It is worthwhile to investigate the behavior of the Gaussian n_σ(π) pion peak. If the centroids are close to zero and the widths close to unity, it suggests that the n_σ(K) and n_σ(p) peaks may be used to set upper limits on the K/p contaminations. This may also allow us to reduce the associated contamination systematic uncertainties.

Figure 1: Pion Peaks

Low pT	Mid pT	High pT

Figure 1 shows the results for low, mid, and high-p_T as a function of z. The fits are those with the "fixed" electron centroids. Within the relevant window of z, namely 0.1-0.8, The pion centroids range from -0.2 to within 0.1. The widths range from about 0.9 to 1. This suggests the pion peaks behave as proper normal distributions.

Figure 2: Kaon/Proton Peaks

Low pT	Mid pT	High pT

Figure 2 presents the centroids and widths of the kaon/proton peaks. Again, the peaks generally behave as proper normal distributions, with widths around 1 and centroids around -2. The 0.1-0.2 bin at low-p_T has a bit of a smaller width than one might expect (0.8 vs. 1.0). However, it should be noted that this bin, in particular, has a large kaon/proton contamination that is missed by the dE/dx fits. So, it is perhaps not a surprise if the fit reveals a kaon/proton peak that does not behave as a proper normal distribution. Outside of this bin, in the 0.1-0.8 range relevant for the present measurement, the K/p peaks appear to be normal distributions.

Figure 3: Electron Peaks

Low pT	Mid pT	High pT

Figure 3 presents the electron peak centroids and widths. The electron peaks show a bit more variation than the pion or K/p peaks. Still, within the range of 0.1-0.8, the peak widths are generally within 0.8-1.2. One notable exception is the 0.1-0.2 bin at low-p_T. The electron contamination in this bin is found to be rather small, around 1.4%. The larger concern in this bin is the state of the K/p contamination. So, generally, the electron peaks also behave as proper normal distributions.

Having demonstrated well-behaved dE/dx peaks, it should be possible to use the n_σ(K) and n_σ(p) peaks for setting limits on the contamination above the range where TOF resolution degrades. These distributions were presented on a previous blog entry. The active range is -1 < n_σ(π) < 2.5, representing 0.819 of the pion Gaussian. Kaons and protons should manifest as normal distributions, centered at zero in n_σ(K) and n_σ(p), respectively. The region of the Gaussian n_σ(K) and n_σ(p) covered can be estimated by marking the horizontal position where the data n_σ(K) and n_σ(p) distributions "turn on." Arbitrarily, this is estimated by point where the data distribution rises to 18% of the maximum. For conservative (but not crazy) estimates, I assume the kaon and proton yields to be 18% and 12% of the pion yield, respectively, and ignore contributions from electrons.

Table 1: High-p_T dE/dx Particle Fractions

Table 1 presents the new contamination estimates from the 4-Gaussian fits and n_σ(K/p) distributions. The estimates are generally quite close to the total K/p fraction found by the three-Gaussian fits. The two largest deviations are the 0.1-0.2 bin and the 0.4-0.5 bin. For the 0.1-0.2 bin, the fit estimate is 3.2% K/p while the estimate is 4.2% K/p. For the 0.4-0.5 bin, the fit estimate is 5.7% K/p while the estimate is 7.2% K/p.

Table 2: Mid-p_T dE/dx Particle Fractions

Table 3: Low-p_T dE/dx Particle Fractions