PCMS interpolation functionality#148
Conversation
Interpolator
searches and adapts radii until all targets get the minimum number of supports and debug queue (counter overflow)
test/test_rbf_interp.cpp
Outdated
|
|
||
| const auto& ntargets = mesh.nverts(); | ||
|
|
||
| Write<Real> radii2( |
There was a problem hiding this comment.
This radii2 variable is never used.
| using namespace Omega_h; | ||
|
|
||
| template <typename Func> | ||
| Write<Real> mls_interpolation(const Reals source_values, |
There was a problem hiding this comment.
We will need to have an overload of this function that also works with ScalarArrayView.
There was a problem hiding this comment.
A test for one mesh vertex to another mesh vertex is also needed.
| "positive but the value is %.16f\n", | ||
| rho_sq); | ||
|
|
||
| OMEGA_H_CHECK_PRINTF(r_sq > 0, |
There was a problem hiding this comment.
Why is it not allowed for r_sq to be 0? I think, it will occur when it has a support location concurrent to a target which has a very high chance of occurrence in regular meshes.
There was a problem hiding this comment.
I have checked (see here), and it works when there is no coincident source and target.
Edit: I removed the last line which was wrong.
with 10% error margin and [temporary] it doesn't work when a source and a target becomes coincident: discussed here: SCOREC#148 (comment)
jacobmerson
left a comment
jacobmerson
left a comment
There was a problem hiding this comment.
I didn't get all the way through this, but there are a lot of changes that look like they are just modifications to the whitespace. We need to remove these whitespace changes from the current pull request. I'm looking into a strategy to have clang-format in CI/CD to make this easier to handle in the future.
]
Outdated
There was a problem hiding this comment.
This file looks like you may have accidentally saved the commit message. Please remove it.
there were nested loops reusing i and given the length of the loop body, this helps readability
use league_rank as loop control var
4 participants
This pull request brings in @abhiyan123 initial changes for dealing with RBF field interpolations. I have gotten this into a state where it can be merged with develop and tests/compilation passes.
However, I have not done a formal review of this code and there are a some issues that need to be addressed before merging with develop.
@Fuad-HH I have requested you as a reviewer. Please take a look and see if you have any input, questions, etc.
@cwsmith relevant code for our discussion on surface fitting is here. Unfortunately, getting everything merged took some time and I have not had time to do the proper review I wanted before our meeting. Let me know if you prefer to push back that discussion until Thursday/Friday.