tag:blogger.com,1999:blog-1952964643237411579.post9119632098271055540..comments2023-08-27T16:45:14.601+02:00Comments on Erling's blog: A question about the conic representation of the power function.Erling D. Andersenhttp://www.blogger.com/profile/07306894197500659436noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-1952964643237411579.post-42893076315994039262010-08-25T19:05:04.124+02:002010-08-25T19:05:04.124+02:00I'd love to know what the answer to this is---...I'd love to know what the answer to this is---because I do some sort of rational decomposition in CVX to convert powers to second-order cones. I try to be as efficient as possible but I do not know if I've succeeded. Check my CVX user guide, Appendix D.2, for a rough description.Anonymoushttps://www.blogger.com/profile/07835164854514314231noreply@blogger.comtag:blogger.com,1999:blog-1952964643237411579.post-30005582910865592992010-04-20T15:13:17.614+02:002010-04-20T15:13:17.614+02:00Imre Polik wrote to me:
This has also been descri...Imre Polik wrote to me:<br /><br />This has also been described in the Alizadeh-Goldfarb SOCP paper (Math. Program., Ser. B 95: 3–51 (2003)). Actually, sums of products of variables with rational powers can be modelled this way. Finding the best representation seems to be a hard problem in this general setting, but maybe (1) is easier. Alizadeh-Goldfarb only note that the representation is not uniqueErling D. Andersenhttps://www.blogger.com/profile/07306894197500659436noreply@blogger.com