Squaw Tuning of Gamma Angles

J.M.Bishop, 3/6/96

We have been attempting to tune the Squaw kinematic fitting program for E852. As has been mentioned at the last few collaboration meetings, a persistant effect has been seen where the gamma angles are being spread away from the forward direction in the process of fitting the events. Rob Lindenbush and I have tried several methods to address this effect. We have effectively moved the LGD closer to the target, we have recalibrated with 2 different angle corrections to the depth correction, and we have explicitly moved the angles out before fitting. We have found that to produce a sufficient effect on the fitted angles, we had to change the angles by too much to be physical, or to change the depth correction to an unphysical value. The pulls on the gamma quantities did respond to large changes in the angles, but realistic changes barely affected the pulls. Also, changing the angles actually made the peak of the missing mass squared move to higher values, the opposite direction of what was needed. The angle changes did move the gamma-gamma effective mass in the right direction.

As we had realized early in this tuning process, the angles of the gammas and the energies of the gammas are coupled. Since the angle spread had been so obvious we tried to cure the problem by changing the angles and waiting until we got that set before tuning the gamma energies. Since we did not have success adjusting the angles, we decided to try adjusting the gamma energies, with no angle change. As one distribution indicated perhaps a 0.4% shift in the energies in the process of fitting, the first attempt was to scale all gamma energies up by 0.4% (egamma = 1.004*egamma). This immediately improved the angle pulls, as well as the gamma-gamma effective mass and the missing mass. The next step tried was to increase the gamma energies by 1.0% (egamma = 1.01*egamma). This further improved the centering of the angle pulls, and overshot the centering of the curvature pulls on the gammas. For several topologies, a 1% energy shift moved the gamma-gamma effective mass and the missing mass to near the correct values, but for some topologies a 1% shift overshoots both a little bit. Rob now agrees that the problem is probably energy related rather than angle related, but he does not think that an across the board shift is necessarily the right correction. Rob is currently looking to see if neutral event selection for calibration is distorting the calibration. My use of a simple 1% increase was primarily to see what the effect would be, not that I necessarily believed that this was the proper correction. The fact that it improves the fits indicates that a correction of this sort is probably appropriate.

The following tables show some of the effects of changing the angles and energies of the gammas. All samples shown are with the current errors and with the MPS field, and thus charged track momenta, scaled up by 0.3%, as was concluded at the last collaboration meeting and implemented in the SCCS libraries shortly thereafter.

			Table I
4 gamma all neutral events (eta-eta to 4 gammas)
		std errors	nom dth	10x dth	20x dth	.4% de	1% de
1c pull, +az	0.135		0.123	0.045	-.035	0.067	-.012
1c pull, -az	-.128		-.127	-.080	-.019	-.098	-.050
mv pull, +az	0.041		0.028	-.106	-.300	0.017	0.009
mv pull, -az	-.037		-.032	0.131	0.269	-.028	-.020
mv curv pull	-.075		-.069	-.030	-.010	0.003	0.114

MMsq		1.00		1.05	1.15	1.25	0.955	0.92
gg efm (eta)	0.541		0.542	0.545	0.547	0.544	0.546

	nom dth	means using the nominal values of angle change (dtheta)
	10x dth means using 10 times the nominal dtheta
	20x dth means using 20 times the nominal dtheta
	.4% de means scaling all gamma energies up by 0.4%
	1% de means scaling all gamma energies up by 1.0%


			Table II
Several data samples at standard errors and at 1% de
		022ee	4g ee	133ee'	133e	022e	6g	3pi
std err
MMsq		0.867	1.00	0.97	0.964	0.923		0.871
gg efm (eta)	0.538	0.541	0.544	0.545	0.522
gg efm (pi0)	0.128

1% de gamma
MMsq		0.85	0.92	0.947	0.979	0.881	0.80
gg efm (eta)	0.551	0.546	0.546	0.546	0.56	0.54
gg efm (pi0)	0.131					0.132

The data samples are
022ee	022 4 gamma to eta eta, 1 eta to 2 gamma, 1 eta to 3pi
4gee	all neutral 4 gamma to eta eta
133ee'	133 4 gamma to eta eta', eta' to eta 2pi, both etas to 2 gammas
133e	133 2 gamma to 3pi eta
022e	022 2 gamma to 2pi eta
6g	all neutral 6 gamma to etas and/or pi0s
3pi	133 no gamma