WooHyun looked at the Z sample in the Stntuples, and found that we are still misaligned. It is not possible to change the Stntuples, but I can do an on-the-fly correction in an Stntuple module. What is plotted is the muon track phi minus the BMU stub phi.
Top row is North and Bottom row is South, and the left is West and right is East.
In the West South you can see a break at about cell number 144
| West | East | |
| North | slope=1.58E-5 Inter=-0.00012 | slope=5.14E-5 Inter=0.00061 |
| South | slope=6.60E-5 Inter=-0.00895 | slope=3.69E-5 Inter=-0.00449 |
To get the above I had to drop out some fliers from NorthWest. 0 is a difficult point since there is wraparound. The slopes and intercepts are for modified phi (-40 to 180 in cell number).
The original corrections made can be found in the following plots, which give ideal minus corrected wire phi positions for front and back. Notice the large break at about phi 135 in East South. Recall that the front mount does not exist for the short BMU, and their positions are interpolated. Notice further that the pattern of corrections, which was taken from survey measurements, does not closely resemble the deviations found above.
I wrote a spot of code to correct the phi returned from TBaseMuonStub:
float phiCorr(float phi, float eta){
// Estimated from Z sample. Not same as survey corrections
int side=0;
if(eta>0) side=1;
float phit=phi;
float correction;
if (phit>225.*PI/180.) phit = phit - 2.*PI;
if(phit<1.54897971){
if(side==0){ // NorthWest
correction=-0.00012+.00072*phit;} // 1.58E-5
else{ // NorthEast
correction=-0.00061+.00236*phit;} // 5.14E-5
}
else{
if(side==0){ // SouthWest
correction=-0.00895+.00303*phit;} // 6.60E-5
else{ // SouthEast
correction=-0.00449+.00169*phit;} // 3.69E-5
}
return phi+correction;
}
Of course the Muon->fBmuDelX has to be corrected also, but that is simply drphi and so using the phi correction with the proper radius should work fine. Since the global position is available, this should not be a problem. A routine which accepts the TStnMuon as an argument seems like the simplest way to go.
The resulting phi differences, West and East, are:
Compare with old versions
Modified 27-June-2008 at 10:30
http://hep.physics.wisc.edu/~jnb/imu/11Jun2007
