CDF Projected Top Mass Uncertainty

The Top Mass Working Group


Download this plot: (eps) (gif) (script)	Download this plot: (eps) (gif) (script)
The plot on the left is an update of our original Run II CDF top mass projection plot, made when we had analyzed 680 pb-1 of data. Additional data points from 1 fb^-1, 2 fb^-1 and 3 fb^-1 combinations show that so far our measured uncertainty does indeed fall between the two blue lines. See below for details.	The plot on the right extrapolates from the current ~3 fb^-1 dataset. See below for details of the procedure used for the projection.

Purpose of the plots

To make a good faith projection about how precisely CDF can measure the top quark mass as a function of integrated luminosity.
To advertise that we're doing much better than extrapolations based on Run I experience. With only 1/3 the data, we surpassed our goal of measuring Mtop to 3 GeV (that's the star, which is the projection from the CDF Run II TDR).
To understand how precisely can CDF measure the top quark mass by the end of Run II

How the plots were made

Since projections are hard, we try to bracket what we might reasonably expect to achieve by the two curves that are shown on each plot.

The solid curve keeps all the systematic uncertainties fixed at their present values, but scales the statistical and in-situ JES uncertainty (from the W->jj constraint in the lepton+jets and all-jets channels) with luminosity. The projected CDF combined top mass uncertainty at a given luminosity is then estimated by repeating the present CDF combination using these projections for the statistical and in-situ JES uncertainties. This curve is then an estimate of the worst we expect to do since it assumes no improvements to the method or in our understanding of the systematics.

The dotted curve scales the total uncertainty with luminosity. It is meant to represent a sort of lower bound on how well we might do. Certainly it would be surprising if the total uncertainty improved faster than that! It is partially motivated by the fact that through 1 fb^-1, this is, in fact, how our total uncertainty was seen to scale. Clearly the methods we use and our understanding of the systematics have to improve to follow this curve.

We DON'T want to claim that we think we'll follow the dotted curve.

We DO want to say that we're reasonably confident that we'll land somewhere in between these two curves.

Other Details

The solid curve is made by repeating the CDF combination as described above assuming luminosities of 1, 2, 3, 4, 6, 8 and 10 fb^-1. This accounts for some of the "kinkiness" in the plot. In particular, the improvement from 1 to 2 fb^-1 comes mostly from the increase in statistics, while the improvement from 2 to 3 fb^-1 is coming from the added data and from refining the all-hadronic measurement technique.
The triangles are CDF combined top mass at 110 pb^-1 (Run I), 320 pb^-1 (first Run II publications), 680 pb^-1, 1 fb^-1, 2 fb^-1 and 3 fb^-1 (most recent preliminary results). The Run II combinations include the Run I measurements, although they carry little weight. See here for more details concerning the CDF combination.
The first time a Run II measurement in the all-hadronic channel was used in a CDF combination was for 1 fb^-1. In the plot on the left, the Run II all-hadronic channel was not included in the blue projection lines. The data points at 1 fb^-1 2 fb^-1 and 3 fb^-1 do include a Run II all-hadronic measurement. The published Run I measurement in the all-hadronic channel is included in all of the combinations and projections.
The solid curve is labelled as scaling the "stat" with luminosity. As explained above it's stat+in-situ JES, whose sum will scale with 1/root(L) as advertised. In our measurements we actually determine stat+iJES from the likelihood contours. The separation into stat and iJES is a bit articifial and the shorthand "stat" made for an easier reading plot.
The DM/M<1% is meant to highlight that we are able to measure the top quark mass with a precision below 1%. The current precision is 0.85%. When presenting you should be sure to mention that we are actively studying potential systematic effects that could become important at that level of precision. Color reconnection effects have already been added, and we are currently investigating the effect of higher order Feynman diagrams.
If CDF alone can go below 1% precision, the combined Tevatron determination will do better still. It's not clear whether there will be a joint CDF/D0 projection to the end of Run II.

Contacts:

( CDF Top Physics Group | CDF Homepage )

Last updated: March 5, 2009