[Help-glpk] GLPSOL issue
Hello everybody. I am having an issue using the standalone glpsol solver: on some particular instances, the solver enters a never-ending elaboration (I let it run for 8 hours, when 20 seconds would be already a huge time for an instance of that size), without finding any solution. I am certain that the feasible solution exists because: 1. two other solvers (minos and Cbc) found it and 2. in case of non-existing solution, the solver should tell me that the instance is empty. Anyway, in the following you can find the model and the instance that gives the issue. model: https://dl.dropboxusercontent.com/u/44504233/glpk_issue/glpk_model.mod instance: https://dl.dropboxusercontent.com/u/44504233/glpk_issue/glpk_input.dat *Interesting property*: the issue occurs on all the instances having Cores['ECONOMIC'] very small (=5 in the sample that I provided). I am using glpsol v4.55 on 64 bit ubuntu. The command line I use is simply glpsol --math --model glpk_model.mod --data glpk_input.dat Thank you for your time and attention. -- Mattia Buccarella ___ Help-glpk mailing list Help-glpk@gnu.org https://lists.gnu.org/mailman/listinfo/help-glpk
[Help-glpk] Objective value issue
Hello everyone, my name is Mattia and I am using glpsol in the development of my master thesis in computer science. I am writing to you all for something I could not find anywhere so far. When I use glpsol to solve my model (written in MathProg, with separate data input file), all the variables are configured properly as I expect (I am referring to a dummy input I use to verify the correctness of my model). Nevertheless, the value of the objective function is shifted and I get the following message: Model has been successfully generated *glp_mpl_build_prob: row obje; constant term 84384 ignored* GLPK Integer Optimizer, v4.52 ...etc... Searching the web, I found somebody had a similar issue but it was not related to the objective function and also involved CPLEX (context very different from mine). Does anybody of you know what is happening? When I try to solve the model (same model, same input) with COIN-OR Cbc, through AMPL, it gives me the non-shifted value. Also, it does not tell me anything about constant terms in the function. Also, I checked again and again my objective function: all the terms are functions of some variable, so nothing is really constant in the calculation. Thanks in advance for your help. Goodbye. -- Mattia Buccarella ___ Help-glpk mailing list Help-glpk@gnu.org https://lists.gnu.org/mailman/listinfo/help-glpk
Re: [Help-glpk] Objective value issue
Thanks for your answer. The link you suggested is the one I already saw and discussed in my first email. My question is about the meaning of constant, since all the terms in my objective function are functions of variables (i.e.: none of them is actually constant). Also, I would like to know if there is any way not to let this shift happen. Thank you again for your time. -- Mattia Buccarella On Fri, Jul 11, 2014 at 5:35 PM, Heinrich Schuchardt xypron.g...@gmx.de wrote: Hello Mattia, please, read http://en.wikibooks.org/wiki/GLPK/Troubleshooting#Objective_shift_term_ignored_when_exported Best regards Heinrich Schuchardt http://www.xypron.de Am 11.07.14 um 17:20 schrieb Mattia Buccarella Hello everyone, my name is Mattia and I am using glpsol in the development of my master thesis in computer science. I am writing to you all for something I could not find anywhere so far. When I use glpsol to solve my model (written in MathProg, with separate data input file), all the variables are configured properly as I expect (I am referring to a dummy input I use to verify the correctness of my model). Nevertheless, the value of the objective function is shifted and I get the following message: Model has been successfully generated *glp_mpl_build_prob: row obje; constant term 84384 ignored* GLPK Integer Optimizer, v4.52 ...etc... Searching the web, I found somebody had a similar issue but it was not related to the objective function and also involved CPLEX (context very different from mine). Does anybody of you know what is happening? When I try to solve the model (same model, same input) with COIN-OR Cbc, through AMPL, it gives me the non-shifted value. Also, it does not tell me anything about constant terms in the function. Also, I checked again and again my objective function: all the terms are functions of some variable, so nothing is really constant in the calculation. Thanks in advance for your help. Goodbye. -- Mattia Buccarella ___ Help-glpk mailing list Help-glpk@gnu.org https://lists.gnu.org/mailman/listinfo/help-glpk ___ Help-glpk mailing list Help-glpk@gnu.org https://lists.gnu.org/mailman/listinfo/help-glpk
Re: [Help-glpk] Objective value issue
I just read again the link you provided before and I noticed there is a posting that suggests a workaround to avoid the problem. I will do more trials to see if it works properly. If it will not work, I will provide an example of my model to investigate further. For the moment, I really thank you for your help: I feel like your page will solve my issue. Thanks again. -- Mattia Buccarella On Fri, Jul 11, 2014 at 5:45 PM, Heinrich Schuchardt xypron.g...@gmx.de wrote: Hello Mattia, please, provide an example model. Best regards Heinrich Schuchardt http://www.xypron.de Am 11.07.14 um 17:41 schrieb Mattia Buccarella Thanks for your answer. The link you suggested is the one I already saw and discussed in my first email. My question is about the meaning of constant, since all the terms in my objective function are functions of variables (i.e.: none of them is actually constant). Also, I would like to know if there is any way not to let this shift happen. Thank you again for your time. -- Mattia Buccarella On Fri, Jul 11, 2014 at 5:35 PM, Heinrich Schuchardt xypron.g...@gmx.de wrote: Hello Mattia, please, read http://en.wikibooks.org/wiki/GLPK/Troubleshooting#Objective_shift_term_ignored_when_exported Best regards Heinrich Schuchardt http://www.xypron.de Am 11.07.14 um 17:20 schrieb Mattia Buccarella Hello everyone, my name is Mattia and I am using glpsol in the development of my master thesis in computer science. I am writing to you all for something I could not find anywhere so far. When I use glpsol to solve my model (written in MathProg, with separate data input file), all the variables are configured properly as I expect (I am referring to a dummy input I use to verify the correctness of my model). Nevertheless, the value of the objective function is shifted and I get the following message: Model has been successfully generated *glp_mpl_build_prob: row obje; constant term 84384 ignored* GLPK Integer Optimizer, v4.52 ...etc... Searching the web, I found somebody had a similar issue but it was not related to the objective function and also involved CPLEX (context very different from mine). Does anybody of you know what is happening? When I try to solve the model (same model, same input) with COIN-OR Cbc, through AMPL, it gives me the non-shifted value. Also, it does not tell me anything about constant terms in the function. Also, I checked again and again my objective function: all the terms are functions of some variable, so nothing is really constant in the calculation. Thanks in advance for your help. Goodbye. -- Mattia Buccarella ___ Help-glpk mailing list Help-glpk@gnu.org https://lists.gnu.org/mailman/listinfo/help-glpk ___ Help-glpk mailing list Help-glpk@gnu.org https://lists.gnu.org/mailman/listinfo/help-glpk
