Track Covariance Map #814
Track Covariance Map #814
Conversation
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
ax3l
force-pushed
the
topic-track-covariance-map
branch
3 times, most recently
from
3b71d60 to
a91748a
Compare
ax3l
force-pushed
the
topic-track-covariance-map
branch
from
a91748a to
ed344c9
Compare
for more information, see https://pre-commit.ci
for more information, see https://pre-commit.ci
| // small trick to force every derived class has to implement a method transport_map | ||
| // (w/o using a purely virtual function) | ||
| T_Element& element = *static_cast<T_Element*>(this); | ||
| cm *= element.transport_map(cm); |
There was a problem hiding this comment.
Is this logic right? The function transport_map in each element takes the reference particle data as input, but this appears to call transport_map with the covariance matrix cm as input. Also, if the intent is to push the covariance matrix by the matrix R, then the math should look like cm = R * cm * R^T, which appears also ImpactX.cpp.
There was a problem hiding this comment.
Oh, that is outdated. I only fixed it in ImpactX.cpp.
|
Regarding user control of the run mode, I propose |
|
We could also make it more general:
That way we have one flag for all planned modes and we can continue to add modes beyond 3 in the future :D (e.g., ensemble/batch variants of these modes or so) |
|
What about |
|
Perfect |
|
Oh, or instead of mode we name it |
|
That works, too. FYI: I used |
For elements that do not yet have maps.
ax3l
force-pushed
the
topic-track-covariance-map
branch
from
ed335ee to
9c39e1f
Compare
* Update DiagnosticOutput & Reduced Beam Diagnostics --------- Co-authored-by: Axel Huebl <[email protected]>
ax3l
force-pushed
the
topic-track-covariance-map
branch
from
9c39e1f to
fb58249
Compare
Co-authored-by: Chad Mitchell <[email protected]>
`chmod a+rx *py`
Temporary/outdated from earlier code draft. Fix column headers.
Unify spellings
ax3l
force-pushed
the
topic-track-covariance-map
branch
from
833c2fa to
eb8a4f7
Compare
Avoid confusing/mixing data.
People might forget to call `sim.init_envelope`, in which case we do not want them to calculate with zero-valued ref or cm.
Move the fields over as well, only used in particle tracking.
This continues to only push particles.
| if (compute_eigenemittances) | ||
| { | ||
| file_handler << " " | ||
| << "emittance_xn" << " " << "emittance_yn" << " " << "emittance_tn"; |
2 participants
Add a new runtime mode, tracking of 2nd moments in a covariance matrix.
Follow-up: add the implementation of space charge kicks (in envelope mode), following Issue #826