New Work
________
Oct 26, 2007
Examining run 5022009.
Wrong drift velocity getting used
_________
Oct 29, 2007
Turns out it was a bug (RT #1082) in the new code for interpolation drift velocities. Solved.
Next problem: central membrane stripes not at z=0:
Took me a while to figure this one out: T0 of laser is simply wrong. I will need to move the hits (not a change in d.v., just a straight shift) to get this right. Since I don't care about the east side for these studies, I can just subtract 6.45cm and that should be good enough.
It's important that the d.v. be correct as it is used in the OmegaTau calculation in the code. It's also important that the hits be at approximately the correct z location so that the fit of the laser data to the distortion map corresponds in z. Being off by a millimeter or two certainly won't make a difference there.
----
still using filterLaser.C with a local version of StLaserEvent to make the data members public:
root4star -l -b -q 'filterLaser.C+("st_junk3_5022007*")'
creates ntuples for each sector. I really only care about the west sectors (2,4,6,8,10,12).
________
Oct 30, 2007
In /star/institutions/bnl/genevb/OmTau/data/analysis
Doing only sectors 2-12 (west side)...
root4star -l -q 'findLasers07.C("../NoCorr/5022006/","Nominal")' > & log.nominal &
removed a few lasers and ended up with: Nominal_good.dat
root -l -q 'plots07.C+("Nominal_good.dat",6,7,706,0.5,"./%03d/")' > & log.706 &
In Hists/Examine7a
root -l -q '../DrawDiff.C("../Hists_7.root","../../Nominal_good.dat")'
fits.Draw("z","re!=0")
shows z values of 15+45+75+105+135+165. Using zr shows the better z value distributions.
currents were measured every 180 seconds (3 minutes).
In /star/institutions/bnl/genevb/IFC_CURRENT/2004/cur.root
cur.Draw("Iow-Iiw:t-1873808","run>5022004&&run<5022011")
cur.Draw("Iiw:t-1873808","run>5022004&&run<5022011","")
cur.SetMarkerColor(2)
cur.Draw("Iow:t-1873808","run>5022004&&run<5022011","same")
There might have been a problem with the very last run, which ends at 1874654. Though the last current measurement is at 1874708, so it's unlikely a problem.
But I don't see any indication that the OFCW current was high for these runs. So I shouldn't see any distortion in the nominal GG run, but I do!!! Run 5022006 is supposed to be at nominal GG voltage (-115), but I still see distortion:
fits.Draw("r+0.05:x:z>>ppp(36,0,180,60,50,200)","re!=0","prof lego2")
Though this is smaller than the distortion seen in the -140V run (5022007):
fits.Draw("d+0.1:x:z>>ddd(36,0,180,60,50,200)","de!=0","prof lego2")
_________
Oct 31, 2007
There is no current measurement for runs 506310*, it wasn't going into the DB. I would have to dig through the slow controls archive again :-(
Processing runs from day 63, but in the meanwhile, here's a difference plot from the GG runs using the -140V distorted minus the -115V nominal:
fits.Draw("r+1.0:x:z>>ref(36,0,180,60,50,200)","re!=0","prof lego2")
fits.Draw("d+2.0:x:z>>dis(36,0,180,60,50,200)","de!=0","prof lego2")
refp = ref
disp = dis
TProfile2D diff(*disp)
diff.Add(refp,-1)
diff.SetMinimum(0.44)
diff.Draw("lego2")
This plot shows very nice curves as I was expecting to see. However, I'm a bit concerned about the offset. I used distortion+2.0, and reference+1.0, so the difference should have a baseline at 2.0-1.0 = 1.0. But the middle of this plot is at 0.5. Did I screw something up? It might have something to do with subtracting profile histograms.
Aha! So it is. Here's the right way:
fits.Draw("d-r+1.0:x:z>>ref2(36,0,180,60,50,200)","de!=0&&re!=0","prof lego2")
ref2.SetMinimum(0.90)
ref2.Draw("lego2")
That's better. Looks like it's the same histogram scaled by 2. Anyhow, it looks good! So I should be able to do my fitting by using (d-r) as my measure of the distortion. Actually, not sure in the end why I need the profile hist to see this. Here it is:
fits.Draw("d-r+1.0:x:z:d-r","de!=0&&re!=0")
For a qualitative (near-quantitative) comparison, here's the expected distortion map for omtau = 2.3 in /star/institutions/bnl/genevb/OmTau/JIMCODE:
TNtuple aa("aa","Aaa","omtau:x:z:d")
aa.ReadFile("phi_140.dat")
aa.Draw("d+1:x:z:d","abs(omtau-2.30)<0.001&&z<170")
I'd say it's looking pretty good :-) Though I didn't in the end really need to add 1.0 to the difference in either plot.
And now for the probability fit.....
/star/institutions/bnl/genevb/OmTau/fit/fit.C
root -l 'fit.C("../data/analysis/Hists/Examine7a/fits.root","../JIMCODE/phi_140.dat",0)'
...
Maximum prob of 0.532667 at: omegatau=2.579999
I had to set adderrr = 0.0063 to get this plot. At 0.008 I get lots of probabily=1 values. And at 0.005 all the probabilities are too small. So it's pretty sensitive. The glitches in the upper end of the chi squares plot is where it decides to drop points based on large chi values, I believe. This happens a lot more when I reduce adderr.
So perhaps all that's left is to do this for all my distortions, multiply the probabilities, and see what we get :-)
Works well....but we can do better...
Looking at where the outliers are, I need to put in a cut on hit errors. Looking more closely, there appears to be a z-dependence on the hit errors:
fits.SetMarkerColor(2)
fits.Draw("re:z","de!=0&&re!=0&&de<0.004&&re<0.004","")
fits.SetMarkerColor(1)
fits.Draw("re:z","de!=0&&re!=0&&de<(0.003-z*0.00001)&&re<(0.003-z*0.00001)","same")
fits.SetMarkerColor(4)
fits.Draw("re:z","de!=0&&re!=0&&de<0.004&&re<0.004","profs same")
But after playing with lots of cuts like the z-dependent one above, I'm really not too worried about these large error hits. They don't contribute too much to the fit in the end because they contribute as the error squared. So removing them doesn't do much at all. I might as well leave them in. I'll just put a cut at 0.005 to remove the wacky ones.
______
Nov 1, 2007
After learning about the z-dependence issues and fixing them, I have redone the day 22 final results. They want:
-65V: 2.33
-90V: 2.35
-140V: 2.41
-165V: 2.40
All rather consistent! The results from multiplying all the probabilities from day 22 gives 50% probability at omegatau = 2.38.