[Cbc] Suboptimal solution as optimal?

[John Forrest]

Fixed hopefully - stable and trunk.

The original model is not well scaled - one element was 2.1e14 and even 
after preprocessing the problem had to be solved applying scaling 
factors.  The small fast branch and bound is deliberately a bit 
forgiving on this.  When the code thinks it has got a solution it double 
checks.  In this case it did that and threw that solution out - but it 
had left some variables fixed when they should not have been.

John Forrest

On 28/12/17 18:30, John Forrest wrote:
> Ryan,
>
> Bug in Cbc.  My first try gave correct result, but I managed to get 
> error.  For small problems, Cbc will do a simpler faster branch and 
> bound on a portion of the tree.  This is returning infeasible - I will 
> look into it.
>
> If you add -depth -100 (which means only go into this simpler version 
> at depth 100) then all looked fine.
>
> John Forrest
> On 27/12/17 18:37, Rhavar wrote:
>> I have a problem:
>> https://gist.github.com/RHavar/dfb9de631363ecb9e1c326fff5ebd09b 
>> <https://urldefense.proofpoint.com/v2/url?u=https-3A__gist.github.com_RHavar_dfb9de631363ecb9e1c326fff5ebd09b&d=DwMDaQ&c=Ngd-ta5yRYsqeUsEDgxhcqsYYY1Xs5ogLxWPA_2Wlc4&r=js2M0T-3OIMIVDvokcKjokJbk0F8QOCd0mT4FsVFE88&m=S4VZrfTJ9vrNsPyEyqDoTZUqb0eG9rdZ8l7-jr8EA9Y&s=dWErV9QjwaNhpQZZkV1ZmdJ_Ofpqgb80lh0-5ORAiSU&e=>
>>
>> (lp file and solution file attached)
>>
>> In the solution it says: "Optimal - objective value"  but I think I 
>> have found a superior solution by hand:
>>
>> o6 = 1
>> i22 = 1
>>
>> Which gives a better objective (70 instead of 825)
>>
>> --
>>
>> So I guess my question is, does "Optimal - objective value"  mean 
>> that it's guaranteed to be the optimal solution? Or am I doing 
>> something wrong? Or is there a bug in cbc?
>>
>>
>> -Ryan
>>
>>
>>
>>
>> _______________________________________________
>> Cbc mailing list
>> Cbc at list.coin-or.org
>> https://list.coin-or.org/mailman/listinfo/cbc
>
>
>
>
> _______________________________________________
> Cbc mailing list
> Cbc at list.coin-or.org
> https://urldefense.proofpoint.com/v2/url?u=https-3A__list.coin-2Dor.org_mailman_listinfo_cbc&d=DwICAg&c=Ngd-ta5yRYsqeUsEDgxhcqsYYY1Xs5ogLxWPA_2Wlc4&r=js2M0T-3OIMIVDvokcKjokJbk0F8QOCd0mT4FsVFE88&m=S4VZrfTJ9vrNsPyEyqDoTZUqb0eG9rdZ8l7-jr8EA9Y&s=SVcz-BvCf0RgvZrRQQxyV89AGoSkVGhfQhv4tx1JKV4&e=


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/cbc/attachments/20171229/5ca74ecc/attachment.html>