Update Mosek to 11.0.24 #23151
Update Mosek to 11.0.24 #23151
Conversation
jwnimmer-tri
added
the
release notes: fix
|
Assigning @jwnimmer-tri for first look or delegation. |
There was a problem hiding this comment.
Reviewable status: 1 unresolved discussion, LGTM missing from assignee jwnimmer-tri(platform), needs at least two assigned reviewers
a discussion (no related file):
Working
@hongkai-dai and I will need to check whether TRI can use Mosek 11 yet.
|
@drake-jenkins-bot linux-jammy-clang-bazel-experimental-everything-release please @drake-jenkins-bot linux-jammy-unprovisioned-gcc-cmake-experimental-packaging please @drake-jenkins-bot mac-arm-sequoia-clang-bazel-experimental-everything-release please |
There was a problem hiding this comment.
Reviewed 1 of 1 files at r1, all commit messages.
Reviewable status: 3 unresolved discussions, needs at least two assigned reviewers
a discussion (no related file):
Previously, jwnimmer-tri (Jeremy Nimmer) wrote…
Working
@hongkai-dai and I will need to check whether TRI can use Mosek 11 yet.
TRI's internal license server only offers v10 right now. We'll need to request an upgrade to v11.
@hongkai-dai should be hold this PR aside until TRI is upgraded, or should be temporarily drop support for MOSEK within TRI until we can upgrade?
a discussion (no related file):
See #20634 for a recent MOSEK upgrade. You'll note that we need to perl pie a bunch of MOSEK urls to point to a new versioned docs base url. Please make that change here, and also improve the upgrade instructions checklist (in repository.bzl) to have a item that mentions this.
tools/workspace/mosek/repository.bzl line 10 at r1 (raw file):
# When the version is updated: # - tools/dynamic_analysis/tsan.supp may also need updating # - LICENSE.third_party may also need updating to match
@mwoehlke-kitware please compare LICENSE.third_party in this directory vs the website URL below, and update our copy if necessary (or just report back here if they are the same).
There was a problem hiding this comment.
Reviewable status: 4 unresolved discussions, needs at least two assigned reviewers
a discussion (no related file):
In mac-arm-sequoia-clang-bazel-experimental-everything-release, the following failures seem real:
[ FAILED ] test/LorentzByLorentzSeparabilityTest.AddMatrixIsLorentzByLorentzSeparableConstraintExpressionChangeSize/0, where GetParam() = (3, 4)
[ FAILED ] test/LorentzByLorentzSeparabilityTest.AddMatrixIsLorentzByLorentzSeparableConstraintExpressionChangeSize/1, where GetParam() = (4, 3)
[ FAILED ] test/LorentzByLorentzSeparabilityTest.AddMatrixIsLorentzByLorentzSeparableConstraintExpressionChangeSize/2, where GetParam() = (5, 5)
[ FAILED ] test/LorentzByLorentzSeparabilityTest.AddMatrixIsLorentzByLorentzSeparableConstraintExpressionChangeSize/7, where GetParam() = (1, 4)
mwoehlke-kitware
force-pushed
the
upgrade-mosek
branch
from
f386ab0 to
f8b2b22
Compare
In order to allow wheels to use Mosek provided from PyPI without version support regressions, we need to use a newer version of Mosek (in particular, 11.x). Therefore, update Mosek to the latest version currently available. Also remove support for x86 macOS, which is not supported by Mosek 11.
mwoehlke-kitware
force-pushed
the
upgrade-mosek
branch
from
f8b2b22 to
e750fc4
Compare
There was a problem hiding this comment.
Reviewable status: 4 unresolved discussions, needs at least two assigned reviewers
a discussion (no related file):
Previously, jwnimmer-tri (Jeremy Nimmer) wrote…
See #20634 for a recent MOSEK upgrade. You'll note that we need to perl pie a bunch of MOSEK urls to point to a new versioned docs base url. Please make that change here, and also improve the upgrade instructions checklist (in
repository.bzl) to have a item that mentions this.
(Perl? Why is Perl necessary?)
Done, partly.
Several links are to 'latest'. Should they be?
mathematical_program.h:2513: https://docs.mosek.com/latest/capi/prob-def-quadratic.html#a-recommendationmosek_solver.cc:261: https://docs.mosek.com/latest/rmosek/tutorial-mio-shared.html#specifying-an-initial-solutionmosek_solver.cc:287: https://docs.mosek.com/latest/capi/debugging-tutorials.html#debugging-tutorialsmosek_solver_internal.cc:955: https://docs.mosek.com/latest/capi/tutorial-sdo-shared.html#example-sdo-lmi-linear-matrix-inequalities-and-the-vectorized-semidefinite-domainmosek_solver_internal.h:180: https://docs.mosek.com/latest/capi/tutorial-acc-optimizer.htmlmosek_solver_internal.h:182: https://docs.mosek.com/latest/capi/tutorial-sdo-shared.htmlmosek_solver_internal.h:608: https://docs.mosek.com/latest/capi/constants.html#mosek.soltype
tools/workspace/mosek/repository.bzl line 10 at r1 (raw file):
Previously, jwnimmer-tri (Jeremy Nimmer) wrote…
@mwoehlke-kitware please compare
LICENSE.third_partyin this directory vs the website URL below, and update our copy if necessary (or just report back here if they are the same).
Done.
However... I notice that we aren't making any effort to preserve formatting from that page.
There was a problem hiding this comment.
Reviewed 11 of 11 files at r2, 1 of 1 files at r3, all commit messages.
Reviewable status: 3 unresolved discussions, LGTM missing from assignee hongkai-dai
a discussion (no related file):
Previously, mwoehlke-kitware (Matthew Woehlke) wrote…
(Perl? Why is Perl necessary?)
Done, partly.
Several links are to 'latest'. Should they be?
mathematical_program.h:2513: https://docs.mosek.com/latest/capi/prob-def-quadratic.html#a-recommendationmosek_solver.cc:261: https://docs.mosek.com/latest/rmosek/tutorial-mio-shared.html#specifying-an-initial-solutionmosek_solver.cc:287: https://docs.mosek.com/latest/capi/debugging-tutorials.html#debugging-tutorialsmosek_solver_internal.cc:955: https://docs.mosek.com/latest/capi/tutorial-sdo-shared.html#example-sdo-lmi-linear-matrix-inequalities-and-the-vectorized-semidefinite-domainmosek_solver_internal.h:180: https://docs.mosek.com/latest/capi/tutorial-acc-optimizer.htmlmosek_solver_internal.h:182: https://docs.mosek.com/latest/capi/tutorial-sdo-shared.htmlmosek_solver_internal.h:608: https://docs.mosek.com/latest/capi/constants.html#mosek.soltype
I'll need to defer to +@hongkai-dai for whether those should remain "latest" or change to "11.0".
There was a problem hiding this comment.
Reviewable status: 3 unresolved discussions, LGTM missing from assignee hongkai-dai
a discussion (no related file):
Previously, jwnimmer-tri (Jeremy Nimmer) wrote…
In
mac-arm-sequoia-clang-bazel-experimental-everything-release, the following failures seem real:[ FAILED ] test/LorentzByLorentzSeparabilityTest.AddMatrixIsLorentzByLorentzSeparableConstraintExpressionChangeSize/0, where GetParam() = (3, 4) [ FAILED ] test/LorentzByLorentzSeparabilityTest.AddMatrixIsLorentzByLorentzSeparableConstraintExpressionChangeSize/1, where GetParam() = (4, 3) [ FAILED ] test/LorentzByLorentzSeparabilityTest.AddMatrixIsLorentzByLorentzSeparableConstraintExpressionChangeSize/2, where GetParam() = (5, 5) [ FAILED ] test/LorentzByLorentzSeparabilityTest.AddMatrixIsLorentzByLorentzSeparableConstraintExpressionChangeSize/7, where GetParam() = (1, 4)
In mac-arm-sonoma-clang-bazel-experimental-everything-release, a subset of those same failures occur:
[ FAILED ] test/LorentzByLorentzSeparabilityTest.AddMatrixIsLorentzByLorentzSeparableConstraintExpressionChangeSize/0, where GetParam() = (3, 4)
[ FAILED ] test/LorentzByLorentzSeparabilityTest.AddMatrixIsLorentzByLorentzSeparableConstraintExpressionChangeSize/1, where GetParam() = (4, 3)
[ FAILED ] test/LorentzByLorentzSeparabilityTest.AddMatrixIsLorentzByLorentzSeparableConstraintExpressionChangeSize/7, where GetParam() = (1, 4)
|
(FYI all of my original jenkins bot suite has finally finished running now, and the only failures were the single test program that I previously noted in a review discussion.) |
In order to allow wheels to use Mosek provided from PyPI without version support regressions, we need to use a newer version of Mosek (in particular, 11.x). Therefore, update Mosek to the latest version currently available.
Also remove support for x86 macOS, which is not supported by Mosek 11.
Toward #20296.
This change is