CliMA/Oceananigans.jl: v0.46.0

Diff since v0.45.2

Breaking changes:

• The Cell location has been renamed to Center which is a clearer name for cell centers on the staggered grid.

Major features:

• Support for online Lagrangian particle tracking with custom particles and tracked field properties (see model setup docs, examples incoming!).
• Support for higher-order advection schemes, forcing functions, and different time steppers for shallow water models.
• New KernelComputedField for fields that need to be computed using a KernelAbstractions.jl CPU/GPU kernel.
• Abstract operations are now conditionally computed as needed to avoid wasted computations.
• Numerous bug fixes and documentation improvements.

Closed issues:

• Change Cell to 'Center' and Face to 'Interface' to specify Field locations? (#414)
• Lagrangian particle trajectories (#511)
• Animations in Docs don't show up on Safari Mac OS X (#944)
• Make announcement post on Discourse (#1111)
• Include units and longname for time in netcdf output (#1208)
• Abstraction for using "custom" kernels to compute fields (plus an example)? (#1246)
• The Great Debate: NetCDF vs JLD2 (#1261)
• Tests fail because shallow water model with h=0 blows up when time stepped (#1262)
• Example in Field docstring is mangled (#1268)
• Aborted (core dumped) on tutorial (#1281)
• ShallowWaterModel needs more options (#1284)
• ERROR: importing Flux into Main conflicts with an existing identifier (#1285)
• "NaN error in u" when trying to simulate open-ocean convection problem (#1289)
• Trying to calculate Richardson number using GPU kernel is failing at the bottom boundary (#1290)
• Do Oceananigans and magnetohydrodynamics mix? (#1304)
• Adding background fields to perturbations fails when writing to NetCDF (#1308)
• Including installation of required packages in examples creates clutter in Docs (#1315)

Merged pull requests:

• Lagrangian particle tracking (#1091) (@ali-ramadhan)
• Set h=1 in shallow water time stepping tests (#1264) (@ali-ramadhan)
• Adding advection schemes into Shallow Water (#1266) (@francispoulin)
• Fixes mangled docstring for Field (#1269) (@glwagner)
• Adds support for advection=nothing (#1270) (@glwagner)
• Timesteppers and forcing functions for shallow water models (#1291) (@ali-ramadhan)
• Adds kernel computed field (#1293) (@tomchor)
• CompatHelper: bump compat for "KernelAbstractions" to "0.5" (#1295) (@github-actions[bot])
• Update to CUDA v2.4.0 (#1296) (@ali-ramadhan)
• Adds explanatory remark on KH power-method (#1298) (@navidcy)
• Creates long_name and units attributes for the time dimension in netcdf outputs (#1299) (@tomchor)
• Implements compute_at! pattern for conditional computation (#1301) (@glwagner)
• Always take positive time steps (#1303) (@ali-ramadhan)
• Animations in Docs go back in using mp4 instead of gif (#1306) (@navidcy)
• Changed every instance of Cell in .jl files to Center using sed (#1314) (@tomchor)
• Avoid executing the lines to install dependencies within examples (#1316) (@navidcy)