Difference between revisions of "Problems with track finding in CDC"
From GlueXWiki
| Line 5: | Line 5: | ||
[[File:lambda4.png]] | [[File:lambda4.png]] | ||
| − | (March 5, 2012) Algorithm for dealing with bogus tracks | + | == (March 5, 2012) Algorithm for dealing with bogus tracks == |
* Compare bit map of hits between two CDC seeds | * Compare bit map of hits between two CDC seeds | ||
* If 2 seeds share more than 90% of the same hits, keep the seed with the best χ^2/ndf from the circle fit | * If 2 seeds share more than 90% of the same hits, keep the seed with the best χ^2/ndf from the circle fit | ||
Revision as of 10:35, 5 March 2012
Generated Lambda events with the particle gun. The following screen grabs from hdview2 illustrate a very common problem with these events: already used segments get associated erroneously with other already used segments to form bogus track candidates.
(March 5, 2012) Algorithm for dealing with bogus tracks
- Compare bit map of hits between two CDC seeds
- If 2 seeds share more than 90% of the same hits, keep the seed with the best χ^2/ndf from the circle fit
- If 2 seeds share more than 50% of the same hits (but less than 90%) and the reduced χ^2 values of the circle fits from each seed differ by more than 1, keep the seed with the best χ^2/ndf from the circle fit; otherwise keep both
The same event as above using the new algorithm:
