Note the time change:  9am Pacific / 12 noon Eastern / 6pm Europe

Agenda:

  1. GBL update                   --- Pelle
  2. Update of the tracking code     --- Norman
  3. Summary of Conditions page   --- Jeremy
  4. Other questions, updates, issues

Quick notes on the meeting:

New people at the meeting:

Jacopo Ferretti, new postdoc with Massimo, in Italy.
Gabriele Similar, new person, University Padova, Italy, has experience with Babar tracking, tracking software interest.

Pelle, New track fitter, GBL:

Added/investigated a generalized broken line (GBL) scattering algorithm, which is also used for the CMS detector alignment.
Make initial fit, then in the second fit, you refit with a broken line.
Millipede is supported with this framework, integrated already, so that all the derivatives are available already for the next step.

"Proposal 2014" v4 detector geometry. 40 MeV A', beam spot at 0,0,0 Current implementation in Python, will need to port to JAVA later.

Chi2 probability plots are flat, with exception of the peak at 0 which are bad tracks. (Look up chi2 prob plots)

Sees improvement on the track momentum resolution of ~25% Bend plane impact parameter improves 13%
Non bend plane impact parameter improves 15%

No fringe field, simulation was with uniform field.

Next step, see if works with test run simulation, and then the test run data. Move on with the alignment.

It is fast, linear fit, not least squares.
The initial fit has to be reasonably close to the true helix. If you are too far away, you can end up with problems. All other experiments use a one step circle fit. So it matters but not that much.

Plans for JAVA, requires linear algebra, matrix decomposition.There are java libraries for doing this, some custom libraries written for this. Estimate 2 weeks if the libraries exist. Implementation is Python and C++, millipede is written in
Fortran.

The low chi2 prob tracks, this is probably a problem with the initial fitter and not GBL.

Norman, Kalman Filter update

Magnetic field, have 2d map, want to go to the full field map, HPS_b18d36_unfolded.dat
Uses linear interpolation.
Classes written in C++ and Java.

Testing, see slides.
SurfZPlane, now can be on the detector z plane.
Needs to use the appropriate multiple scattering for electrons, and appropriate materials.
Next steps, SLIC simulation of HPS, then test data.

Within the dipole, the field is very flat, but this is from the mid plane measurements. It could well be that we won't notice much difference. If needed we could also add the fringe fields to the edges of the uniform field, which may be faster.

Need a better measurement of the magnet?

All:

We will look into how to best get the new members in the collaboration started with the lcsim software. 

  • No labels