1. (+) m.apollonio@imperial.ac.ukPoster session, PAC09 Vancouver – TH6PFP056 Introduction The Muon Ionisation Cooling Experiment (MICE, fig. 1c) at RAL[1] will demonstrate ionization cooling in a variety of initial emittances and momenta ( ,p). Protons in the ISIS synchrotron hit a titanium target producing pions which are focussed with an F-D- F quadrupole triplet and steered by means of a dipole towards a 5T solenoid where they decay into muons. Hence a second dipole transports them towards the experimental apparatus by means of two other quadrupole F-D-F triplets. This system constitutes the MICE beam line (fig.1a). The transverse phase space emittance of the initial muon beam depends on the production mechanism and is estimated to be around 1.5 mm rad. In order to change this value to the desired ones (up to 10 mm rad) multiple scattering can be used in a controlled fashion, by Optimisation with a Genetic Algorithm placing a layer of material, like a lead disc of defined thickness, at the proper location (fig. 1b). This element constitutes the diffuser, and the choice of its thickness is ruled by three basic requirements: inflate the initial emittance to some desired value, cover most of the amplitudes accessible by the tracking devices (and not cut off by the actual cooling channel), be flexible enough to work for all the configurations of the ( ,p) matrix. The best position for the diffuser is inside the bore of the first spectrometer solenoid (fig. 1b) which poses several mechanical challenges. Emittance Generation in MICE M.Apollonio +, Imperial College of SciencTechnology and Medicine, London, UK [1] the Scienc and Technology Facility Council Rutherford Appleton Laboratory, Didcot OX11 0QX (UK) fig. 1c MICE experiment fig. 1a MICE beamline Optical Matching Requirements Matching Procedure Our goal is reaching the matching condition inside the spectrometer solenoid:  K=1 (K=qB z /2P z ), which corresponds to a flat  within the region of uniform B z. The  function is propagated from this initial value back to the diffuser and to TOF 1 (downstream Q 9 ). The beamline is then tuned to reach these values. Table 1 show the Twiss parameters on the upstream fac of the diffuser requred for a proper matching [1] Evolution of the Twiss parameters (a,b) from the centre of the spectrometer solenoid to TOF1. Conclusions alpha  0   DK Solenoid Q1 Q2 Q3 Dipole 1Dipole 2 Q4 Q5 Q6Q7 Q8 Q9 (  ) at TOF1 (  ) at diffuser Upstream beamline: magnet currents fixed by the condition on the final momentum and matching inside the DK solenoid. Downstream beamline: triplets Q4-5-6 and Q7-8-9 are tuned to fulfil thematching requirements on (a,b) at some Z-positions (TOF1 and diffuser) After fixing the optics for the upstream part (which is dictated by the final momentum we want to achieve) we need to optimise the downstream section. The code Decay-TURTLE[2] is used to track pions from the target to the DKsolenoid. Muons generated after  -decays are collected and used as a starting file for the tracking inside the downstream section. The covariance matrix of the beam at chosen Z-positions is used to calculate the Twiss parameters subsequently used in the matching algorithm. Automatic Tuning: to speed the tuning procedure up a Genetic Algorithm[3] is used. The six currents of the last two quadrupole triplets form a genotype that produces a specific optics (phenotype). A fitness function is associated to it which describes the degree of goodness of the optics (nearness to goal Twiss parameters). Pairs of individuals are mated exchanging sequences of genes to produce new genotypes. The lower fitness ones are discarded in favour of better individuals. beamline/experiment interface: the MICE diffuser evolution of the quadrupole currents for the downstream beamline: (left) Qi as a function of the optics fitness: high values correspond to a good match of the optics with the required values (right) Qi as a function of the number of generations. After an initial erratic behaviour evolution brings currents to stable values evolution of the Twiss parameters (top)  (left) and  (right) (bottom left) Transmission is monitored (not optimised) through the process (bottom right)  versus  plot with bins weighted by the fitness function for a phenotype. The brighter spot corresponds to the target pair(s)  CASE-1: eN=10 mm rad, P=207 MeV/c, tdiff=15.5 mm Optimisation on (  ) at the upstream face of the diffuser The GA reaches convergence after 10 generations and Transmission is slighlty improved as long as stable values are found. Fitness is high on average indicating a good convergence. CASE-2: eN=10 mm rad, P=207 MeV/c, tdiff=15.5 mm Optimisation on (  ) at three Z-positions: 1)upstream diffuser face 2)downstream TOF1 face 3)upstream TOF1 face Convergence turns out to e more difficult. This is seen in the final Fitness which is much lower than in CASE-1. Generating the correct emittance in MICE requires a proper matching of the optics from the beamline. The use of a GA associated to particle tracking looks promising for our purposes. In the studied cases convergence is reached within 2 hours ½ for an initial population of 70K muons. The procedure is flexible enough to allow multiple constraints on the optics. The natural prosecution of this work is the completion of the (eN,P) matrix required in the MICE program. Verification with high definition tracking codes (e.g. G4Beamline) is envisaged. [2] D.C.Carey, K.L.Brown, Ch.Iselin “Decay TURTLE”, SLAC-246, UC-28 (I/A) [3] P. Charboneau, B. Knapp “A User’s Guide to PIKAIA 1.0”, NCAR/TN-418+IA, December 1995 (http://www.hao.ucar.edu/Public/models/pikaia/pikaia.html) [1] M. Apollonio et al. “The MICE Diffuser System, PAC08, Genoa, Italy, 23-27 Jun 2008, pp WEPP108