Search for a command to run...
Boscia v0.1.0 This release has been identified as a backport. Automated changelogs for backports tend to be wildly incorrect. Therefore, the list of issues and pull requests is hidden. <!-- **Merged pull requests:** - active set size and discarded set size added to output callback (#9) (@hannahtro) - Throw error in case of over splitting (#10) (@dhendryc) - Clean code (#11) (@dhendryc) - Check example (#14) (@dhendryc) - Cleanupversions (#15) (@matbesancon) - interface toplevel (#16) (@matbesancon) - adjusting printing (#17) (@pokutta) - pre and post solving reporting + placeholder for solution status (#25) (@pokutta) - dual gap (#27) (@matbesancon) - print iter as parameter, default 100 (#29) (@matbesancon) - [WIP] Improving_reporting_further (#30) (@pokutta) - added a tricky example and relative gap computation function (#31) (@pokutta) - homogenization (#32) (@pokutta) - removed node id from reporting (#33) (@pokutta) - removing debug code (#35) (@pokutta) - minor fix and odd problem (#36) (@pokutta) - Use bonobo optimize (#38) (@hannahtro) - [WIP] Examples Basics (#39) (@pokutta) - limited heursitic check to when LMO was called (#42) (@matbesancon) - add params to dict (#45) (@hannahtro) - rename header (#46) (@hannahtro) - add list for lmo_calls_per_layer (#47) (@hannahtro) - Final calculation and clean up (#48) (@dhendryc) - homogenized header naming with FW package (#49) (@pokutta) - Reset LMO and SCIP solve (#52) (@dhendryc) - Mps example (#53) (@matbesancon) - Clean examples (#54) (@dhendryc) - slight renaming of strategies (#55) (@pokutta) - relative gap without % (#59) (@matbesancon) - strong branching example and fix (#60) (@matbesancon) - return vertices set size per layer (#61) (@hannahtro) - Some changes on the final reporting and other minor changes (#62) (@dhendryc) - add fw_epsilon treshold to node (#63) (@hannahtro) - set minimum epsilon (#64) (@matbesancon) - Fix relative gap termination issue (#65) (@dhendryc) - renamed package (#67) (@matbesancon) - Incorrect primal dual (#70) (@dhendryc) - Lasso Example (#72) (@dhendryc) - Callback result (#73) (@hannahtro) - Custom optimize function and fix in the lb update (#74) (@dhendryc) - do not stop FW early in first iteration (#75) (@hannahtro) - Fix min number lower (#76) (@hannahtro) - As fix (#77) (@dhendryc) - Fix dual gap issue (#78) (@dhendryc) - Reporting (#80) (@dhendryc) - Added time limit as criterion. (#81) (@dhendryc) - DRAFT: Portfolio fix (#82) (@dhendryc) - add updated_incumbent to root (#84) (@hannahtro) - use FrankWolfe solution (#85) (@hannahtro) - added SCIP (#86) (@matbesancon) - HiGHS in test deps (#90) (@matbesancon) - Deps (#91) (@matbesancon) - Use FW callback (#92) (@dhendryc) - Portfolio fix (#95) (@dhendryc) **Closed issues:** - Fix broken output (#12) - Include Bonobo deeper into BranchWolfe (#13) - # nodes currently open in log (#18) - add final iteration (#19) - BB is not stopped when relative gap / absolute gap is reached (#20) - Solution sometimes numerically close to integer only -> we should probably polish them before reporting (#21) - When returning solution add test whether solution is feasible, i.e., it satisfies the integrality requirements (#22) - fix issue with FW package 0.2.5 (#23) - produce more detailed solution stastitics after call (#24) - Remode Node ID from printing (#26) - Need for clarification: 0 FW iterations (#34) - Do correctness tests via enumeration for _small_ examples (#37) - Merge of bonobo.optimize et al broke behavior again (#40) - Interface: return a statistics object / dict from the run to test behavior (#41) - Polish solution returned (#43) - Cumulative LMO in printing (#44) - add relative_gap as criterion (#50) - Cleanup phase: reset LMO optimizer bounds to the global bounds after solve (#51) - Add example that uses strong branching (#56) - add support for time limit as stopping criterion (#57) - Extend initial reporting of algorithm querying the underlying lmo information (#58) - problem with empty tree and incorrect primal dual values (#66) - Add safety asserts to detect regressions (#68) - abs gap / rel gap = infinity and stopping (#71) - Add "*" is as very first character whenever a new best primal solution is found (#79) - Add the source of primal solutions when adding them (#83) - Issue with portfolio example (#93) -->