1. Lars Arge 1/12 Lars Arge

2. 2/12  Pervasive use of computers and sensors  Increased ability to acquire/store/process data → Massive data collected everywhere  Society increasingly “data driven” → Access/process data anywhere any time  Increased data availability: “nano-technology-like” opportunity Nature issue 2/06: “2020 – Future of computing”  Trend continues as e.g. sensors increasingly pervasive → Exponential growth of scientific data  Paradigm shift: Science will be about mining data → Computer science paramount in all sciences Massive Data

3. Lars Arge 3/12 Algorithm Inadequacy  Importance of scalability/efficiency → Algorithmics core computer science area  Traditional algorithmics: Transform input to output using simple machine model  Inadequate with e.g.  Massive data  Small/diverse devices  Continually arriving data → Software inadequacies!  Communities addressing inadequacies have emerged  But much work still needs to be done

4. Lars Arge 4/12  Established 2007  Initially funded for 5 years by the national research foundation  High level objectives:  Advance algorithmic knowledge in “massive data” processing area  Train researchers in world-leading international environment  Be catalyst for multidisciplinary/industry collaboration  Building on:  Strong international team  Establish vibrant international environment (focus on people)  Focus areas:  I/O-efficient, streaming, cache-oblivious algorithms  Algorithm engineering Center for Massive Data Algorithmics

5. Lars Arge 5/12 I/O-Efficient Algorithms  Problems involving massive data on disk  Disk access is 10 6 times slower than main memory access  Large access time amortized by transferring large blocks of data → Important to store/access data to take advantage of blocks  I/O-efficient algorithms:  Move as few disk blocks as possible to solve given problem “The difference in speed between modern CPU and disk technologies is analogous to the difference in speed in sharpening a pencil using a sharpener on one’s desk or by taking an airplane to the other side of the world and using a sharpener on someone else’s desk.” (D. Comer)

6. Lars Arge 6/12 I/O-Efficient Algorithms Matters  Example: Traversing linked list (List ranking)  Array size N = 10 elements  Disk block size B = 2 elements  Main memory size M = 4 elements (2 blocks)  Difference between N and N/B large since block size is large  Example: N = 256 x 10 6, B = 8000, 1ms disk access time  N I/Os take 256 x 10 3 sec = 4266 min = 71 hr  N/B I/Os take 256/8 sec = 32 sec Algorithm 2: N/B=5 I/OsAlgorithm 1: N=10 I/Os 1526 7 34108912 9 8 54763

7. Lars Arge 7/12 Streaming Algorithms  Problems involving truly massive data  Sequential read of disk blocks much faster than random read  In many modern (sensor) applications data arrive continually → ( Massive) problems often have to be solved in one sequential scan  Streaming algorithms:  Use single scan, handle each element fast, using small space

8. Lars Arge 8/12 Cache-Oblivious Algorithms  Problems to be solved on unknown and/or changing devices  Block access important on all levels of memory hierarchy  But memory hierarchies are very diverse  Cache-oblivious algorithms:  Use blocks efficiently on all levels of any hierarchy

9. Lars Arge 9/12  Algorithm engineering:  Design/implementation of practical algorithms  Experimentation  Center motivated by theory inadequacy  Center wants to promote interdisciplinary/industry work → Natural to consider algorithm engineering  Algorithm engineering work can lead to practical breakthroughs  Example: Flow simulation on massive terrain models ~18 billion points at 1 meter (>>1TB) Implementation of I/O-efficient alg. → two weeks to three hours!! Algorithm Engineering

10. Lars Arge 10/12 Center Team  International core team of algorithms researchers  Including top ranked US and European groups  Leading expertise in focus areas  AU: I/O, cache and algorithm engineering  MPI: I/O (graph) and algorithm engineering  MIT: Cache and streaming AU MPI MIT Arge Brodal Mehlhorn Meyer Demaine Indyk

11. Lars Arge 11/12 Center Activities  Visits of core researchers  Exchange of AU, MPI and MIT Post Docs and PhD students  Visiting faculty, Post doc and students from other institutions  Frequent summer schools: Streaming algorithms this week!  Major international events:  25 th Annual Symposium on Computational Geometry in 2009  New conference: Symposium on Algorithms for Massive Datasets  Theme Workshops to foster multidisciplinary/industry collaboration

12. Lars Arge 12/12  Initially funded for 5 years by the National Research Foundation  Collaboration between three leading international groups  Addresses inadequacies of traditional theory  When processing massive data  When using diverse devices  Initial research focus on I/O-efficient, streaming, cache-oblivious  Also focus on algorithm engineering and multidisciplinary work  Focus on people: Establish vibrant international environment www.madalgo.au.dk