Detection of Clustering Instabilities for Sequential Recombination Algorithms

We explore clustering stability of sequential recombination jet reconstruction algorithms. Events are reconstructed many times, using random variations of kinematic properties of the jet fragmentation process. Sensitivity of different algorithms to initial conditions are quantified by introducing probabilistic assignment of initial particles to jets (fuzzy clustering). A criterion detecting unstable configurations (bifurcation points) is proposed, based on the overall fuzziness of the event.