Closed JunoRavin closed 1 year ago
One quick comment about the Regression system: there are various instances where if we change something in the code (say something that would constitute non-commutativity of floating point arithmetic) and we compare 0 and 0 (say comparing M1i in some of the LBO tests where there is no M1i) then it will say the test fails and it will appear like the errors are large. Not sure what to about this in general. Just wanted to make this note that I did check that this slight change to the Vlasov LBO through G2 and everything looks good modulo ostensibly "large" differences comparing 1.0e-18 and 1.0e-18.
We now set the pointers in the LBO once at initialization, similar to how Vlasov works (and VlasovDG.lua in G2). Goes with branch https://github.com/ammarhakim/gkylzero/tree/app-cleanup which mostly focuses on the G0 level.
Note that this branch also removes some unneeded link flags on Macs that cause issues and I am not sure why they were added back in.
Regression tests pass.