Datenschutzerklärung|Data Privacy
Impressum

18.12.2019
L. Friedel

Mo 06.01.2020 16.00 Uhr, TU Berlin Einsteinufer 17, Raum EN719, 10587 Berlin: "Thrill: Distributed Algorithmic Framework in C++ and COBS: A Compact Bit-Sliced Signature Index" (Timo Bingmann, Karlsruhe Institute of Technology)

Abstract:
In this talk I present two recent projects: Thrill and COBS.

Thrill is a prototype of a general purpose distributed Big Data processing framework in C++.
It is similar to Apache Flink and Spark in the sense that it abstracts data and algorithms into building blocks.
But the internal execution is completely different and thus an experimental field for pushing distributed processing of scalable algorithms to exascale computers.
The future challenges therein are manifold and contain distributed dynamic load balancing, fault tolerance, and a scalable hierarchical data exchange.

COBS is an excursion of my previous text indexing research into practical bioinformatics.
In cooperation with the EBI in Cambridge (UK) I developed a new $k$-mer search index for approximate matching in large DNA databases.
This data structure is a crossover between Bloom filters and inverted indices and by allowing false positives in the results we are able to reduced the index size such that it becomes feasible for large datasets.

Short Bio:
Timo Bingmann is currently a PostDoc in Prof. Peter Sanders' Group at the Institute of Theoretical Informatics, Algorithmics, which is part of the Karlsruhe Institute of Technology. He received a Dr. rer. nat. from KIT in 2018 for his dissertation on "Scalable String and Suffix Sorting". Previously, he received a Diploma degree in Computer Science (Dipl. Inform.) from the same university in 2010, and a second Diploma degree in Mathematics (Dipl. Math.) from the FernUniversity in Hagen in 2016.