ParGraph 2013 - Workshop on Parallel Algorithms and Software for Analysis of Massive Graphs (ParGraph)

Combinatorial algorithms play an important enabling role in several areas of science and engineering. Many combinatorial, machine learning, and graph algorithms in particular, are characterized by irregular data-access patterns that make efficient implementation challenging on both traditional HPC and commodity platforms. Architectural features, algorithmic design and software engineering techniques, and characteristics of the input collectively determine the performance of these algorithms.
In addition, runtime support for irregular applications, programming models, the large-scale graph structure (sparse matrix) deserves attention for efficient storage, graph query and visualization for data analysis. With this workshop we aim to bring together system builders in industry and academia interested in combinatorial (graph), machine learning algorithms, data intensive applications involving them, and parallel computing.
Topics
We seek submissions on broad topics which include, but not limited to:
Theory and Algorithms
Combinatorial (graph) and machine learning algorithms and libraries
Algorithms for analysis of large complex networks (social network analysis)
Approximation algorithms and bounds
Design and performance evaluation of algorithms on
Traditional architectures: massively parallel distributed-memory systems (IBM BlueGene, Cray XE), shared-memory multicore systems (Intel, AMD)
Nontraditional architectures: massively multithreaded architectures (Cray XMT), many core and tiled platforms (Nvidia, AMD, Tilera, Intel)
Applications
Combinatorial Scientific Computing; Data Mining; Graph Mining
Big Data Application ? Social graph analysis, Graphical Model, pattern matching, time-evolving graphs
Bioinformatics; Information (Cyber) security; Mobile computing
Visual Analytics; Graph Visualization
Programming paradigms and runtime support for irregular applications