Abstract

The aim of this paper is to provide a detailed description of a framework for the estimation of time-sliced origin-destination (OD) trip matrices in a transit network using counts and travel time data of Bluetooth Smartphone devices carried by passengers at equipped transit-stops. A Kalman filtering formulation defined by the authors has been included in the application. The definition of the input for building the space-state model is linked to network scenarios modeled with the transportation planning platform EMME. The transit assignment framework is optimal strategy-based, which determines the subset of paths related to the optimal strategies between all OD pair Document type: Part of book or chapter of book

Abstract

It is widely acknowledged that, to create models for transportation planning that recognize the essential dynamic character of passenger network flows, one must consider two time scales: the so-called within-day time scale and the day-to-day time scale. Substantial progress has been made in modelling within-day dynamic flows for fixed trip matrices; one of the most widely acknowledged models for this purpose is the dynamic user equilibrium model proposed by Friesz et al. (1993) and studied by Xu et al. (1999), Wu et al. (1998), Friesz et al. (2001), Bliemer and Bovy (2003), and Friesz and Mookherjee (2006). In this chapter we propose two day-to-day models of demand growth compatible with a differential variational inequality formulation of the Friesz et al. (1993) model. The first of these employs dynamics inspired by evolutionary game theory, while the second uses the perspective of preferential attachment familiar from the network science and social network literature to create a model of demand growth. Additionally, numerical experiments to compare and contrast the two proposed theories of demand growth are described, along with hypotheses that one might address via such experiments. Document type: Part of book or chapter of book

Abstract

We consider the class of “left-looking” sequential matrix algorithms: consumer-driven algorithms that are characterized by “lazy” propagation of data. Left-looking algorithms are difficult to parallelize using the message-passing or distributed shared memory models because they only possess pipeline parallelism. We show that these algorithms can be directly parallelized using mobile pipelines provided by the Navigational Programming methodology. We present performance data demonstrating the effectiveness of our approach. Document type: Part of book or chapter of book