Policy gradient algorithms are among the few learning methods successfully applied to demanding real-world problems including those found in the field of robotics. While Likelihood Ratio (LR) methods are typically used to estimate the gradient, they suffer from high variance due to random exploration at each timestep during the rollout. We therefore evaluate several policy gradient methods with...

We provide sufficient conditions for synchronization by noise, i.e. under these conditions we prove that weak random attractors for random dynamical systems consist of single random points. In the case of SDE with additive noise, these conditions are also essentially necessary. In addition, we provide sufficient conditions for the existence of a minimal weak point random attractor consisting of...

Shared disks environment (SDE) refers to a distributed architecture for high performance transaction processing. In SDE, the computing nodes are locally coupled via a high-speed network and share a common database at the disk level. To avoid frequent disk accesses, each node caches database pages in its main memory buuer. However, since a particular page may be simultaneously cached in diierent...

Abstract. We present various versions of the maximum principle for optimal control of forwardbackward stochastic differential equations (SDE) with jumps. Our study is motivated by risk minimization via g-expectations. We first prove a general sufficient maximum principle for optimal control with partial information of a stochastic system consisting of a forward and a backward SDE driven by Lévy...

The roadmap gives an overview on status and future developments in Superconducting Digital Electronics (SDE). Key areas in SDE under focus are applications, circuit simulation and design, circuit fabrication, interfacing and testing, cooling and system aspects, and new devices and materials. Care was taken to establish the vital link between research and development on the one hand and the indu...

We describe a new family of weak p-th order accurate SDE time integration schemes, called the Direct Richardson p-th order accurate (DRp) schemes. The DRp schemes use the idea of Richardson extrapolation on Euler time steps, performed by way of an acceptance-rejection algorithm. Previous applications of Richardson extrapolation to the Euler scheme are applicable only when the objective is to es...

Relatively little is known about the ability of numerical methods for stochastic differential equations (SDEs) to reproduce almost sure and small-moment stability. Here, we focus on these stability properties in the limit as the timestep tends to zero. Our analysis is motivated by an example of an exponentially almost surely stable nonlinear SDE for which the Euler–Maruyama (EM) method fails to...

It is well-known that a stochastic differential equation (SDE) on a Euclidean space driven by a Brownian motion with Lipschitz coefficients generates a stochastic flow of homeomorphisms. When the coefficients are only locally Lipschitz, then a maximal continuous flow still exists but explosion in finite time may occur. If – in addition – the coefficients grow at most linearly, then this flow ha...

Differential equations have long been used to describe the motion of particles. Stochastic differential equations (SDE)s have been employed for situations where randomness is included. This present work is motivated in part by seeking to describe the motion of mammals moving in a constrained region. Interesting questions that arise include: how to write down a pertinent (bivariate) SDE, how to ...

1 Department Environmental Science and Energy Research, Weizmann Institute of Sciences, Rehovot 76100, Israel 2 Max Planck Institute for Chemistry, Biogeochemistry Department, P.O. Box 3060, 55020 Mainz, Germany 3 The Remote Sensing Laboratory, Jacob Blaustein Institute for Desert Research, Ben-Gurion University of the Negev, Sde Boker Campus, 84990, Israel 4 The Desertification and Restoration...