Optim v1.9.1

Diff since v1.9.0

Merged pull requests:

• Use zeros instead of fill in Adam (#1075) (@pkofod)

Optim v1.9.0

Diff since v1.8.0

Merged pull requests:

• Add Adam and AdaMax (#1069) (@pkofod)
• Add docs for adam and adamax (#1072) (@pkofod)

Closed issues:

• Documentation outer configurable options (#914)
• Newlines in two-argment show (#915)
• Feature request: Include the Adam algorithm (#1012)
• Typo on Optim.simplexer() function (#1064)
• Docs Latex expressions displays as raw (#1067)

Optim v1.8.0

Diff since v1.7.8

Merged pull requests:

• Add MOI wrapper (#929) (@blegat)
• Update to new MOI nonlinear interface (#1052) (@blegat)
• Add option for disabling warnings (#1058) (@MilesCranmer)
• Update README example to new MOI nonlinear interface (#1059) (@blegat)
• Fix compat bounds in Project.toml (#1060) (@odow)
• Grammar in documentation of GradientDescent() (#1061) (@lukketotte)

Closed issues:

• Problem with optimization with Optim.GradientDescent (#1018)
• Absolute function tolerance (#1063)

Optim v1.7.8

Diff since v1.7.7

Merged pull requests:

• restore allocation free maxdiff (#1051) (@annuges)

Closed issues:

• Not showing candidate solutions? (#1047)

Optim v1.7.7

Diff since v1.7.6

Closed issues:

• patch 1.7.4 -> 1.7.5 breaks Accelerated/MomentumGradientDescent() (#1038)
• NamedTuple switch broke add_default_options (#1041)
• Verbose option (#1042)
• Optim with BFGS, LBGFS gives wrong result (#1045)

Merged pull requests:

• Change outer tolerances default to nothing (#936) (@ArnoStrouwen)
• Add docs for outer boxconstrained options (#937) (@ArnoStrouwen)
• fix reset_search_direction! failure when training with GPU (#1034) (@wei3li)
• Fix bug in add default options (#1043) (@pkofod)
• Math rendering in the README (#1044) (@DanielVandH)
• exit early when nan in grad or hessian (#1046) (@mohamed82008)

Optim v1.7.6

Diff since v1.7.5

Closed issues:

• For ParticleSwarm, upper default to type that errors if lower is passed (#1033)

Merged pull requests:

• Rework lower/upper in ParticleSwarm for stability. (#1037) (@pkofod)
• CompatHelper: bump compat for StatsBase to 0.34, (keep existing compat) (#1040) (@github-actions[bot])

Optim v1.7.5

Diff since v1.7.4

Closed issues:

• Rosenbrock example: Type instability (#1021)
• How to log additional information after each iteration? (#1024)
• JuliaNLSolvers has only 1 member? (#1025)

Merged pull requests:

• Use namedtuple for default options. (#1026) (@pkofod)
• CompatHelper: bump compat for FillArrays to 1, (keep existing compat) (#1035) (@github-actions[bot])

Optim v1.7.4

Diff since v1.7.3

Closed issues:

• Support for GPU codes? (#1002)
• A way to force at least one iteration (#1004)
• Wrong syntax for Fminbox in example (#1006)
• Saving intermediate results during optimization (#1007)
• Univariate optimisation options (#1008)
• Exact line search (#1009)
• Optimization with (L)BFGS gives DimensionMismatch(“dimensions must match”) (#1011)
• Document what is passed to the callback (#1014)
• Help: Generic non-linear constraint with multiple inputs (#1015)
• LoadError: MethodError: no method matching Optim.IPNewtonState (#1019)
• LoadError: MethodError: no method matching Optim.BarrierLineSearchGrad (#1020)
• Would it be possible to add an argument to the optimization function? (#1022)

Merged pull requests:

• remove unbound type parameter (#1005) (@nsajko)

What's Changed

• Fix incorrect URL for build badges by @cynddl in #998
• improve SAMIN convergence message, fix issue #960 by @mcreel in #988
• Use LinAlg to speed up particle swarm by @alyst in #993

New Contributors

• @cynddl made their first contribution in #998

Full Changelog: v1.7.2...v1.7.3

Optim v1.7.2

Diff since v1.7.1

Closed issues:

• Autodiff using Enzyme (#1001)

Merged pull requests:

• CompatHelper: bump compat for Compat to 4, (keep existing compat) (#986) (@github-actions[bot])