Goal: 10% statistic error

•  “Statistics” is not pure stats ( sqrt(N) ) but error from fits with correlated parameters
• “experimental” observation: still error = A*sqrt(N), 1.5<A<2 for any good significant Y peak no matter of species and algorithm

 (see table at the end)

Solving 0.1 N=A*sqrt(N) for A=1.5 .. A=2 gives N=250…400

Scaling from d+Au

R(pA)/R(dA)=X(dA)/X(pA) * Ncoll(dA)/Ncoll(pA) * sigY(pA)/sigY(dA)

•       Assumption: R(dAu) ~ R(pAu)

•       Ncoll(dAu)/Ncoll(pAu) ~ 1.5

(1.5 used previously by STAR. I tried the Phobos Glauber run gives 1.6 (with default parameters, 200 GeV)

•       Glauber inelastic x-sections: X(dA) = 2.11, X(pA) =  1.74

==> sigY(dA)/sigY(pA)=1.8

•       Assuming similar efficiencies, with intLum(dA)  = 28.2 nb^-1 ,

intLum(pA) = intLum(dA) * N(pA)/N(dA) * sigY(dA)/sigY(pA)
 

estimate

-       For the |y|<0.5 bin (66 dAu hits): 200 … 300 nb^-1

-       For the |y|<1 bin: (115 dAu hits): 100 … 175 nb^-1

-       For the -1<y<0.5 bin: (21 dAu hits): 600 … 950 nb^-1

appendix