Open jonathan-taylor opened 2 years ago
Hi, Jonathan! This will need changing the convex solver at every node. We use CD, which will not work anymore. I think this can be done but will need some work especially since we need dual bounds for BnB termination.
Hi Rahul,
What is the convex sub problem you solve? Would proximal gradient / FISTA be fast enough?
Get Outlook for Androidhttps://aka.ms/ghei36
From: Rahul Mazumder @.> Sent: Tuesday, October 12, 2021 9:20:51 AM To: alisaab/l0bnb @.> Cc: Jonathan Taylor @.>; Author @.> Subject: Re: [alisaab/l0bnb] bound form (#1)
Hi, Jonathan! This will need changing the convex solver at every node. We use CD, which will not work anymore. I think this can be done but will need some work especially since we need dual bounds for BnB termination.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/alisaab/l0bnb/issues/1#issuecomment-941167112, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AACTM247R76L256B6THFJMLUGRN6HANCNFSM5F2562CA. Triage notifications on the go with GitHub Mobile for iOShttps://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Androidhttps://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.
Replying to you by email.
Wondering how hard it would be to get a version with bound on l_0 instead of Lagrange form