-
Notifications
You must be signed in to change notification settings - Fork 28
Thetis meeting agenda 2020 04 17
Date/time of meeting 2020-04-17 13:00 UTC (14:00 BST)
Present: David Ham, Tuomas Karna, Stephan Kramer, Mariana Clare, Koki Sagiyama, Joe Wallwork
2D tracer model currently only supports Crank-Nicolson. See tracer-timesteppers branch for implementation of other schemes.
Tuomas: Progress on wetting-drying, pressure gradient cancellation method, conservative swe, HLL fluxes
See Slack for Balzano test video. Cancelling pressure gradient term is sufficient for drying phase, but there is a problem with wetting case.
Roe fluxes can dry out a cell, whereas HLL would not.
All the pieces are there for a very good wetting and drying scheme, but it might be quite slow due to the slope limiters. Also not obvious how to extend to implicit time integrators.
SK asked about the status.
DH: all the bits are there to make this happen, but it hasn't been implemented yet. Masters student may work on this in June.
SK: Addressed in PR #206
DH: See Firedrake docs for info on generating parallel-safe random fields in Firedrake.
See Firedrake Issue #1648 for discussion of the coefficient-like Filtered object introduced by KS.
Filtering conceptually similar to indexing. Allows assembly of certain Jacobian rows or certain residual components.
Will allow us to solve problems on sub-domains.