Transport of single-channel spinless interacting fermions (Luttinger liquid) through a barrier has been studied by numerically exact quantum Monte Carlo methods. A novel stochastic integration over the real-time paths allows for direct computation of nonequilibrium conductance and noise properties. We have examined the low-temperature scaling of the conductance in the crossover region between a very weak and an almost insulating barrier.PACS numbers: 72.10. 73.40.Gk Studies of transport in one-dimensional (1D) interacting Fermi systems have gathered novel physical insights and led to the development of several new techniques over the past few years. Instead of the usual Fermi liquid theory, the low-temperature properties of an idealized 1D quantum wire are described by the Luttinger liquid model [1,2]. Remarkable transport behaviors are expected for a Luttinger liquid in the presence of impurities or barriers, and several interesting theoretical [3][4][5][6][7] and experimental [8,9] studies have emerged recently. In this Letter, we describe a real-time quantum Monte Carlo (QMC) method that enables us to directly address the dynamics of a Luttinger liquid, and use it to investigate the low-temperature transport properties of a 1D quantum wire.So far the only numerical approach to this problem has been the one given by Moon et al. [6]. These workers examined transport in a Luttinger liquid using Euclideantime QMC simulations by analytically continuing numerical data to real time with Padé approximants. Mathematically, this analytic continuation is ill-posed and small statistical errors in the imaginary-time data could be magnified immensely [10]. Physically, imaginarytime data collected at low temperatures contain predominantly ground-state information whereas real-time dynamics is controlled by the excitation spectrum. Although the findings of Ref.[6] are certainly reasonable, a direct real-time simulation which avoids the possibly troublesome analytic continuation is desirable. Furthermore, a real-time simulation needs not rely on the Kubo formula and allows for direct computations of nonequilibrium transport and noise properties beyond the reach of previous methods.Common to all real-time QMC simulations is the ubiquitous and fundamental dynamical sign problem [10][11][12][13][14][15]. It arises in the stochastic summation over the system paths when the real-time propagators are oscillatory. The quantum-mechanical interference between different paths leads to a vanishing signal-to-noise ratio at very long real times and the simulation becomes effectively unstable. However, it is generally possible to treat intermediate-to-long times by employing a partial summation scheme [14] which reduces the effective number of variables subject to the Monte Carlo sampling. If guided by physical intuition, such a partial summation scheme can largely circumvent the dynamical sign problem. Previous applications to dissipative tight-binding models in the quantum-chemical context [14,15] have demonstrated the practical usefu...