot useful.
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: Y17P_Karthikeswar_Ivaturi

I was solving an objective function and I got this as the output --
CPLEX 12.3.0.0: optimal integer solution within mipgap or absmipgap; objective 16492019410
10 MIP simplex iterations
0 branch-and-bound nodes
absmipgap = 145.402, relmipgap = 8.81652e-09

What does it mean by optimal integer solution within mipgap or absmipgap ? And also what is relmipgap.

Thank you
#DecisionOptimization
#OPLusingCPLEXOptimizer