Extraction of Fragmentation Functions from Charged Kaon and Pion Production at COMPASS

Quark helicity distributions can be accessed by measuring spin asymmetries in polarised deep-inelastic scattering, but for a full flavour separation the precise knowledge of quark fragmentation functions is essential. Those can only be inf erred from experimental data, and are still poorly determined today. The few existing para metrisations of fragmentation functions are derived from world data (mainly on electron-positron annihilation), and often differ considerably, most notably for strange quarks. This thesis presents an independent evaluation of fragmentation functions from deep- inelastic scattering data recorded at the COMPASS experiment. A method of extraction was developed, based on the relation between hadron multiplicities, r h ( x, z ), unpolarised parton distribution functions, q ( x ), and quark fragmentation functions into hadrons, D h d ( z ). (In this work x stands for the Bjorken scaling variable, and z for the fraction of the quark momentum that is transferred to the pro duced hadron.) Multiplicities for charged kaons and pions from 2004 data extracted by t he COMPASS collaboration and two sets of unpolarised parton distribution functi ons from global parametrisations (CTEQ6 and MSTW) were used as input for a system of equations, and five different fragmentation functions ( D fav , D unf for pions and D str , D fav , D unf for kaons) were extracted in four z bins. To evaluate the stability of the results a number of tes ts was performed, among them limiting the x range of the multiplicities and unpolarised parton distribu- tion functions used for the analysis, and comparing them to results derived with a set of artificial multiplicities generated with the particle physics event generator PYTHIA. Four of the five extracted fragmentation functions - D fav and D unf for both pions and kaons - are quite stable with regard to the parton distribution used as well as to the x range in which the analysis was performed, and agree with an earlier evaluation by the EMC collaboration. The fifth fragmentation function, D str for charged kaons, shows in- compatible results for the different x intervals of the multiplicities used in the analysis, and also differs considerably for the two parametrisations of unpolarised parton distribu- tion functions. The extracted values for D str are often negative - and therefore physically impossible, and do not agree with predictions from parametrisations.