I read that cbc can solve quadratic problem or I didn't understand.
At the page http://www.coin-or.org/Cbc/cbcuserguide.html, point 7, topic Quadratic MIP, it's written "it is possible to do quadratic MIP with CBC". furthermore, Cbc provide some quadratic problem exemples : quad.mps and quad2.mps with the keyword "QUADOBJ".
So, if I try to solve quad.mps I have some problems.
First case (result file : firstcase.txt) : I open cbc with shell and I import quad.mps
Coin:import quad.mps ... Coin:solve ... Result - Optimal solution found
Objective value: 27797.14684147 Enumerated nodes: 30530 Total iterations: 3410826 Time (CPU seconds): 208.77 Time (Wallclock seconds): 210.57
But if I want to print the result I have a segment fault :
Coin:solution solution.txt Erreur de segmentation (core dumped)
Second case (result file : secondcase.txt) : With a command line in the shell, I have no solution :
cbc quad.mps -solution stdout
Welcome to the CBC MILP Solver Version: 2.8 Build Date: May 7 2014 Revision Number: 2031
command line - cbc quad.mps -solution stdout (default strategy 1) At line 1 NAME At line 2 ROWS At line 299 COLUMNS At line 778 RHS At line 878 BOUNDS At line 985 QUADOBJ Problem no_name has 295 rows, 260 columns and 869 elements Coin0008I no_name read with 0 errors At line 1000 ENDATA Status unknown - objective value 0.00000000 Total time (CPU seconds): 0.00 (Wallclock seconds): 0.00
But if and don't ask the solution it seems to work but impossible to know the details of the solution. I added that Gurobi find different objective value (file : gurobi.txt) :
cbc quad.mps
Result - Optimal solution found
Objective value: 27797.14684147 Enumerated nodes: 30530 Total iterations: 3410826 Time (CPU seconds): 208.56 Time (Wallclock seconds): 211.71
Total time (CPU seconds): 208.56 (Wallclock seconds): 211.71
So, is it a bug ? Cbc can solve quadratic model ? Do you have any ideas ?
Thanks for your help and for your work. You provide very good products.
Attachments: https://github.com/s-c-e/cbc-trac-migration-attachments/blob/master/trac-ticket-151.zip
Hello,
I read that cbc can solve quadratic problem or I didn't understand.
At the page http://www.coin-or.org/Cbc/cbcuserguide.html, point 7, topic Quadratic MIP, it's written "it is possible to do quadratic MIP with CBC". furthermore, Cbc provide some quadratic problem exemples : quad.mps and quad2.mps with the keyword "QUADOBJ".
So, if I try to solve quad.mps I have some problems.
First case (result file : firstcase.txt) : I open cbc with shell and I import quad.mps
Coin:import quad.mps ... Coin:solve ... Result - Optimal solution found
Objective value: 27797.14684147 Enumerated nodes: 30530 Total iterations: 3410826 Time (CPU seconds): 208.77 Time (Wallclock seconds): 210.57
But if I want to print the result I have a segment fault :
Coin:solution solution.txt Erreur de segmentation (core dumped)
Second case (result file : secondcase.txt) : With a command line in the shell, I have no solution :
cbc quad.mps -solution stdout
Welcome to the CBC MILP Solver Version: 2.8 Build Date: May 7 2014 Revision Number: 2031
command line - cbc quad.mps -solution stdout (default strategy 1) At line 1 NAME At line 2 ROWS At line 299 COLUMNS At line 778 RHS At line 878 BOUNDS At line 985 QUADOBJ Problem no_name has 295 rows, 260 columns and 869 elements Coin0008I no_name read with 0 errors At line 1000 ENDATA Status unknown - objective value 0.00000000 Total time (CPU seconds): 0.00 (Wallclock seconds): 0.00
But if and don't ask the solution it seems to work but impossible to know the details of the solution. I added that Gurobi find different objective value (file : gurobi.txt) :
cbc quad.mps
Result - Optimal solution found
Objective value: 27797.14684147 Enumerated nodes: 30530 Total iterations: 3410826 Time (CPU seconds): 208.56 Time (Wallclock seconds): 211.71
Total time (CPU seconds): 208.56 (Wallclock seconds): 211.71
So, is it a bug ? Cbc can solve quadratic model ? Do you have any ideas ?
Thanks for your help and for your work. You provide very good products.