\title{Parton fragmentation studies in ATLAS} \author{Jiri Dolejsi for ATLAS collaboration} \address{Charles University, Faculty of Mathematics and Physics,\\ Institute of Particle and Nuclear Physics, V Holesovickach 2, 180 00 Prague 8, Czech Republic\\ jiri.dolejsi@mff.cuni.cz} ATLAS installation and commissioning proceeds towards ability to record first $pp$ collisions during the year 2008. The ATLAS heavy ion working group has examined the detector performance in conditions of $PbPb$ collision at 5,5 TeV per nucleon, based on full simulations of the detector response, and continues in detailed studies. The focus of this talk is on the jet studies. A dedicated jet finder package has been developed and integrated into the standard ATLAS simulation framework. It contains the background subtraction, where the background is estimated over the region outside jet candidate, and iteratively performs the jet cone algorithm using the calibrated energy depositions in individual cells of the ATLAS calorimeters. This package has been tuned on Pythia jets embedded into HIJING $PbPb$ collision with a hard-cut of high $p_T$ processes. The simulations show an efficiency of jet reconstruction by the cone algorithm steeply rising from about 50\% at $E_T$ = 60 GeV up to 90\% at $E_T$ larger than 90 GeV. The fake fraction is about 0.04 at $E_T$ = 60 GeV and decreases with higher $E_T$. These results were obtained for events with moderately high d$N$/d$\eta$ = 2700. The axis of the jet can be measured with the angular resolution of about 0.05 (both in $\phi$ and $\eta$) for jets with $E_T$ of 60 GeV and better for higher $E_T$ in these events. The energy resolution is about 30\% at 60 GeV and is flat over accessible $\eta$ range . The precise position resolution enables a reliable extraction of the jet shapes. The distributions of $j_T$ (particle momentum perpendicular to the jet axis) as well as jet fragmentation functions $D(z)$ have also been studied using the tracks reconstructed in the ATLAS Inner Detector. The tracking performance was found to be sufficiently good for these distrubutions. The agreement between the true and reconstructed distributions is good, providing that the tracks from the underlying event are subtracted. The Fast $K_T$ algorithm is studied as one of the possible alternatives. Its efficiency and energy resolution is comparable to that of the cone algorithm and the more detailed studies continue. References: Heavy Ion Physics with the ATLAS Detector (LoI), ATLAS Collaboration, CERN/LHCC/2004-009, LHCC I-01322 March 2004 ATLAS Physics Prospects, N. Grau, plenary talk at Quark Matter 2008 (http://www.veccal.ernet.in/~pmd/qm2008/webpage/Program/9Feb/Grau.pdf)