Multi-cultural visualization : how functional programming can enrich visualization (and vice versa)

Abstract

The past two decades have seen visualization flourish as a research field in its own right, with advances on the computational challenges of faster algorithms, new techniques for datasets too large for in-core processing, and advances in understanding the perceptual and cognitive processes recruited by visualization systems, and through this, how to improve the representation of data. However, progress within visualization has sometimes proceeded in parallel with that in other branches of computer science, and there is a danger that when novel solutions ossify into `accepted practice' the field can easily overlook significant advances elsewhere in the community. In this paper we describe recent advances in the design and implementation of pure functional programming languages that, significantly, contain important insights into questions raised by the recent NIH/NSF report on Visualization Challenges. We argue and demonstrate that modern functional languages combine high-level mathematically-based specifications of visualization techniques, concise implementation of algorithms through fine-grained composition, support for writing correct programs through strong type checking, and a different kind of modularity inherent in the abstractive power of these languages. And to cap it off, we have initial evidence that in some cases functional implementations are faster than their imperative counterparts.