Implement linear interpolation for both cell-centered and nodal data types #3638
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Closing and re-opening to trigger CI. |
EZoni
reviewed
Jan 25, 2023
Source/Parallelization/WarpXComm_K.H
Outdated
Comment on lines
60
to
70
| wj = (sj == 0) ? 1.0_rt : (rj - amrex::Math::abs(j - (jc + jj) * rj)) | ||
| wj = (sj == 0) ? (rj - amrex::Math::abs(j + 0.5 - (jc + jj + 0.5) * rj)) | ||
| / static_cast<amrex::Real>(rj) | ||
| : (rj - amrex::Math::abs(j - (jc + jj) * rj)) | ||
| / static_cast<amrex::Real>(rj); | ||
| wk = (sk == 0) ? 1.0_rt : (rk - amrex::Math::abs(k - (kc + kk) * rk)) | ||
| wk = (sk == 0) ? (rk - amrex::Math::abs(k + 0.5 - (kc + kk + 0.5) * rk)) | ||
| / static_cast<amrex::Real>(rk) | ||
| : (rk - amrex::Math::abs(k - (kc + kk) * rk)) | ||
| / static_cast<amrex::Real>(rk); | ||
| wl = (sl == 0) ? 1.0_rt : (rl - amrex::Math::abs(l - (lc + ll) * rl)) | ||
| wl = (sl == 0) ? (rl - amrex::Math::abs(l + 0.5 - (lc + ll + 0.5) * rl)) | ||
| / static_cast<amrex::Real>(rl) | ||
| : (rl - amrex::Math::abs(l - (lc + ll) * rl)) |
There was a problem hiding this comment.
What do you think about aligning the code in the following way? I feel like it'd make it quite more readable (it doesn't matter too much if we go over 100 characters in these few lines):
wj = (sj == 0) ? (rj - amrex::Math::abs(j + 0.5_rt - (jc + jj + 0.5_rt) * rj)) / static_cast<amrex::Real>(rj)
: (rj - amrex::Math::abs(j - (jc + jj ) * rj)) / static_cast<amrex::Real>(rj);
wk = (sk == 0) ? (rk - amrex::Math::abs(k + 0.5_rt - (kc + kk + 0.5_rt) * rk)) / static_cast<amrex::Real>(rk)
: (rk - amrex::Math::abs(k - (kc + kk ) * rk)) / static_cast<amrex::Real>(rk);
wl = (sl == 0) ? (rl - amrex::Math::abs(l + 0.5_rt - (lc + ll + 0.5_rt) * rl)) / static_cast<amrex::Real>(rl)
: (rl - amrex::Math::abs(l - (lc + ll ) * rl)) / static_cast<amrex::Real>(rl);Please remember to add the _rt suffix to the real constants, too.
There was a problem hiding this comment.
Yes, that looks better. Thanks, @EZoni !
for more information, see https://pre-commit.ci
EZoni
reviewed
Jan 25, 2023
Co-authored-by: Edoardo Zoni <59625522+EZoni@users.noreply.github.com>
EZoni
reviewed
Jan 25, 2023
Source/Parallelization/WarpXComm_K.H
Outdated
Comment on lines
60
to
65
| wj = (sj == 0) ? (rj - amrex::Math::abs(j + 0.5_rt - (jc + jj + 0.5_rt) * rj)) / static_cast<amrex::Real>(rj) | ||
| : (rj - amrex::Math::abs(j - (jc + jj ) * rj)) / static_cast<amrex::Real>(rj); | ||
| wk = (sk == 0) ? (rk - amrex::Math::abs(k + 0.5_rt - (kc + kk + 0.5_rt) * rk)) / static_cast<amrex::Real>(rk) | ||
| : (rk - amrex::Math::abs(k - (kc + kk ) * rk)) / static_cast<amrex::Real>(rk); | ||
| wl = (sl == 0) ? (rl - amrex::Math::abs(l + 0.5_rt - (lc + ll + 0.5_rt) * rl)) / static_cast<amrex::Real>(rl) | ||
| : (rl - amrex::Math::abs(l - (lc + ll ) * rl)) / static_cast<amrex::Real>(rl); |
There was a problem hiding this comment.
At this point, I wonder if it would be slightly better to remove the if conditions on sj, sk and sl from these nested loops, and have them only once instead, before the loops. sj, sk and sl do not vary depending on jj, kk and ll, so in principle there would be no need to check them within these nested loops.
I think we could do something like the following:
wj = (rj - amrex::Math::abs(j + hj - (jc + jj + hj) * rj)) / static_cast<amrex::Real>(rj);
wk = (rk - amrex::Math::abs(k + hk - (kc + kk + hk) * rk)) / static_cast<amrex::Real>(rk);
wl = (rl - amrex::Math::abs(l + hl - (lc + ll + hl) * rl)) / static_cast<amrex::Real>(rl);where hj, hk and hl are defined, before the nested loops, as
amrex::Real hj = (sj == 0) ? 0.5_rt : 0._rt;
amrex::Real hk = (sk == 0) ? 0.5_rt : 0._rt;
amrex::Real hl = (sl == 0) ? 0.5_rt : 0._rt;What do you think?
There was a problem hiding this comment.
yes, I like it this way. Thank you! I will make the change and push it. Also, thank you for resetting the CI benchmarks!
EZoni
approved these changes
Jan 31, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR implements linear interpolation for both cell-centered and nodal arrays. Before this, we were doing nearest grid point interpolation for cell-centered directions, which was resulting in "stair-casing" in the
auxfield. This effect was particularly intensified with larger mesh refinement ratio (e.g. 4).