Closed marcelo-alvarez closed 9 months ago
The 2lpt implementation is in the 2lpt branch (https://github.com/exgalsky/xgfield/tree/2lpt). This was failing to run on Perlmutter because it was going out of memory.
@1cosmologist can we go ahead and create a PR to track this branch even if it's not passing tests? Also, do you recall whether it was going out of memory in 1lpt mode?
It seems reasonable to me to do the running and testing as a part of the PR. I did succesfully run the 1lpt code but not sure if the grid sizes were same.
2LPT branch successfully ran on Perlmutter. I think we are ready to evaluate if the 2LPT implementation works as intended and then move on.
https://github.com/exgalsky/xgfield/pull/8#issuecomment-1933156551
Need to implement 2LPT given the input linear density contrast realization from Websky, for which there already exists first and second order displacements to compare to.
(transferred from marcelo-alvarez/lpt4py#1)