[Paper Review] Combination of Single Top Quark Production Results from CDF

R. C. Group, C. I. Ciobanu+2 more|ArXiv.org|Sep 26, 2008

Particle physics theoretical and experimental studies1 references3 citations

TL;DR

This paper presents a combined measurement of the single top quark production cross section from CDF using two independent methods—NEAT (neuro-evolution of augmenting topologies) and BLUE (Best Linear Unbiased Estimate)—to improve precision by leveraging complementary multivariate analyses. The combined result yields σₛ₊ₜ = 2.2 ± 0.7 pb, corresponding to |Vₜᵦ| = 0.88 ± 0.14 (exp) ± 0.07 (theory), with a 9% improvement in expected sensitivity over individual analyses.

ABSTRACT

Recently, the CDF experiment at the Fermilab Tevatron has used complementary methods to make multiple measurements of the singly produced top quark cross section. All analyses use the same dataset with more than 2 fb^-1 of CDF data and event selection based on W+2 or W+3 jet events with at least one b-tagged jet. However, due to differences in analysis techniques these results are not fully correlated and a combination provides improved experimental precision. Two independent methods are used to combine the results. This combination results in an improved measurement of the single top production cross section and also the CKM matrix element V_tb.

Motivation & Objective

• To improve the experimental precision of the single top quark production cross section measurement at the Tevatron by combining multiple complementary CDF analyses.
• To reduce systematic uncertainties and enhance sensitivity by combining results from different multivariate analysis techniques.
• To validate consistency among three independent single top analyses using statistical combination methods.
• To extract a more accurate determination of the CKM matrix element |Vₜᵦ| from the combined cross section measurement.
• To demonstrate the effectiveness of NEAT and BLUE as robust combination techniques for high-energy physics measurements with correlated uncertainties.

Proposed method

• The NEAT method constructs a neural network super-discriminant optimized via neuro-evolution of augmenting topologies to combine discriminants from likelihood, matrix element, and neural network analyses.
• The BLUE method performs a weighted average of the three cross section measurements using a covariance matrix that accounts for correlations and multiple error types.
• Correlations between analyses are determined using pseudo-experiments from fully simulated events, with correlation coefficients summarized in Table 1.
• Asymmetric Iterative BLUE (AIB) is applied to handle non-Gaussian and asymmetric uncertainties, ensuring unbiased combination results.
• Both NEAT and BLUE use event-by-event output templates for signal and background to fit the data and extract the single top signal component.
• The combination improves expected sensitivity by ~9% (NEAT) and ~7% (BLUE), with consistency checks via pseudo-experiment testing.

Experimental results

Research questions

• RQ1 Can combining multiple multivariate analyses improve the precision of the single top quark cross section measurement beyond individual analysis limits?
• RQ2 How do the NEAT and BLUE combination methods compare in performance and robustness when applied to correlated CDF analyses?
• RQ3 To what extent are the three independent CDF single top measurements consistent with one another and with the Standard Model expectation?
• RQ4 What is the impact of correlated uncertainties and asymmetric errors on the combined cross section and |Vₜᵦ| determination?
• RQ5 Does the combination method enhance sensitivity to deviations from the Standard Model in single top production?

Key findings

• The NEAT combination yields a single top quark cross section of σₛ₊ₜ = 2.2 ± 0.7 pb, representing a 9% improvement in expected sensitivity over individual analyses.
• The BLUE combination confirms the NEAT result with a consistent cross section measurement and achieves a 7% improvement in expected sensitivity.
• The combined measurement of |Vₜᵦ| is determined to be 0.88 ± 0.14 (experimental) ± 0.07 (theoretical), consistent with the Standard Model.
• The χ² test of consistency shows that the combined measurement is better than 87% of pseudo-experiments, indicating high compatibility among the three analyses.
• About 15% of pseudo-experiments with the Standard Model expectation measured a cross section below 2.1 pb, indicating the observed result is consistent with a 1σ deviation from the SM.
• The combination method successfully reduces uncertainty and enhances sensitivity by leveraging complementary discriminants and robust statistical treatment of correlations and asymmetric errors.