Multiverse data-flow control

Abstract

In this paper we present a data-flow system which supports comparative analysis of time-dependent data and interactive simulation steering. In such a system data is created on-the-fly and the user can introduce branching events to investigate multiple scenarios. The novel data-flow model automatically generates the required data for user requests without further interaction. In particular, we generalize the notion of capabilities, jobs and settings of a node which steer the data-flow execution. Based on a notion of scope and validity it becomes straightforward for the user to specify settings which vary over time or scenarios. Intuitive navigation concepts let the user select what he wants to see, the decision which data is required throughout the data-flow is then handled automatically. The approach is based on a graph-based model of a topologically sorted data-flow. Based on that we show that the notion of a scope allows to describe the computation of jobs and validity using set intersections and set unions. We demonstrate the approach by discussing two small case studies.