Repository navigation
v0.116.0
Oceananigans v0.116.0
Breaking changes
LambertConformalConicGridis renamed toConformalConicGrid, with no deprecation (#6168).reset!(simulation)no longer resets the stop criteria:stop_iteration,stop_time, andwall_time_limitare kept (#5521).NonhydrostaticModelnow throws for a halo that is too small for its advection schemes, likeHydrostaticFreeSurfaceModelandShallowWaterModel, instead of rebuilding the grid. Bounds-preserving WENO now requires a horizontal halo ofbuffer + 1(4 for WENO5), and per-tracer advection schemes are now included in halo validation (#6315).interpolate!throws anArgumentErrorwhen interpolating to a grid that isFlatin a dimension where the source grid is not (#5474).- Biased (WENO/upwind) interpolations use their reduced-order buffer schemes only near boundaries. Results change near boundaries and immersed boundaries (#6212).
- The sign of
BulkDragon right boundaries is fixed. Results change for simulations that use it there (#5973).
New features
IMEXFluxBoundaryConditionworks with any flux, with a Patankar fallback for the implicit split, andBulkDraghas an IMEX option (#5973).conformal_conic_rotation_anglegives the convergence angle of a conformal conic projection at a longitude (#6321).
Bug fixes
set!rotates geographic vdirections on immersedOrthogonalSphericalShellGrids, such as an immersedTripolarGrid(#6316).- The horizontal pair of the tor` is rotated the right way(#6320).
- Fixed a distributed deadlocds between architectures(#6332).
Performance
- Cheaper halo fills: unused boundary-condition arguments are skipped, and tuples of fields are filled with one ke4).
masked_multiplyin the batched tridiagonal solver is simpler (#6261).
Merged pull requests:
- Validate Flat dimensions in interpolate! (#5474) (@ewquon)
- Don't reset stop_time or stop_iteration in reset! (#5521) (@glwagner)
- Evaluate the reduced-order buffer schemes of biased interpolations only near boundaries (#6212) (@glwagner)
- Simplify masked_multiply (#6261) (@glwagner)
- Use Julia v1.13.2 in CI (#6309) (@navidcy)
- Rotate the horizontal pair the right way in the 3-component
extrinsic_vector(#6320) (@ewquon) - Add conformal_conic_rotation_angle for the convergence angle at a longitude (#6321) (@glwagner)
Closed issues: