Backward search implementation #345
Conversation
…h pickup points are defined.
…as there are a lot of seeds with hits only in TOB1 and TOB2.
- introduce new member in IterationConfig - IterationParams m_backward_params to be used during backward search - add --no-backwards-search commandline option for standalone
|
what about the changes in RecoTracker/MkFit/plugins/createPhase1TrackerGeometry.cc ? are they still necessary for the PR to work? |
|
oh yes, the same ... there is extra stuff in there for bkw-search params since the previous commit |
|
do we have validation plots for this PR? |
|
@leonardogiannini had tested this PR few weeks ago. @leonardogiannini, could you post a link? |
| if (do_backward_fit) | ||
| { | ||
| // a) BackwardFitBH works on MkBuilder::m_tracks | ||
| builder.BackwardFitBH(); | ||
| builder.BackwardFit(); |
There was a problem hiding this comment.
this looks like the place that can change the physics results in the initialStep (the rest looks more related to the backward search, which is not used in the initialStep).
@osschar is my guess correct?
I can make an incremental diff to confirm that some time later.
There was a problem hiding this comment.
Hmmh, could be. They used to be the same ... will cross check the code, we can bisect afterwards if I don't find anything.
There was a problem hiding this comment.
I have a check and a proposal for a "revert" in #349
Validation / MTV has been run by Leonardo two weeks back.
CMSSW json files will have to be remade as there is a new member in IterationParams.
At some point we should revisit what propagate-to-PCA does(and when):