tag:blogger.com,1999:blog-593563533834706486.post3481502982688638207..comments2024-03-28T10:35:10.453-04:00Comments on Yet Another Math Programming Consultant: Continuous max sum rectangle: MIQP vs GAErwin Kalvelagenhttp://www.blogger.com/profile/09496091402502236997noreply@blogger.comBlogger4125tag:blogger.com,1999:blog-593563533834706486.post-91603552432766987562021-01-24T07:11:03.058-05:002021-01-24T07:11:03.058-05:00That is a great observation. I get a similar perfo...That is a great observation. I get a similar performance with Cplex. Erwin Kalvelagenhttps://www.blogger.com/profile/09496091402502236997noreply@blogger.comtag:blogger.com,1999:blog-593563533834706486.post-60620407974794106182021-01-23T12:25:39.661-05:002021-01-23T12:25:39.661-05:00Thanks for sharing the data. My colleague Imre Po...Thanks for sharing the data. My colleague Imre Polik suggested treating the points as elements of a sparse matrix and using your formulation from http://yetanothermathprogrammingconsultant.blogspot.com/2021/01/submatrix-with-largest-sum.html. This yielded a 20x speedup in solution time for us. For n points, the model has 4n variables and 2n+2 constraints. After linearizing the n products (not n^2), the model has 5n variables and 5n+2 constraints.Rob Pratthttps://www.blogger.com/profile/16525877394541155854noreply@blogger.comtag:blogger.com,1999:blog-593563533834706486.post-29286829458670420572021-01-23T03:14:45.942-05:002021-01-23T03:14:45.942-05:00https://amsterdamoptimization.com/data/p1000.csvhttps://amsterdamoptimization.com/data/p1000.csvErwin Kalvelagenhttps://www.blogger.com/profile/09496091402502236997noreply@blogger.comtag:blogger.com,1999:blog-593563533834706486.post-16774658762153412752021-01-22T16:30:56.589-05:002021-01-22T16:30:56.589-05:00Can you please share the larger data?Can you please share the larger data?Rob Pratthttps://www.blogger.com/profile/16525877394541155854noreply@blogger.com