- 175 - Real-Time User Interaction within Socio-Physical Systems Corina Bianca Alb Technical University of Cluj-Napoca Str. Memorandumului, nr.28, Cluj-Napoca corina7bianca@gmail.com Dorian Gorgan Technical University of Cluj-Napoca Str. Memorandumului, nr.28, Cluj-Napoca dorian.gorgan@cs.utcluj.ro ABSTRACT In this sample paper you can find information regarding the study of human interaction with a computer application that manipulates massive amount of information in real-time. Here are discussed the matters of assuring high performance and responsiveness of the system when interacting with a human user. It is also presents a way to establish efficient communication between entities that form crowds and the user input: how to apply external forces to dynamic entities. Author Keywords Entity; GPU; interaction; atomic; obstacle; target; parallel computation; User input; manipulation; cluster; ACM Classification Keywords HCI General Terms Human-Computer Interaction; Socio-physics; Clustering; Entity; User input; GPU; Parallel and distributed computations; Interface responsiveness; INTRODUCTION Human-Computer Interaction is a wide area of study that brings together multiple disciplines, like: computer science, behavioral science, design and media studies. Those are just a few fields of interest that are also tangent to the subject of this paper. The need for a fast means of computation in case of massive amounts of data is a matter of computer science innovation. GPU uses parallel and distributed methods in order to increase the computational performance. This paper shows examples of calculus based on those GPU methods of parallelism. Another aspect is represented by the behavioral sciences. This is a field that requires the analysis of huge quantities of data. This is where GPU comes in handy. The analysis of information is done real-time and the responsiveness of the application assures a good user experience. Socio-physics is a branch of behavioral sciences that also uses mathematical and physical tools in order to understand the behavior of human crowds. Socio-physical systems are representations of complex, real life interaction processes between diverse dynamic entities The interactions can be abstracted into simple mathematical operations. The individuals/entities are characterized by a series of forces and relations, out of which we differentiate in this study: (a) attraction towards a target or another individual and (b) rejection between an individual and another individual or obstacle. The study of those processes requires a means of simulation and real time visualization of scenarios; the interaction between entities must also allow to be manipulated by a user. This paper has as purpose the development and analysis of some flexible and scalable solutions for real-time execution of socio-physical scenarios using graphical processing units (the GPU provides a good environment for real-time manipulation of massive quantities of data by using parallel computation mechanisms). In order to achieve our purpose, we simulate generalized functional patterns applied on entities. Those are described as attraction forces, rejection forces, movement towards an end point (objective point/target) and the concept of obstacle and. Regarding crowd simulation, multiple entities can be grouped in so-called clusters. Entities in the same cluster are characterized by similar behaviors in response to external events; see the figure 1 below for a graphical representation. Each group is defined by one or more atomic actions, just as the ones presented earlier (rejection, attraction etc.). An entity can be part of multiple clusters or none, thus the increasing diversity: there are multiple possible combinations of atomic functions/actions resulting in the creation of complex behaviors. Figure 1. Entities grouped in clusters defining similar behaviors.