[Cbc] Infeasible in CBC, optimal for commercial solvers

[John Forrest]

Kevin,

Both you and Haroldo are right.  The preprocessing changed some values 
slightly.  The correct solution was found but when double checking it a 
tighter tolerance is used for a solve but the standard tolerance should 
have been used on final check for feasibility.  The largest 
infeasibility found was then epsilon over allowed.

I have modified to put back standard tolerance a bit earlier.

John Forrest

On 20/11/17 11:52, Kevin van Blokland wrote:
> Dear CBC community,
>
> We are faced with a problem formulation that is infeasible for CBC but 
> optimal for commercial solvers. My best guess is that the problem is 
> related to integer/floating point precision issues.
>
> Is someone able to look at the provided mps files and give us any 
> direction on a solution? The mps file can be downloaded from the 
> following link: 
> https://cqmshare.blob.core.windows.net/uploads/model3.mps?st=2017-11-20T11%3A25%3A00Z&se=2018-11-21T11%3A25%3A00Z&sp=r&sv=2016-05-31&sr=b&sig=ecTeCY%2BaqgbGr%2By7CHxw6p6jCWtIR9k2U02b3KT7CTg%3D 
> <https://urldefense.proofpoint.com/v2/url?u=https-3A__cqmshare.blob.core.windows.net_uploads_model3.mps-3Fst-3D2017-2D11-2D20T11-253A25-253A00Z-26se-3D2018-2D11-2D21T11-253A25-253A00Z-26sp-3Dr-26sv-3D2016-2D05-2D31-26sr-3Db-26sig-3DecTeCY-252BaqgbGr-252By7CHxw6p6jCWtIR9k2U02b3KT7CTg-253D&d=DwMBAg&c=Ngd-ta5yRYsqeUsEDgxhcqsYYY1Xs5ogLxWPA_2Wlc4&r=js2M0T-3OIMIVDvokcKjokJbk0F8QOCd0mT4FsVFE88&m=TKEW7I8nzLYXl9nCBltb6mFBfMpcLYrWwH8zi90kRmg&s=i644nvRBcwNp5sgd0rIjQyYJ5O8mdidHvIpMSOsHOoY&e=> 
>
>
> Help is very much appreciated!
>
> Kind regards,
>
> Kevin van Blokland
>
>
> _______________________________________________
> 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=TKEW7I8nzLYXl9nCBltb6mFBfMpcLYrWwH8zi90kRmg&s=-ARtpjY-lYAVQc9qjq_NWC9iWXROW_7NXdOb3fANyQY&e=


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/cbc/attachments/20171120/58c435f6/attachment.html>