Abstract

In the last few decades, successful new methodologies based on path integrals have emerged for including nuclear quantum effects in simulations of condensed-phase molecular dynamics. These include centroid molecular dynamics (CMD) and similar methods that combine exact, quantum Boltzmann statistics with classical dynamics. Recently, such methods were placed on a firm theoretical footing with the conception of Matsubara dynamics, although this is not itself a practical method since it suffers a severe sign problem. The fluctuation terms that arise in the Matsubara dynamics framework are approximated only crudely by CMD and other approaches, in order to circumvent the sign problem. Using a simplified model of an OH molecule, we assess the regimes in which it is necessary to take account of these fluctuations more explicitly for the accurate prediction of infrared spectra.