October 16, 1996 Gerry Lynch Description of the DIRC simulation in Aslund. The calculation of the track separation power of the DIRC that is done in Aslund is done in a subroutine called Dirc. This subroutine is called by Aslund whenever Aslund finds that a track hits the cylinder that approximates the Dirc position in the Aslund geometry. Dirc behaves in Aslund that same way that other PID routines behave. Aslund passes to Dirc the track properties and Dirc gives back the number of photons and the number of standard deviations that each mass hypotheses is from its expected value. Actually Dirc can be used in two modes, a "nominal" mode and a Monte Carlo mode, which is the one used in Aslund. The Dirc routine that is now available is essentially the same one that was used for the TDR. It has not been modified to include the changes to the DIRC design that have been made since then. It still has the mirror at the bottom of the stand-off-box, a 60 degree mirror at the top of the stand off box, and does not have the quartz wedge at the end of the quartz bar. It does try to include the effects of the light catchers on the phototubes. The present DIRC design is expected to yield results that are close to the ones that are simulated here. The new design will have more photons, especially at small dip angles, due to the wedge. The way that Dirc proceeds is to calculate from the input track the expected number of photons produced, the fraction of these that get internally reflected at the sides and not internally reflected at the end, and the absorption in the quartz and at the glue joints, to get the expected number of detected photons. Then it combines errors from many sources to calculate the error on ThetaC. Random numbers are used to get the observed number of photons and the measured ThetaC, and standard deviations are calculated from these values. The calling sequence for Dirc is: subroutine dirc(mode,xyz,pxyz,mass,dphi,ds,enpho,sigma,enstd) c input quantities integer*4 mode ! is zero for nominal values ! and non-zero for monte carlo values real*4 pxyz(3) ! three-momentum of the particle at the dirc real*4 xyz(3) ! position of the particle at the dirc real*4 mass ! mass of the particle real*4 dphi ! error in the track azimuth real*4 ds ! error in the tangent of the dip angle c when dphi or ds is zero, nominal errors are used. c output quantities real*4 enpho ! number of photons detected ! when mode is non-zero, enpho is an integer. ! when enpho=0, other output quantities are ! undefined. real*4 sigma ! accuracy of measurement of theta_c real*4 enstd(5) ! number of standard deviations from e,mu,pi,k,p The code for using ds and dphi has never been used nor adequately checked. Aslund has never calculated these quantities. Therefore, Dirc has always used nominal track errors. Many of the properties of the DIRC are hard-wired into Dirc. A few are put into the PIDpostTDR.input file and can easily be changed by the user. The quantities that can be set in PIDpostTDR are: DETECTOR DIR ! Np Ty Index dI/I Dist Ang Nzero Abs30 Side #DIRC 144 1.70 1.474 0.0058 120.0 0.0238 137.0 0.11 -1 ! mat r thick sig zf zb stereo LAYER Si 84.5000 .0000 .1000E+04 175.58 -163.00 .00 Quartz pipe The LAYER line has the radius and end points in z of the sensitive region of the DIRC. The DIRC line has Np = 144 is the number of quartz bars placed on a cylindrical radius. Ty = 1.70 cm is the bar thickness Index = 1.474 is the average index of refraction of the Quartz. DI/I = 0.0058 is a measure of the dispersion of the Quartz. It is the rms of the distribution of indices for the detected photons. Dist = 120 is the distance from the end of the bar to the detector plane. ANG = 0.0238 is the effective angle subtended by collection area of a phototube. Abs30 = 0.11 is the absorption per meter at a 30 degree dip in the first DIRC prototype geometry. Side = -1 says that the readout is in the backward direction. Some quantities that are hard-wired are a 4% loss in geometrical coverage due to gaps in azimuth, a 90% reflection by the mirrors, a 150cm distance between the end of the sensitive region of the bar and the stand- off-box, a 1.33 index of refraction of the water, and a loss loss of 0.8% per meter due to glue joints. Although the method in Dirc is an attempt to get the resolution right, it makes many approximations and cannot be expected to be accurate enough. Therefore, a more detailed stand-alone Monte Carlo (called MCDIRC5) was run to tune the parameters that are in Dirc. MCDIRC5 also approximates the the quartz bars as being on a cylinder. It assumes that the mirror at the bottom of the stand-off box is one mirror that extends to the detector surface, and the same for the 60 degree mirror. The phototubes are nearly close packed on a toroidal detector plane. Individual photons are propagated to the detector plane. In the analysis of these data, all of the ambiguous interpretations were considered, and for each mass hypothesis the interpretation that was closest to the expected position was chosen. From this selection a chi-squared was make for both the pion and kaon hypotheses. From the results of MCDIRC5 three polynomial corrections were added to Dirc that made corrections to quantities calculated there. One polynomial provides a dip angle correction to the absorption. For most angles the correction is less than one, that is, the absorption is less than the input quantity. Another polynomial is one that corrects the number of photons by dip angle. A third polynomial corrects the resolution as a function of momentum. Other more ugly corrections are made in addition to these. The Dirc routine simulates the k-pi separation in the TDR design for isolated tracks above one GeV/c quite well. For low momentum tracks, especially tracks that are near the Cerenkov threshold, the results are less reliable.