v0.95.8

Oceananigans v0.95.8

Diff since v0.95.7

Merged pull requests:

• More generalization of tuples (#4049) (@wsmoses)
• Fix bug in validation of MutableVerticalDiscretization() (#4050) (@tomchor)
• Disambiguate interpolation of ::Number on grids with Flat dimensions (#4051) (@simone-silvestri)
• Remove all hardcoded Float64 from the Advection module (#4053) (@simone-silvestri)
• Some changes to supported_float_types (#4056) (@simone-silvestri)
• Bugfix for advection on distributed topologies (#4058) (@simone-silvestri)
• Add a fallback set_timestepper! (#4059) (@simone-silvestri)
• (0.95.8) Bump to 0.95.8 (#4060) (@glwagner)
• Use 1.10.8 in tests (#4062) (@glwagner)

Closed issues:

• Unable to pickup from checkpoint when using SplitRungeKutta3Timestepper (#4057)

v0.95.7

Oceananigans v0.95.7

Diff since v0.95.6

Merged pull requests:

• (0.95.7) Yet another ZStar implementation (#3956) (@simone-silvestri)
• Allows properties of StaticVerticalCoordinate to have different types (#4029) (@wsmoses)
• Generalize ntuple type constraints to be tuple{vararg} (#4035) (@wsmoses)
• Optimize masking xy planes (#4040) (@simone-silvestri)
• Add compat entries for Makie 0.22 and MakieCore 0.9 (#4045) (@navidcy)

v0.95.6

Oceananigans v0.95.6

Diff since v0.95.5

Merged pull requests:

• Correct tupled_halo_filling for distributed fields (#4017) (@simone-silvestri)
• Support interpolating onto Windowed fields (#4021) (@simone-silvestri)
• Add info for Julia version used for testing (#4022) (@navidcy)
• Minor enhancement in OceananigansMakieExt (#4023) (@navidcy)
• Makie extension: tweaks the xticks for domains that wraps around the globe (#4025) (@navidcy)
• Enzyme: mark prettytime as non-differentiable (#4026) (@wsmoses)
• Add default Cb for CATKE (#4031) (@glwagner)
• Inference problem on the CPU (#4032) (@simone-silvestri)
• (0.95.6) Pin GPUArrays version + fix GPU testing pipeline (#4037) (@simone-silvestri)

Closed issues:

• Default CATKE configuration (#4015)
• Use xticks to multiples of 60 or 30 for a domain that wraps around the globe in Makie extension (#4024)
• "Free surface" is mentioned in the vertical coordinate summary (#4030)

v0.95.5

Oceananigans v0.95.5

Diff since v0.95.4

v0.95.4

Oceananigans v0.95.4

Diff since v0.95.3

Merged pull requests:

• Allow FieldTimeSeries to load data split across multiple files (#3976) (@glwagner)
• Avoid flux computations when not needed in the IsopycnalSkewSymmetricDiffusivity (#3992) (@simone-silvestri)
• Dont import Grids.coordinates (#4000) (@glwagner)
• Try to add plot labels automagically within Makie extension (#4001) (@glwagner)
• Generalized Checkpointing plus interface for set!(::Simulation, pickup) (#4003) (@glwagner)
• Add masking methods for ConstantFields (#4006) (@simone-silvestri)

Closed issues:

• Mid-Level Enzyme + Oceananigans Integration Test (#3346)
• Automatically adding labels to Makie's plots in Oceananigans (#3998)
• WARNING: could not import Grids.coordinates into ImmersedBoundaries (#3999)

v0.95.3

Oceananigans v0.95.3

Diff since v0.95.2

Merged pull requests:

• (0.95.3) Reset model clock and skip time_step!ing if next actuation time is tiny (#3606) (@tomchor)
• Add parameter to epsilon equation so buoyancy flux term may always be positive (#3886) (@glwagner)
• (0.95.3) Restores open boundary condition functionality (#3937) (@jagoosw)
• Actually test instantiation on GPU and also skip LagrangianAveragedDynamicSmagorinsky test (#3989) (@glwagner)
• Add support for loading data generated prior to 0.95.0 with FieldTimeSeries (#3990) (@glwagner)
• Update the docstring for HydrostaticFreeSurfaceModel (#3991) (@simone-silvestri)
• Remove grid requirements from a Model in a Simulation (#3993) (@simone-silvestri)

Closed issues:

• Pressure has extremely high gradients in random chunks of simulation using NonhydrostaticModel with ImmersedBoundaryGrid and BuoyancyTracer (#3593)
• Unexpected FlatExtrapolationBoundaryCondition behaviour in validation script cylinder.jl (#3975)
• Testing the instantiation of LagrangianAveragedDynamicSmagorinsky on GPU takes 1 hour (#3988)

v0.95.2

Oceananigans v0.95.2

Diff since v0.95.1

Merged pull requests:

• (0.95.2) Bugfix in the Polar BC (#3985) (@simone-silvestri)

v0.95.1

Oceananigans v0.95.1

Diff since v0.95.0

Merged pull requests:

• Fix horizontally Flat WENO advection (#3979) (@simone-silvestri)
• Fix exported names from BuoyancyFormulations module (#3984) (@glwagner)
• Bump to 0.95.1 (#3987) (@glwagner)

v0.95.0

Oceananigans v0.95.0

Diff since v0.94.3

Merged pull requests:

• Correct AveragedTimeInterval to use actuations (#3721) (@liuchihl)
• Adds momentum equation test for Enzyme extension (#3822) (@jlk9)
• Add test for autodifferentiating hydrostatic turbulence simulation (#3867) (@glwagner)
• Fix bug in advect_particle (#3923) (@simone-silvestri)
• Reinstate DynamicSmagorinsky tests (#3926) (@glwagner)
• RK3 for the HydrostaticFreeSurfaceModel version 2 (#3930) (@simone-silvestri)
• Fix Integral reductions on immersed grids (#3949) (@ali-ramadhan)
• Fix positivity preserving WENO (#3952) (@simone-silvestri)
• Refactor of spacings and correction of spacings-connected bugs (#3955) (@simone-silvestri)
• Allow Tuple of closures with CATKE (#3960) (@simone-silvestri)
• Polar boundary conditions (#3965) (@simone-silvestri)
• Some corrections in the timestepper constructors (#3969) (@simone-silvestri)
• advective form of the GM skew diffusivity (#3971) (@simone-silvestri)
• (0.95.0) Buoyancy -> BuoyancyForce and BuoyancyModels -> BuoyancyFormulations (#3973) (@glwagner)
• (0.95.0) Refactor grids to allow generalized vertical coordinate (#3974) (@simone-silvestri)

Closed issues:

• Add Multi GPU testing (#2526)
• Should we make Conjugate Gradient + FFTBased-preconditioner the default for HydrostaticFreeSurfaceModel? (#2635)
• Is Architectures.next_stream() redundant? (#2748)
• Problem using immersed boundary with PartialCellBottom and Flat horizontal dimensions (#2798)
• with_halo is broken for ImmersedBoundaryGrid (#2988)
• Boundary condition ordering (#3342)
• xspacings is broken for ImmersedBoundaryGrid (#3701)
• Implementing AB2 for variable time-step size (#3738)
• SplitExplicitFreeSurface code improvements (#3873)
• Why do we only build distributed test_architectures if MPI.Initialized? (#3879)
• Some Lagrangian particles like to travel Southwest on an immersed lat-lon grid + hydrostatic model (#3917)
• Error when constructing Integral of a field on an immersed rectilinear grid: Azᶜᶜᵃ not defined (#3948)
• minimum_xspacing does not account for the immersed boundary (#3954)
• Duplicate ψspacing and Δψ (#3957)
• Why are the prognostic fields deep-copied in the constructor for HydrostaticFreeSurfaceModel? (#3967)

v0.94.3

Oceananigans v0.94.3

Diff since v0.94.2

Merged pull requests:

• Reduce parameter space for velocity tendency kernels (#3924) (@simone-silvestri)
• Leftover from split explicit refactor (#3931) (@simone-silvestri)
• CompatHelper: bump compat for StructArrays to 0.7, (keep existing compat) (#3934) (@github-actions[bot])
• (0.94.3) Vector rotation for immersed boundary grids (#3939) (@simone-silvestri)
• Bugfix in compute bottom height (#3940) (@simone-silvestri)
• Better message for regularize_immersed_boundary_condition (#3942) (@navidcy)
• Fix a typo in show(io, Simulation) (#3944) (@glwagner)