Scalable Data Structures

Motivation

Data is at the core of computing: Computing is about processing, exchanging, and storing data. The organization of data profoundly influences the performance of accessing and manipulating data. By optimizing the way data is stored, performance can be improved by several orders of magnitude when data scales.

Even if the field of data structures is quite mature, new trends and limitations in computer hardware together with the ever-increasing amounts of data that need to be processed raise new questions with respect to efficiency and continuously challenge the existing models of computation. Thermal and electrical power constraints have caused technology to reach the “the power wall” with stagnating single processor performance, meaning that all nontrivial applications need to address scalability with multiple processors, a memory hierarchy and other communication challenges. Scalable data structures are pivotal to this process since they form the backbone of the algorithms driving these applications.

Over the last three decades there has been a successful sequence of Dagstuhl seminars on Data Structures held approximately biennially. This Dagstuhl Seminar shall bring together researchers from several research directions to illuminate solutions to the scalability challenge of data structures. Experts from algorithm theory and algorithm engineering can contribute the latest results from approximate data structures, hashing, streaming data analysis, space-efficient and succinct data structures, sublinear space data structures, fine-grained complexity, computation on noisy data, and fault tolerant computing. On the other hand, experts on machine models, parallel computing, and grand challenge applications will illuminate the context in which these data structures will be used, and associated results and techniques.

Particular emphasis will be on the growing number of processing elements in multicore processors, GPUs, and supercomputers. This Dagstuhl Seminar shall catalyze this direction of research by bringing practitioners on parallel computing together with the data structure community.