2016-11-09 Coupler Breakout Meeting notes
Date
Time
- MST: 11am
Call Info
- See "Keep Away" breakout room info
Attendees
Goals
- Gather requirements for future version of ACME coupler.
Discussion items
| Time | Item | Who | Notes |
|---|---|---|---|
What Actually Happened
List of requirements for a future ACME coupler.
- Regridding (aka interpolation aka mapping) needs:
- vector fields interpolated correctly on a sphere
- staggered grids handled natively without interpolation to cell centers
- higher-order methods choosable on a per-field basis
- interpolate gradient AND field for some fields
- possibly iterative to optimize some fields
- Allowing for dynamic grids
- Some algorithmic work necessary such as vector fields in spherical coordinates
- AMR support? not needed soon. Bicycles has support
- support for components that share an interface, 3D for ice-sheet-ocean
- Time interpolation
- option for implicitness
- get away from lagging output
- understand/control time integration error
- Coupling design
- Allow pair wise coupling between components if they want (same grid)
- coupler functionality callable from other models. For atm/dyn physics for example
- coupling of models with 3D-partial overlapping interfaces.
- more runtime configuration and runtime loadbalancing
- handle dynamic domains: wetting and drying of land and ocean from sea-level rise.
- Go away from flexible coupler that allows any atmosphere, any ocean. Specific coupling for ACME components.
- Force models to agree on boundaries, decomposition to lower workload of coupler.