Conversation
|
The author of this PR, JopHendrikx, is not an activated member of this organization on Codecov. |
|
@JopHendrikx it seems that commits 8d54025 and 923ac8b should not be part of this PR. Is that correct? |
|
@JopHendrikx, I was going through the open pull requests, then stumbled on this one. I noticed 1) a question of @daanboer about some patches that are (possibly) mistakenly part of this request, and 2) merge conflicts. Will you need help with any of this? Apart from the technical issues: we should think carefully about this change. This will result in deviations w.r.t. to the MATLAB code, so it would be better to first agree with @IST-Lisbon and @AntonioTejero about the desired operation. (A convincing before vs. after testcase will help, of course.) |
daanboer
force-pushed
the
cs-cell-interpolation
branch
3 times, most recently
from
11d85ef to
5790f0e
Compare
|
@janvdijk, @JopHendrikx the commit history has been cleaned up and merge conflicts are resolved (rebased onto main). In theory this could be merged (perhaps with some additional testing), but I agree that it is better to first discuss with @IST-Lisbon, and @AntonioTejero. |
@daanboer, after #168 gets merged, I plan to implement the parts of #116 that are described in #169 first. That will significantly reduce the size of #116 and avoids the controversial parts of it. It will just: 1) calculate the cell-values of This should be ... the same in many other places. I will add that change to the cs-cell-interpolation branch. UPDATE: @daanboer, I just looked, but that is too messy. I will just wait for your review of #168 (and hopefully its merger), make these change afterwards on a separate branch, then merge those back to #116. |
daanboer
force-pushed
the
cs-cell-interpolation
branch
from
5790f0e to
dbd151b
Compare
* Grid.h: add free function interpolateNodalToCell * EedfCollision.*: use that, store the totalCellCrossSection and provide an accessor. * ElectronKinetics.cpp: use totalCellCrossSection(), do not repeat this task multiple times. See issue #169. Note that this does the non-controversial parts of #116.
* Store totalCellCrossSection, centralize interpolation. * Grid.h: add free function interpolateNodalToCell * EedfCollision.*: use that, store the totalCellCrossSection and provide an accessor. * ElectronKinetics.cpp: use totalCellCrossSection(), do not repeat this task multiple times. See issue #169. Note that this does the non-controversial parts of #116. * Add a test for interpolateNodalToCell. Test that grid.getNodes() is correctly interpolated up to grid.getCells(), up to essentially the machine accuracy. * Add documentation for interpolateNodalToCell. * Integrals.h -> GridOps.h, add cell-derivation File Integrals.h has been renamed into GridOps.h. It now hosts various grid-related operations: interpolation, differentiation and integration. The inline interpolation function has been moved there from Grid.h. Tests have been added for the new cell->cell differentiation function cellDerivative, which has been extracted from ElectronKinetics (function calculateFirstAnisotropy). A test has been added (that u'=1). * Minor comment fixes and additions. * Remove unused variable 'n'. --------- Co-authored-by: Jan van Dijk <jan@epgmod.phys.tue.nl>
daanboer
force-pushed
the
cs-cell-interpolation
branch
from
dbd151b to
b25b7a2
Compare
daanboer
force-pushed
the
cs-cell-interpolation
branch
from
b25b7a2 to
a8c96b3
Compare
|
I will close this for now, as this might be fixed implicitly by merging #232, or similar changes. |
3 participants
Calculate cell cross sections on the fly for inelastic/ionization/attachment operators, together with corresponding output parameters. Removes double interpolation, such that linear interpolation errors are as minimized as possible. Also with this the numerical threshold is found on the grid cells instead of at the nodes.
Operator distribution is now defined from target cell towards source cell.