Fix offset of interpolation and use second order interpolation in x and y #570
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
MaxThevenet
approved these changes
Jul 26, 2021
src/fields/Fields.cpp
Outdated
| const amrex::IntVect refinement_ratio = m_ref_ratio; | ||
|
|
||
| const int interpol_order = 2; | ||
| const int nguards_xy = std::max(1, Hipace::m_depos_order_xy); |
There was a problem hiding this comment.
It looks like this has to be the same value as m_slices_nguards, so it is dangerous to re-define it (if the other definition changes, this code might break). Could you use directly m_slices_nguards[0] and m_slices_nguards[1]? Or just
AMREX_ALWAYS_ASSERT(m_slices_nguards[0] == m_slices_nguards[1]);
const int nguards_xy = m_slices_nguards[0];
src/fields/Fields.cpp
Outdated
Comment on lines
371
to
377
| if (i==nx_fine_low) { | ||
| x = plo[0] + (i+lo[0]+nguards_xy-0.5_rt)*dx[0]; | ||
| } else if (i== nx_fine_high) { | ||
| x = plo[0] + (i+lo[0]+nguards_xy+1.5_rt)*dx[0]; | ||
| } else { | ||
| x = plo[0] + (i+lo[0]+nguards_xy+0.5_rt)*dx[0]; | ||
| } |
There was a problem hiding this comment.
Could you add some comments to explain these values? Why always add guard cells? Why -0.5/+1.5? In particular, what puzzled me is that:
i x
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
nx_fine_low plo[0] + (nx_fine_low+lo[0]+nguards_xy-0.5_rt)*dx[0];
nx_fine_low+1 plo[0] + (nx_fine_low+lo[0]+nguards_xy+1.5_rt)*dx[0]; # the one above + 2*dx
nx_fine_low+2 plo[0] + (nx_fine_low+lo[0]+nguards_xy+2.5_rt)*dx[0]; # the one above + 1*dx
nx_fine_low+3 plo[0] + (nx_fine_low+lo[0]+nguards_xy+3.5_rt)*dx[0]; # the one above + 1*dx
etc.
Is this expected? Same question around nx_fine_high, and for the y direction.
src/fields/Fields.cpp
Outdated
Comment on lines
392
to
393
| const int j_cell = compute_shape_factor<interpol_order>(sx_cell, | ||
| xmid+nguards_xy-0.5_rt); |
There was a problem hiding this comment.
Could you explain why this is xmid+nguards_xy-0.5_rt? And maybe add a comment to clarify?
This was referenced Jul 27, 2021
MaxThevenet
approved these changes
Jul 28, 2021
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 fixes an issue in the interpolation:
Not the value of the position of the boundary of the fine grid must be interpolated from the coarse grid to fine grid, but the value at the position of the guard cell (on the fine grid) must be interpolated from the coarse to the fine grid.
A second order interpolation in x and y via a stencil was implemented. It is much clearer to read and will allow for a simplification in the upcoming longitudinal refinement in #565.
By fixing the guard cells, this PR is now correct for all deposition orders. Additionally, in the definition of the Poisson solver, the length of the problem was assessed in a level-agnostic way, which gave the wrong result for level 1, leading to the ripples we saw at the boundary. Using the correct length resolves this issue.
It can be seen, that there is still some very slight shift. This is caused by the imperfection in rho. If I hard-code rho to 1, it gives flawless results:
The correct handling of rho at the boundaries will be fixed in an upcoming PR.
constisconst)