Dynamical approach to analytic continuation of quantum Monte Carlo data

A new method of analytic continuation from a Matsubara single-particle Greens function to a spectral function is presented. We recast this problem onto a new one where we dynamically minimize a suitably defined potential. Our method allows the imposition of physical constraints such as smoothness and adherence to the sum rule. The method is applied to the symmetric Anderson impurity model. We show how the spectral function changes with the Kondo temperature TK, the hybridization witdth ", and the Coulomb potential U.