NVIDIA/cuopt v26.02.00 on GitHub

What's Changed

Fix issue stemming because Rocky 8.10 is EOL and replaced by 8.11 by @rgsl888prabhu in #427
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #429
Integrate CliqueMerging presolver and tune presolver settings by @rg20 in #415
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #433
Add optional timestamps enabled by env CUOPT_EXTRA_TIMESTAMPS by @tmckayus in #392
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #435
Remove left over kernels by @akifcorduk in #430
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #436
GF2 presolve infeasibility fix + CI occasional failure fix by @aliceb-nv in #437
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #439
Parallel Branch-and-Bound by @nguidotti in #412
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #441
Barrier method for LPs with cuDSS sparse Cholesky / LDLT on the GPU by @chris-maes in #272
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #446
Short run fine tuning by @aliceb-nv in #444
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #447
Fixed delayed termination after timeout by @nguidotti in #451
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #452
Explicitely create new stream for Barrier concurrent by @hlinsen in #453
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #455
Fix compile and log by @akifcorduk in #450
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #456
Don't eliminate free variables in dual simplex by @chris-maes in #457
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #460
Enable green context in Barrier by @hlinsen in #448
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #461
More numerically robust CPUFJ feasibility checking by @aliceb-nv in #454
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #464
Barrier post-solve, find_dense_columns fix, folding fix by @chris-maes in #462
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #465
Further CPUFJ assert and numerical fixes by @aliceb-nv in #466
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #467
Disable TBB because of a bug in CliqueMerging presolver by @rg20 in #463
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #469
Add setting to change the dual initial point used by barrier by @chris-maes in #468
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #471
Fix an UMA on ns1952667 due to ignoring rank in factorize_basis when no singletons present by @chris-maes in #470
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #472
Remove raft-dask dependency by @rgsl888prabhu in #475
Don't access free variables in the original problem on the folded problem by @chris-maes in #477
Add docs and example tests to barrier by @rgsl888prabhu in #449
Barrier log fixes by @chris-maes in #478
Enable TBB and use fixed commit version by @akifcorduk in #481
[BUG] Fixed starting variable bounds for diving by @nguidotti in #474
Gracefully handle OOM in Barrier by @hlinsen in #483
Fix issue with barrier incorrectly returning numerical in concurrent by @chris-maes in #484
Release notes for v25.10 by @cwilkinson76 in #432
Print scaling information about the user model by @chris-maes in #488
Fix return on size mismatch in Barrier by @hlinsen in #489
Fix handling zero sized array in case of presolve by @rgsl888prabhu in #491
Switch to using papilo develop branch by @rg20 in #493
Branch 25.12 merge branch 25.10 by @rgsl888prabhu in #490
Branch 25.12 merge branch 25.10 by @rgsl888prabhu in #505
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #476
Fix TSP when order locations are set by @hlinsen in #503
Fix nightly container testing suite by @rgsl888prabhu in #495
Add default accuracy for each method in LP by @rgsl888prabhu in #514
Forward-merge branch-25.10 into branch-25.12 by @rapids-bot[bot] in #515
Reduce package sizes by @rgsl888prabhu in #507
Add code-rabbit config file to the repo by @rgsl888prabhu in #516
Refactor build scripts to use RAPIDS_BRANCH and remove RAPIDS_VERSION by @bdice in #517
Disable flaky barrier tests by @rgsl888prabhu in #520
Add release notes to main repository for quick view by @rgsl888prabhu in #512
Use main shared-workflows branch by @bdice in #523
Improve Diversity Loop by @akifcorduk in #518
Update coderabbit config to enable autoreview by @rgsl888prabhu in #522
Remove papilo from getting added to packaging by @rgsl888prabhu in #521
Fix cudf breaking change for build_column which is used to create cudf series from buffer by @rgsl888prabhu in #535
Carry over implied integers from papilo presolve reductions by @aliceb-nv in #534
Handle concurrent mode oom and avoid unnecessary deep copy of problem_t by @hlinsen in #541
Fix papilo assertion mip by @hlinsen in #542
Disable netlib testing by @rgsl888prabhu in #550
Use one cusparse handle per thread to avoid race condition on cuspars… by @hlinsen in #544
Fix bug in logger when log_to_console is set to false by @rg20 in #538
Persistent basis_update_mpf_t in the Dual Simplex Phase 2 by @nguidotti in #383
Skip version check for cuOpt in Julia code by @rgsl888prabhu in #539
Update CCCL to align with rapids-cmake by @bdice in #487
Fix compilation for the solve_MIP executable for benchmarking by @nguidotti in #557
Reduce concurrent overhead from right_looking_lu in dual simplex by @hlinsen in #555
Fix nightly notebook testing by @rgsl888prabhu in #549
Use current system architecture in conda environment creation command by @bdice in #567
Do not rebuild from scratch when git hash changes by @rg20 in #563
Use ruff-check, ruff-format instead of black, flake8, isort by @KyleFromNVIDIA in #571
Use latest pre-commit-hooks, clean up copyright notices by @KyleFromNVIDIA in #573
Fix issue with build_column in cudf which is being deprecated by @rgsl888prabhu in #578
RINS heuristic by @aliceb-nv in #527
Fix minor printf bugs. by @ahehn-nv in #574
refactored update-version.sh to handle new branching strategy by @rockhowse in #583
Update RMM includes from <rmm/mr/device/*> to <rmm/mr/*> by @bdice in #588
Move doc examples to individual files and add CI test to test those by @rgsl888prabhu in #565
Fix cusparse leak by @Kh4ster in #590
Update to 26.02 by @AyodeAwe in #596
Forward-merge release/25.12 into main by @rapids-bot[bot] in #597
Revert "Forward-merge release/25.12 into main" by @AyodeAwe in #598
Make API tested library static since it is needed for only testing by @rgsl888prabhu in #594
Forward-merge release/25.12 into main by @rapids-bot[bot] in #600
Enable nightly build and testing for 25.12 by @rgsl888prabhu in #604
Forward-merge release/25.12 into main by @rapids-bot[bot] in #605
Test lean CI PR run option by @rgsl888prabhu in #540
Add review guide-lines for coderabbit and remove unnecessary details in review. by @rgsl888prabhu in #572
Main merge branch 25.12 by @rgsl888prabhu in #607
Forward-merge release/25.12 into main by @rapids-bot[bot] in #606
Disable TSP tests which are failing due to dataset download issue. by @rgsl888prabhu in #611
Forward-merge release/25.12 into main by @rapids-bot[bot] in #614
Fix nightly build to build for both main and release and one shouldn't cancel other by @rgsl888prabhu in #608
Fix memory leaks in barrier by @hlinsen in #584
Forward-merge release/25.12 into main by @rapids-bot[bot] in #618
Improve Numerical Stability in Dual Simplex by @nguidotti in #591
Forward-merge release/25.12 into main by @rapids-bot[bot] in #619
Remove cuvs dependency from cuOpt by @rgsl888prabhu in #615
Add missing parameters to documentation by @rgsl888prabhu in #613
Use strict priority in CI conda tests by @bdice in #617
Propagate the bounds from the parent to the child nodes by @nguidotti in #473
Reuse of the basis factorization from the parent node by @nguidotti in #492
Main merge branch 25.12 by @rgsl888prabhu in #624
Fix explored nodes counter by @nguidotti in #570
Restart helper function fix by @Kh4ster in #629
Fix Crash in the Incumbent Test by @nguidotti in #632
Disable tsp dataset download by @hlinsen in #633
Main merge release/25.12 by @rgsl888prabhu in #639
Forward-merge release/25.12 into main by @rapids-bot[bot] in #620
Handle inherited module doc string issues by @rgsl888prabhu in #637
Forward-merge release/25.12 into main by @rapids-bot[bot] in #640
Turn off building with assertions by default by @tmckayus in #635
Forward-merge release/25.12 into main by @rapids-bot[bot] in #643
Use strict priority in CI conda tests by @bdice in #636
FIx nightly to build only single architecture image instead of manifest list by @rgsl888prabhu in #638
Forward-merge release/25.12 into main by @rapids-bot[bot] in #645
Add Multi-GPU concurrent LP solve by @hlinsen in #601
Forward-merge release/25.12 into main by @rapids-bot[bot] in #646
Set the linker to use RUNPATH instead of RPATH to allow for LD_LIBRARY_PATH overrides by @aliceb-nv in #631
Forward-merge release/25.12 into main by @rapids-bot[bot] in #648
Add more preemption points to stop the solver after optimal is found by @aliceb-nv in #647
Forward-merge release/25.12 into main by @rapids-bot[bot] in #649
add a note on contents of ~/.local/lib for Python builds by @tmckayus in #651
Forward-merge release/25.12 into main by @rapids-bot[bot] in #655
Add nim operator documentation on for using cuOpt by @rgsl888prabhu in #642
Forward-merge release/25.12 into main by @rapids-bot[bot] in #659
try adding cuda_version to dependencies.yaml by @tmckayus in #661
Forward-merge release/25.12 into main by @rapids-bot[bot] in #662
Fix pdlp_solver_settings method bug in pdlp by @Kh4ster in #657
Forward-merge release/25.12 into main by @rapids-bot[bot] in #663
Fix OOM when running very large MIP instances by @nguidotti in #658
Forward-merge release/25.12 into main by @rapids-bot[bot] in #664
Fix cvxpy failure in nightly testing by @nguidotti in #656
Forward-merge release/25.12 into main by @rapids-bot[bot] in #666
Add support for quadratic programming to the barrier method by @chris-maes in #500
Forward-merge release/25.12 into main by @rapids-bot[bot] in #669
Add concurrent lp solve and crossover at root node by @hlinsen in #602
Forward-merge release/25.12 into main by @rapids-bot[bot] in #672
cuopt service correct definition for status value in lp result spec by @tmckayus in #671
Forward-merge release/25.12 into main by @rapids-bot[bot] in #675
Add details of new branching strategy and release cycle details to README and CONTRIBUTING.md by @rgsl888prabhu in #641
Rounding and crash fixes by @aliceb-nv in #670
Forward-merge release/25.12 into main by @rapids-bot[bot] in #679
Improve iterative refinement by @rg20 in #677
Forward-merge release/25.12 into main by @rapids-bot[bot] in #682
add Quadratic expression class and Objective by @rgsl888prabhu in #673
Forward-merge release/25.12 into main by @rapids-bot[bot] in #683
Fix infinite average pseudocost + Handle numerical issues in strong branching by @nguidotti in #678
Forward-merge release/25.12 into main by @rapids-bot[bot] in #684
Fix root LP solution copy in diversity_manager_t by @aliceb-nv in #685
Fix warp divergence by @hlinsen in #681
Update release notes for 25.12 by @rgsl888prabhu in #665
Update the QP documentation by @chris-maes in #688
Main merge release/25.12 1 by @rgsl888prabhu in #689
Forward-merge release/25.12 into main by @rapids-bot[bot] in #687
Forward merging CI crash fixes (#691) by @rgsl888prabhu in #716
Fixes mismatch cuda-numba version between cuopt and cudf by @nguidotti in #717
Fix non implemented case in LinearExpression class by @hlinsen in #724
Fix missing headers for thrust::pair, add trailing return types where needed by @bdice in #732
Populate dual solution when dual post solve is enabled by @hlinsen in #726
Fix device code to provide trailing return type. by @rgsl888prabhu in #738
Fix synchronization issue originated from non-blocking stream by @rgsl888prabhu in #740
Use SPDX license identifiers in pyproject.toml, bump build dependency floors by @jameslamb in #742
Use static GTest by @bdice in #729
Remove alpha specs from non-RAPIDS dependencies by @bdice in #660
Modernize dependency pinnings by @bdice in #723
Adding best practices and details by @rgsl888prabhu in #692
remove dask dependencies and configuration, reduce duplication in dependencies.yaml by @jameslamb in #744
Create separate test run scripts by @bdice in #722
Typo fix in pip install instructions by @mlubin in #743
Use static linkage for CUDA runtime by @bdice in #680
Implement batch solve for routing models by @hlinsen in #728
build and test against CUDA 13.1.0 by @jameslamb in #747
Add clang host build and ThreadSanitizer support, race condition fixes by @aliceb-nv in #603
Update release branch to 26.02 for upcoming release nightly build by @rgsl888prabhu in #751
doc: Adding FQDN to versions1.json by @aschilling-nv in #756
Add message when optimal solution is found during presolve by @adityasinghz in #623
Adjusted column spacing in B&B logs by @nguidotti in #714
Disable consistent updates of max dynamic shared memory by @hlinsen in #750
Move datasets to s3 to reduce dependency on third-party download sites by @rgsl888prabhu in #748
add branch and commit time info to image by @Iroy30 in #737
Update doc and cusparse check by @Iroy30 in #735
Use augmented system when there are no constraints by @rg20 in #765
Make README consistent with other docs by @rg20 in #768
Use main shared-workflows branch by @jameslamb in #777
Explicitely use cudaStream_t for cub::DeviceTransform and use non blocking stream by @hlinsen in #771
remove deprecated params by @Iroy30 in #770
Move augmented sytem computations in barrier to GPU by @rg20 in #746
Fix container build and test by @rgsl888prabhu in #780
Add Probing Implications and Improve Variable Order by @akifcorduk in #580
Diving Improvements by @nguidotti in #697
Replace cudf Column and Buffers APIs with pylibcudf and public cuDF APIs by @rgsl888prabhu in #781
wheel builds: react to changes in pip's handling of build constraints by @mmccarty in #785
Add support for c++20 by @rgsl888prabhu in #486
Fix typo in QP problem instantiation by @mtanneau in #758
Propose fix a typo by @jeis4wpi in #761
Fix find dense columns out of bounds by @hlinsen in #793
Add skills and other helper structure for agentic work by @rgsl888prabhu in #778
Add sonarqube for cuopt - static analysis by @rgsl888prabhu in #762
Several bug fixes by @nguidotti in #786
Add support for --split-compile and --jobserver in build.sh by @aliceb-nv in #626
Add matrix interface to problem modeling by @Iroy30 in #775
Batch PDLP by @Kh4ster in #805
C api additions by @aliceb-nv in #721
Add C api callbacks for getting and setting solutions by @akifcorduk in #779
Add regression testing by @Iroy30 in #556
Add cuts to the MIP solver by @chris-maes in #814
Adjust basis repair for super basic variables. Repeat root solve if cuts make it infeasible by @chris-maes in #831
Scheduler-Worker Model + Reliability Branching by @nguidotti in #766
FJ numerical fix + fix clang build by @aliceb-nv in #835
Integrate PSLP presolver in cuOpt by @rg20 in #816
update solver_settings by @Iroy30 in #840
Fix an issue with the sign of dual variables on maximization problems by @chris-maes in #841
B&B CPU determinism by @aliceb-nv in #798
Adding doc updates for release notes and other items by @rgsl888prabhu in #808

New Contributors

@cwilkinson76 made their first contribution in #432
@rockhowse made their first contribution in #583
@mlubin made their first contribution in #743
@aschilling-nv made their first contribution in #756
@adityasinghz made their first contribution in #623
@mmccarty made their first contribution in #785
@mtanneau made their first contribution in #758
@jeis4wpi made their first contribution in #761

Full Changelog: v25.12.00a...v26.02.00