Simone Silvestri
Simone Silvestri
From the discussion #3587 @scott-conn found out that our definition of immersed derivatives ~~does not work very well~~ is **wrong** when using reduced fields. Consider, for example, a field reduced...
this PR allows using a `CuArray` times in conjunction with a `Cyclical{Nothing}` time_indexing
This PR starts the implementation of the conversion between frames described in #3625 cc @glwagner @siddharthabishnu
1. I was able to run the example provided in https://gaelforget.github.io/MITgcm.jl/dev/examples/ on a Linux computer but not on my Mac M1. After some investigation, most likely it was a problem...
This PR is just a test to check that the documentation build triggers correctly on the GPU on the Caltech cluster and the results are literated correctly. The setup now...
At the moment, the tendencies were updated at each call of `update_state!`. This PR makes it so that the tendencies are computed only inside the time-stepping where we need them....
This was already possible for tracers with the `TracerAdvection` scheme. This PR changes the `TracerAdvection` scheme into a `FluxFormAdvection` scheme that can be used also for momentum closes #3699 closes...
closes #3681 We can probably use this PR to add also a validation for the `PartialCellBottom` and the bug-fixes in the implementation. What do people think? cc @jm-c
The Adams-Bashforth time stepper relies on an approximate time-integral to calculate the tendencies. This time integral, when using a constant time step, is simply approximated as ```math G^{n+1} = \left(\frac{3}{2}...
To asynchronously fill the halos of distributed fields, the code uses an incremental counter to track how many MPI requests are live and update the MPI send and receive tag....