Improved $b$-jet Energy Correction for $H \to b\bar{b}$ Searches at CDF

TL;DR

This work addresses the limited $b$-jet energy resolution in $H\to b\bar{b}$ searches by developing a neural-network regression that combines calorimeter, tracking, and secondary-vertex information to correct $b$-jet energies. Applied to $WH\to \ell\nu b\bar{b}$ at CDF, the method yields a dijet mass resolution improvement from ~15% to ~11% and a roughly 9% gain in expected Higgs sensitivity in the most sensitive category. It demonstrates data/MC validation of the correction variables, linearity across Higgs masses 100–150 GeV/$c^2$, and substantial background rejection improvements for the dominant $b\bar{b}$ processes. The approach also shows potential applicability to related measurements such as $WZ\to\ell\nu b\bar{b}$ and single-top production analyses.

Abstract

We present a method for improving the $b$-jet energy resolution in order to improve the signal sensitivity in searches for particles decaying to a $b$ quark and anti-$b$ quark. A correction function is computed for individual jets, which combines information from the secondary vertex tagger, the offline tracking and standard calorimeter-based jet-energy reconstruction algorithm in order to provide a more accurate measurement of the true $b$-quark energy. We apply the correction to Monte-Carlo-simulated jets in the process $WH \rightarrow \ell νb\bar b$ and find an improvement in both the mean and the resolution of the $b$-jet energy with respect to the $b$-quark energy. The correction improves the measured Higgs dijet invariant mass resolution from $\sim$ 15%(standard jet corrections) to $\sim$ 11%(improved jet corrections) in the Higgs mass range from 100 GeV/$c^2$ - 150 GeV/$c^2$. Using the corrected $b$-jet energies instead of the standard calorimeter-based $b$-jet energies results in a $\sim$ 9% improvement in the expected sensitivity for Higgs boson production cross section in the most sensitive search region of the $WH \rightarrow \ell νb\bar b$ analysis, which is two tagged jets and one charged central lepton.

Figures (20)

• Figure 1: Schematic view of CDF Run II detector.
• Figure 2: Figure demonstrating the reconstructed primary vertex, the secondary vertex and displaced tracks resulting from the $B$-hadron decay.
• Figure 3: The left plot shows that the measured raw jet Et is correlated with the generated true quark energy. The blue (solid) line shows the raw jet Et for generated quarks Et $<$ 50 GeV, the red (dotted) line for Et $>$ 50 GeV. The right plots shows the correlation for the level 5 corrected jet Et.
• Figure 4: The left plot shows that the level-5 corrected jet Pt is correlated with the generated true quark energy. The blue (solid) line shows the level-5 corrected jet Pt for generated quarks Et $<$ 50 GeV, the red (dotted) line for Et $>$ 50 GeV. The right plots shows the correlation for the fitted secondary vertex Pt.
• Figure 5: The left plot shows that the measured secondary vertex position in the xy-plane has a correlation with respect to the generated true quark enertgy. The blue (solid) line shows the secondary vertex position in the xy-plane for generated quarks Et $<$ 50 GeV, the red (dotted) line for Et $>$ 50 GeV. The right plots shows the correlation for the uncertainty of the secondary vertex position.