[Coin-discuss] RE: Running the BranchAndCut example

[Jesper Hansen]

Hi Laszlo

Thank you. Now Mkc and MaxCut is working fine. However, the Branch&Cut still
gives an assert. Did you find out if it is a roundoff problem or...? Does
there exist a good Osi tutorial or do I have to look at the code? The Lehigh
site is a bit outdated and at least for me difficult to follow without
additional comments.

'Jesper

-----Original Message-----
From: Laszlo Ladanyi [mailto:ladanyi at us.ibm.com]
Sent: Monday, April 07, 2003 3:47 PM
To: Jesper Hansen
Cc: coin-discuss at www-124.southbury.usf.ibm.com
Subject: RE: [Coin-discuss] RE: Running the BranchAndCut example


Uh-oh... I haven't looked at the examples in a while. I'll check all today.

I can already tell what the problem with the Branch&Cut example:
  "Cutting" is done by designating some of the rows as being in the matrix
to
start with and after solving each LP the rest of the rows are tested for
feasibility. Whichever is violated is added to the formulation. This is the
"cutting". Now when a feasible solution is found it is sent back to the tree
manager and there, just to make sure, it is checked whether it violates any
of
the constraints and that assertion fails. Roundoff errors can make that
happen
(though not likely). I'll check it out in any case.

Can you tell me what operating system and compiler (including version) have
you used?

Thanks,
--Laci


> >
> > --------------------------
> >
> > BCP_parameters::read_from_stream   Scanning parameter stream.
> > BCP_parameters::read_from_stream   Scanning parameter stream.
> > BCP_parameters::read_from_stream   Scanning parameter stream.
> > BCP_parameters::read_from_stream   Scanning parameter stream.
> > BCP_parameters::read_from_stream   Scanning parameter stream.
> > BCP_parameters::read_from_stream   Scanning parameter stream.
> > Coin0001I At line 15 NAME          P0033
> > Coin0001I At line 16 ROWS
> > Coin0001I At line 34 COLUMNS
> > Coin0001I At line 109 RHS
> > Coin0001I At line 118 BOUNDS
> > Coin0001I At line 152 ENDATA
> > Coin0002I Problem P0033 has 16 rows, 33 columns and 98 elements
> > Coin0008I P0033 read with 0 errors
> > Clp0005I 0  Objective 0 Primal infeas 27.2175 (10)
> > Clp0005I 19  Objective 2520.57
> > Clp0000I Optimal - objective value 2520.57
> >
> > --------------------
> >
> > LP: *** Starting iteration 2 ***
> > Clp0005I 0  Objective 3053.44 Primal infeas 2.16911 (9)
> > Clp0005I 6  Objective 3089
> > Clp0001I Primal infeasible - objective value 3089
> > LP:   Matrix size: 33 vars x 33 cuts
> > LP:   Solution value: 3089.0000 / 36 , 6
> >  LP: Default test_feasibility() executed.
> >  LP: Default test_full() executed.
> > LP:   Terminating and fathoming due to proven high cost.
> > LP:   Deleting 5 rows from the matrix.
> > TM: Pruning NODE 8 LEVEL 4 instead of sending it.
> > TM: Pruning NODE 11 LEVEL 6 instead of sending it.
> > TM: Pruning NODE 13 LEVEL 7 instead of sending it.
> > TM: Pruning NODE 10 LEVEL 5 instead of sending it.
> > TM: Pruning NODE 16 LEVEL 8 instead of sending it.
> > TM: Pruning NODE 15 LEVEL 8 instead of sending it.
> > TM: Running time: 0.340
> > TM: search tree size: 31   max depth: 8
> > LP statistics:
> >    time in cut generation  :        0.060 sec
> >    time in var generation  :       -0.000 sec
> >    time in heuristics      :       -0.000 sec
> >    time in solving LPs     :        0.140 sec
> >    time in strong branching:        0.110 sec
> >
> > TM: The best solution found has value 3089.000000
> > bcps: /home/jesper/COIN/Examples/BranchAndCut/TM/BB_tm.cpp:234: void
> > BB_tm::display_feasible_solution (const BCP_solution *): Assertion
`lhs[i]
> > >= desc.rlb_core[i] - 1e-6 && lhs[i] <= desc.rub_core[i] + 1e-6' failed.
> > Aborted
> >
> >