Algorithm Engineering (Dagstuhl Seminar 13391)

Abstract

This report documents the program and the outcomes of Dagstuhl Seminar 13391 "Algorithm Engineering". The algorithm engineering approach consists of a cycle of algorithm design, analysis, implementation, and experimental evaluation, with the aim of bridging the gap between theory and practice in the area of algorithms. This cycle of phases is driven by falsifiable hypotheses validated by experiments. Moreover, real-world instances often have direct impact on this cycle since they often expose modeling and analysis shortcomings. Algorithm engineering touches other research areas such as algorithm theory, combinatorial optimization, computer architecture, parallel and distributed 
computing, high-performance computing, and operations research. Prominent success stories in  algorithm engineering include the linear program solver CPLEX, the traveling salesman suite CONCORDE, speed-up techniques for shortest paths computation, for example, in route planning, as well as graph partitioning and the computation of Steiner trees. All these topics are driven by the need for efficient algorithms and libraries for problems that appear frequently in real-world applications. 

In the last fifteen years, this approach to algorithmic research has gained increasing attention. There is an ACM Journal on Experimental Algorithmics
and several annual conferences (WAE/ESA applied track since 1997, Alenex since 1998, and WEA/SEA since 2001) and the series of DIMACS implementation challenges  where people meet to compare implementations for a specific problem. From 2007 to 2013 the German Research Foundation also funded a special priority program on algorithm engineering (DFG SPP 1307).