tag:blogger.com,1999:blog-593563533834706486.post1774782970643117083..comments2024-03-28T10:35:10.453-04:00Comments on Yet Another Math Programming Consultant: MPS filesErwin Kalvelagenhttp://www.blogger.com/profile/09496091402502236997noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-593563533834706486.post-69748860925084060462011-10-19T15:27:17.123-04:002011-10-19T15:27:17.123-04:00No, I have not looked into that. MPS files have no...No, I have not looked into that. MPS files have no standard facilities for maximization and for a constant term in the obj; these are well-known issues.<br /><br />I think the situation I described is somewhat different: there we see incorrect MPS files being generated.Erwin Kalvelagenhttps://www.blogger.com/profile/09496091402502236997noreply@blogger.comtag:blogger.com,1999:blog-593563533834706486.post-65641864665318509002011-10-19T08:13:37.609-04:002011-10-19T08:13:37.609-04:00Have you checked translation of a constant in the ...Have you checked translation of a constant in the objective too?<br />lp_solve and I think glpsol<br /> do not change the sign for something like<br />min: 10 + X ;<br /> to mps.<br />But Coin clp ( presumably IBM standard )<br /> expects the constant as -10 in the mps file to give a matching objective value of the solution.<br />lp_solve dealt with this in 5.5.0.15 (2009) with new options.<br /><br />- Changed the second parameter to read_mps, read_MPS, read_freemps and<br />read_freeMPS from verbose to options.<br />These routines now supports via this options parameter the native IBM<br />interpretation of<br />integer variables and allows to negate the objective constant.<br />This is also supported by the lp_solve command line program via the -mps_ibm<br />and -mps_negobjconst options.<br />This is also supported by the IDE in the options tab.William Pattonnoreply@blogger.com