Skip to content

Releases: CliMA/Oceananigans.jl

📒 Oceananigans v0.14.1

25 Oct 15:02
Compare
Choose a tag to compare

v0.14.1 (2019-10-25)

Diff since v0.14.0

Merged pull requests:

🧮 Oceananigans v0.14.0

24 Oct 22:02
Compare
Choose a tag to compare

v0.14.0 (2019-10-24)

Diff since v0.13.0

Main features:

  • Support for abstract operations to define new diagnostics and equations!
  • Support for β planes.
  • Can specify domain for grids.

Closed issues:

  • Output example for NetCDF errors (#499)
  • Behavior of data is inconsistent with behavior of Base.parent (#454)
  • Clean up legacy constructors (#429)
  • Diagnostic for computing horizontal averages of fluxes (#428)
  • Grids should allow us to specify the domain (#413)
  • β-plane and full Coriolis (#28)

Merged pull requests:

🔃 Oceananigans v0.13.0

23 Oct 02:02
Compare
Choose a tag to compare

v0.13.0 (2019-10-23)

Diff since v0.12.0

Main feature:

  • The k index (used for indexing the vertical level in 3D arrays) now increases as you go up so that k=1 corresponds to the bottom of the domain.

Closed issues:

  • Round-off error in RegularCartesianGrid produces ranges of wrong length (#480)
  • Rename Pearson vortex test to Taylor-Green vortex (#468)
  • Vertical index k increases as z coordinate decreases (#90)

Merged pull requests:

🐾 Oceananigans v0.12.0

22 Oct 21:02
Compare
Choose a tag to compare

v0.12.0 (2019-10-22)

Diff since v0.11.1

Main features:

  • Support for specifying 0-2 active tracers and an arbitrary number of passive tracers.
  • Documentation overhaul.
  • Faster and better NetCDF output writer using NCDatasets.jl

Closed issues:

  • LES regression test (#473)
  • Each tracer should have its own diffusivity (#430)
  • Documentation, examples/tutorials, and cleanup for JOSS paper (#332)
  • Numerical methods that need to be documented (#327)
  • Reference or explanation of DCT-FFT translation used for GPU Poisson solver method (#240)
  • Equations in documentation are not dimensionally consistent (#222)
  • Arbitrary tracer fields. (#25)

Merged pull requests:

🗜️ Oceananigans v0.11.1

16 Oct 21:57
Compare
Choose a tag to compare

v0.11.1 (2019-10-15)

Diff since v0.11.0

Main features:

  • Stratified Couette flow verification experiment.
  • Much improved examples.

Closed issues:

  • Error when trying the hot bubble with GPU (#451)
  • First example (hot bubble) fails when defining the model (#449)
  • We should use CI to test that examples and verification tests run without errors (#415)
  • Stratified plane Couette flow test (#310)

Merged pull requests:

⚓ Oceananigans v0.11.0

20 Sep 15:24
Compare
Choose a tag to compare

v0.11.0 (2019-09-19)

Diff since v0.10.1

Main feature:

  • Massive API cleanup and stabilization!

Closed issues:

  • New abstraction for buoyancy and equation of state? (#403)
  • Convenience 'parameters' field in Model for user-defined functions (#394)
  • User-defined forcing function signature (#390)
  • Model constructor not general enough + different model constructors for different use cases (#372)
  • Too many exported names (#364)
  • Use dictionaries rather than arrays to store output writers and diagnostics? (#362)
  • Purpose of Diagnostic abstract type? (#361)
  • Utilities for diagnosing diffusive CFL numbers. (#360)
  • Use Literate.jl for examples? (#358)
  • Generalize JLD2OutputWriter (#357)
  • We should only import what we need from packages (#351)
  • ardata and ardata_view (#313)
  • The future of the Field abstraction (#298)
  • Manual conversions in Grid constructor are not necessary (#285)
  • Different model constructors for different common use cases (#258)
  • Better organization of rotation-related and buoyancy-related constants and parameters (#217)
  • Tidal sloshing forcing function (#212)
  • Idealized nonlinear equation of state (#191)
  • Revise physical constants (#190)
  • Do not install/load CUDA packages if no GPU is detected. (#178)
  • Physics tests with analytic solutions for the statistics (#159)
  • Physics tests for channel or box models with analytic solutions (#158)
  • Physics tests for doubly-periodic domains with analytic solutions (#157)
  • Validation and performance benchmarks. (#136)
  • Ability to turn off/on certain terms in the momentum equations. (#36)

Merged pull requests:

🧭 Oceananigans v0.10.1

10 Sep 08:12
Compare
Choose a tag to compare

v0.10.1 (2019-09-09)

Diff since v0.10.0

Main feature:

  • Bug fixes for diffusivisty halo regions and Value/Gradient boundary conditions to play nice with LES models.

Closed issues:

  • Register Oceananigans v0.10.0 (#384)
  • Better Value and Gradient boundary condition implementation (#371)

Merged pull requests:

📼 Oceananigans v0.10.0

30 Aug 09:33
Compare
Choose a tag to compare

v0.10.0 (2019-08-29)

Diff since v0.9.5

Main feature:

  • New and more correct turbulence closures.

Closed issues:

  • Switch from HAVE_CUDA to new CUDAapi.has_cuda() (#376)
  • Register Oceananigans v0.9.5 (#374)
  • Register Oceananigans v0.9.4 (#373)
  • Clearer Model instantiation error when HAVE_CUDA=false (#295)

Merged pull requests:

🏺 Oceananigans v0.9.5

28 Aug 20:25
Compare
Choose a tag to compare

v0.9.5 (2019-08-28)

Diff since v0.9.4

Main feature:

  • Horizontal average diagnostic and we kind of decided on what diagnostics should look like.

Closed issues:

  • Fast vertical profile diagnostic (#186)
  • Allow forcing to be an array. (#110)
  • Use @propagate_inbounds (#58)

Merged pull requests:

🛂 Oceananigans v0.9.4

28 Aug 12:38
Compare
Choose a tag to compare

v0.9.4 (2019-08-28)

Diff since v0.9.3

Main features:

  • Model checkpointing via the Checkpointer output writer.
  • Verbose and maximum filesize options for the JLD2 output writer.
  • Much cleanup.

Closed issues:

  • Upgrade TimeStepWizard to support model spinup? (#356)
  • Register Oceananigans v0.9.3 (#345)
  • Split JLD2 output files by size (#344)
  • Possibly incorrect halo region filling and missing boundary condition (#328)
  • Barebones checkpointing (#324)
  • Halo regions filling functions are not general (#307)
  • What is the purpose of the 'dim' property of RegularCartesianGrid? (#287)
  • HDF5 and plotting utilities are heavy dependencies (#284)
  • More tuples! (#262)
  • Constant Smagorinsky halo size (#260)
  • Types of kernel arguments do not seem to be inferred (#251)
  • Eliminate dichotomy between GPU and CPU tests (#242)
  • Deep convection example and golden master test should switch to BCs. (#220)
  • Dynamically install packages like Plots.jl and ArgParse.jl for examples. (#195)
  • Better user API for interacting with fields (#174)
  • isbits Fields and FieldSets (#163)
  • Change AB-2 parameter to 0.5 + χ to offer first order forward time stepping (#160)
  • Elegant unpacking (#155)
  • set!(u::Field{G}, f::Function) should use xC, yC, zC (or xF, yF, zF) depending on Field. (#23)
  • Using Reqiures.jl for optional GPU usage (#14)

Merged pull requests: