😽 v0.90.9

Oceananigans v0.90.9

Diff since v0.90.8

Merged pull requests:

• Use Julia v1.10 for CI (#3403) (@navidcy)
• Scaling improvement for the distributed HydrostaticFreeSurfaceModel (#3404) (@simone-silvestri)
• (0.90.9) Fix a sign error in CATKE's implicit dissipative buoyancy flux (#3484) (@glwagner)

Closed issues:

• Are we actually @unrolling when we think we are? (#3374)
• maximum(abs, v) doesn't work on GPU in Julia 1.10.0 with grid size larger than (10, 10, 10) (#3427)

v0.90.8

Oceananigans v0.90.8

Diff since v0.90.7

Merged pull requests:

• (0.90.8) Rules for Time interpolation / extrapolation for FieldTimeSeries and InterpolatedField (#3450) (@glwagner)

Closed issues:

• FieldTimeSeries does not adapt correctly on GPU (#3472)

v0.90.7

Oceananigans v0.90.7

Diff since v0.90.6

Merged pull requests:

• Some type stabilization to get a new gradient from Enzyme (#3360) (@glwagner)
• Fixes error when reducing Fields with a condition on ImmersedBoundaryGrids (#3440) (@tomchor)
• Better docstring rendering (#3448) (@navidcy)
• Bump up SeawaterPolynomials to v0.3.4 (#3451) (@glwagner)
• Remove Enzyme from core dependency (#3452) (@navidcy)
• Update Docs/Model setup pages + fix bug with callsite=TendencyCallsite() in Callbacks (#3454) (@navidcy)
• Enhance Checkpointer + TendencyCallsite callback example (#3461) (@navidcy)
• More illustrative example for checkpointing (#3469) (@navidcy)
• (0.90.7) Remove argument splatting in hydrostatic free surface tendency kernel entry functions (#3477) (@glwagner)

Closed issues:

• Fusing kernels for calculating diagnostics to improve performance (#1483)
• The definition of gravity_unit_vector apparently doesn't match it's name (#2356)
• Building WENO advection with immersed grid fails due to grid size (#3137)
• Doc page for Callbacks outdated and does not appear under Docs/Model Setup (#3275)
• Ambiguous method error when calculating a conditional Average() with an ImmersedBoundaryGrid (#3439)
• 2D turbulence example missing fontsize keyword in Label call (#3442)
• WENOVectorInvariant method cleanup? (#3449)
• Callback with callsite=TendencyCallsite() doesn't work (#3456)
• Can we rewrite the pressure equation and calculate each term in it? (#3459)
• restore_from_checkpoint is deprecated? (#3462)
• Checkpointer required properties includes :architecture despite it can be inferred from grid (#3463)
• @fluidnumerics-joe the CI doesn't work... something's wrong with the Manifest you pushed. (#3470)

🎿 v0.90.6

Oceananigans v0.90.6

Diff since v0.90.5

Merged pull requests:

• Refactor Fields.interpolate to make sense for Nothing locations and Flat directions - take 2 (#3395) (@simone-silvestri)
• Enhance docstring for Integral (#3446) (@tomchor)
• (0.90.6) Bump (#3447) (@glwagner)

Closed issues:

• interpolate doesn't work with Nothing locations (#3356)

📟 v0.90.5

Oceananigans v0.90.5

Diff since v0.90.4

Merged pull requests:

• Change StokesDrift example in docs to a wavepacket moving with group velocity (#3425) (@navidcy)
• Improve references/citations in docs + enhances 3D Stokes drift example (#3430) (@navidcy)
• Add few papers in the "Papers and preprints using Oceananigans" (#3437) (@navidcy)
• Add missing adapt_structure for stokes drift (#3441) (@jagoosw)
• Use CairoMakie v0.11 for example/docs (#3443) (@navidcy)

Closed issues:

• Three-dimensional StokesDrift docstring example may be misleading (#3422)
• Non-standard function throws an error when used in a KernelFunctionOperation on the GPU (#3438)
• Question for 3D Stokes example (#3444)

⛲️ v0.90.4

Oceananigans v0.90.4

Added three-dimensional implementation of Stokes drift.

Diff since v0.90.3

Merged pull requests:

• Fixes bug with pre-defining diffusivity fields (#3367) (@jagoosw)
• (0.90.4) Implement three-dimensional StokesDrift 🌊 (#3384) (@glwagner)
• Fix docs for @apply_regionally (#3387) (@navidcy)
• Improve docstring example for seawater density (#3398) (@glwagner)
• Add convenience functions for conjuring TimeStepWizards and adding Callbacks (#3405) (@glwagner)
• Few fixes in few examples (#3414) (@navidcy)
• Adds some missing @inbounds (#3417) (@navidcy)
• Add @inbounds to ab2_step_Gu,v (#3421) (@navidcy)

Closed issues:

• compute_regionally must be deprecated (#3386)
• Split-explicit inbounds checks? (#3419)

👅 v0.90.3

Oceananigans v0.90.3

Diff since v0.90.2

Merged pull requests:

• (0.90.3) Fix bug in adapting ScalarDiffusivity and ScalarBiharmonicDiffusivity (#3413) (@glwagner)

🧑🏾‍🍳 v0.90.2

Oceananigans v0.90.2

Diff since v0.90.1

Merged pull requests:

• Default NaN-Checker for Particle Tracking models (#3376) (@simone-silvestri)
• Add OceanBioME.jl paper (#3380) (@navidcy)
• Updates Relaxation to work with grids with Flat dimensions (#3383) (@jagoosw)
• Tiny conversion of code to English (#3385) (@glwagner)
• Changes property "longname" to "long_name" for NetCDFOutputWriter (#3391) (@tomchor)
• Correction of three-dimensional Stokes drift implementation (#3392) (@BrodiePearson)
• Add link to arXiv for scalings in README (#3393) (@navidcy)
• Added adapt method for scalar and scalar biharmonic diffusvity + test (#3401) (@jagoosw)
• Fix CUDA compat entry (#3402) (@jagoosw)

Closed issues:

• Documentation for latest versions does not seem to be working (#3377)
• Prescribing realistic bottom bathymetry (#3389)
• Nonhydrostatic pressure correction broken? (#3397)

🕴️ v0.90.1

Oceananigans v0.90.1

Diff since v0.90.0

Merged pull requests:

• Non uniform partitioning for Distributed architectures (#3339) (@simone-silvestri)
• Fix advection timescale calculation (#3362) (@zhihua-zheng)
• Update regression test data (#3373) (@simone-silvestri)
• Removes pole warning from LatitudeLongitudeGrid (#3375) (@navidcy)
• (0.90.1) Bump to 0.90.1 to fix docs issue (#3378) (@glwagner)

Closed issues:

• Docs don't build for tagged releases (#3191)
• Regression test(s) for SplitExplicitFreeSurface (#3335)
• Should the time step wizard take into account of background velocity in calculation of CFL number? (#3357)

😑 0.90.0

Oceananigans v0.90.0

This version introduces a major change (#3355) to the API for initial conditions, background fields, and boundary conditions based on functions on grids with Flat directions. When using functions, we previously would supply all three coordinates (eg f(x, y, z) regardless of whether the function was being used on a grid with a Flat direction. But now, the coordinates associated with Flat directions are omitted. For example, if y is Flat, then initial conditions are set with functions of f(x, z).

One more example. Before this release, one might write

grid = RectilinearGrid(size=64, z=(0, 1), topology=(Flat, Flat, Bounded))
c = CenterField(grid)
random_numbers(x, y, z) = rand()
set!(c, random_numbers)

After this release, this code has to change because the x, y directions are Flat. Thus after this release we write

grid = RectilinearGrid(size=64, z=(0, 1), topology=(Flat, Flat, Bounded))
c = CenterField(grid)
random_numbers(z) = rand()
set!(c, random_numbers)

As a rule, coordinates within functions should never be provided for Flat directions or for Nothing locations.

Diff since v0.89.3

Merged pull requests:

• Add Y-partition and XY-partition tests (#3338) (@simone-silvestri)
• (0.90.0) Define WENOVectorInvariant + some cleanup (#3341) (@glwagner)
• Move Enzyme extension tests to dedicated test group (#3347) (@glwagner)
• Update Docs/Model/Output Writers to replace FieldSlicer -> indices (#3350) (@xkykai)
• (0.90.0) Omit entries in Flat direction from node (#3355) (@glwagner)
• Explicit OffsetNDRange vs NDRange (#3359) (@simone-silvestri)
• Better defaults for WENOVectorInvariant (#3365) (@glwagner)
• CompatHelper: add new compat entry for Statistics at version 1, (keep existing compat) (#3366) (@github-actions[bot])
• Add compat entries for Stdlibs (#3371) (@navidcy)
• Add compat entry for Random (#3372) (@navidcy)

Closed issues:

• node should omit nodes from Flat dimensions (#3285)
• Update docs to change FieldSlicer to indices (#3348)
• Broadcasting slices of fields doesn't work as expected (#3361)