[Coin-discuss] Mistaken infeasibility when using reduce-and-split cut generator

[Esben Mose Hansen]

Hi,

I came across this problem: For both /devel and /trunk branches, running with 
reduce-and-split active, even at root, makes coin believe the problem is 
infeasible.  A short select  from the output is included below, while the 
problem itself is available at 

http://ange.dk/public/mistaken_infeasibility_447.mps

it's about 1000 variables.

The output:

$ cbc mistaken_infeasibility_447.mps -reduce off -solve
[...]
Cbc0013I At root node, 37 cuts changed objective from 341649 to 1.21167e+06 in 
17 passes
[...]
Result - Finished objective 1.21174e+06 after 2 nodes and 684 iterations - 
took 1.34008 seconds

$ cbc mistaken_infeasibility_447.mps -reduce root -solve
[...]
Cbc0013I At root node, 298 cuts changed objective from 341649 to 593399 in 1 
passes
[...]
Result - Finished objective 1e+50 after 0 nodes and 243 iterations - took 
0.760047 seconds

What is going on here? 

-- 
regards, Esben

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.