Output

[skygering]

Improve output speed and functionality

• JLD2 outputs for floes
• Switch plotting code to use JLD2
• Identify which methods depend on a regular grid and make sure that we are dispatching on grid type for those functions
• Add periodic element (IDs)
• Add stress and strain output to eulerian data writer
• Add attributes to JLD2 file
• Dump initial simulation status to output file
• Checkpoint output writer
• Optional outputs in separate output struct - e.g. energy balance?

[skygering]

Progress made with PR #26

[skygering]

It would be good to add all of the output writers to an output writer holder struct that is then a field in the simulation.

[skygering]

Right now we stop writing when there are no floes. This might not be the optimal behavior, especially once we have packing as more floes could re-enter the simulation

[skygering]

I want to re-write calc_eulerian_data! to be more efficient as it requires so much unnecessary polygon clipping right now. However, this will have to wait until after the switch to GO clipping.

[skygering]

We are swapping over to saving the polygon rather than coordinates for floes.

This then means that we won't have to store the coordinates anymore, or really anything that can be determined from the floe's polygon (area, centroid, area, rmax, angles).

I think maybe we should re-do the floe writers to depend on the Floe's sub-structs... If I break it down so that the floe has very simple writing mechanism, this will make it easier for users to write new writers that give them the information that they need. I will need to think more on how to do this.