[Cgl] length of gomory cuts

Ashutosh Mahajan asm4 at lehigh.edu
Fri Sep 19 10:30:02 EDT 2008


Thanks for the clarification. i would prefer sticking with just one limit.
that would mean either leaving the code unchanged with an additional
explanation in the comments or make the limit on the final number (your second
option).

ashutosh

On Fri, 19 Sep 2008, John J Forrest wrote:

>    Ashutosh,
> 
>    It is a feature rather than a full blown bug.
> 
>    The Gomory cut is generated from the tableau and it is the length of that
>    cut which is tested. So the first cut generated for acc3 has 499 elements
>    and that is less than 600. However the cut has to be converted back to
>    just have column elements rather than slacks. Normally this does not make
>    very much difference. However with this problem - 3249 rows, 1620 columns
>    and continuous and integer objectives of 0.0, the converted cut has 1554
>    elements.
> 
>    I can modify the code to have another limit - which would not impact
>    current users or I could make the limit on final number, which would
>    change actions for current users. What would people prefer?
> 
>    John Forrest
> 

--
regards
Ashutosh Mahajan
http://coral.ie.lehigh.edu/~asm4



More information about the Cgl mailing list