[Coin-discuss] BCP procedures

Heesu Hwang hxh9528 at omega.uta.edu
Tue Jul 4 11:12:12 EDT 2006


Hi, all.

This is related to BCP, however, somehow bcp-discuss mailing list for JULY
doesn't take my posting, so I am posting this in coin-discuss.
Please, pardon me if this annoys you.

I changed the order of procedures as followings by modifying
BCP_lp_main_loop.cpp
in short...
First:     generate_vars_in_lp()
Second: generate_cuts_in_lp()

     The output of my code shows...
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
DCG_tm::initialize_core()
DCG_tm::create_root()
DCG_lp::initialize_solver_interface()
############################################################
TM: Starting phase 0
############################################################
DCG_tm::init_new_phase
DCG_lp::vars_to_cols
DCG_lp::cuts_to_rows
DCG_lp::initialize_new_search_tree_node
DCG_lp::modify_lp_parameters
##### matrix-0.0.1
LP:   Matrix size: 3486 vars x 3435 cuts
LP:   Solution value: -1085053.0482 / 2 , 603 
DCG_lp::test_feasibility
DCG_lp::compute_lower_bound
DCG_lp::generate_vars_in_lp
DCG_lp::vars_to_cols
DCG_lp::generate_cuts_in_lp
DCG_lp::cuts_to_rows
DCG_lp_branch::select_branching_candidates
DCG_lp::modify_lp_parameters

##### matrix-0.0.2
LP:   Matrix size: 4851 vars x 4785 cuts
LP:   Solution value: -1085053.0482 / 2 , 109 
DCG_lp::test_feasibility
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

   My code runs about 25 seconds and finds first feasible solution.
   After that the BCP procedures change, and my code dies.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
##### matrix-1.136.271
LP:   Matrix size: 11096 vars x 10973 cuts
LP:   Solution value: -486256.4960 / 2 , 29 
DCG_lp::test_feasibility
TM: Default unpack_feasible_solution() executed.
TM: Solution found at 24.760 sec.
TM: Solution value: -486256.496000 (best solution value so far: infinity)
DCG_lp::compute_lower_bound
DCG_lp::vars_to_cols        <-----  skips DCG_lp::generate_vars_in_lp
DCG_lp::cuts_to_rows       <----- skips DCG_lp::generate_cuts_in_lp

./DCG: line 1:   539 Segmentation fault      
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Could you give me some idea, why the procedures change after the solution?

Thanks,

Heesu Hwang.




More information about the Coin-discuss mailing list