Elzbieta Krepska - Homepage

PhD student, Computer Systems Group

E-mail:
Office: P4.32
Tel: +31 20 598 7765

VU University Amsterdam
Dept. of Computer Science
De Boelelaan 1081
1081 HV Amsterdam
The Netherlands
Route description

I am a Ph.D. student in Computer Science at the VU University Amsterdam. I work with Henri Bal and Thilo Kielmann in the Computer Systems group in the area of high-performance computing. We collaborate with Wan Fokkink in the Theorethical Computer Science group and Jaap Heringa and Anton Feenstra in the Bioinformatics institute.

I received an M.Sc. diploma in Computer Science from VU University Amsterdam and from Warsaw University. At Warsaw University I also obtained an M.Sc. diploma in Mathematics. I had some great time working as a Summer Student for CERN (Geneva, CH), as an intern at NEC (Princeton, NJ, US) and twice as an intern at Microsoft Research (Cambridge, UK).

 

Research | Publications | Teaching | Links

Research

Interdisciplinary Computer Science

It has always deeply fascinated me, how interdisciplinary Computer Science is - indispensable in physics, chemistry, biology. Not only are scientists hungry for computing power and storage but also they need to handle enormous complexity emerging from ever-increasing amounts of data. I am interested in applying high-performance computing and parallel programming techniques, along with formal modelling methods, to gain insight in natural sciences. In my Ph.D. I am working in the field of bioinformatics and cell modelling.

Parallel Computing: Hierarchical Parallem Graph Algorithms (HipG)

I work on a design and implementation of a framework called HipG for high-level distributed processing of large-scale graphs. We target graph algorithms such as Strongly Connected Components, clustering or social network analysis metrices. This project is done with Henri Bal, Thilo Kielmann and Wan Fokkink.
The website, tool and papers: HipG

Executable Biology: Cell Termination (BioCheck)

I collaborate with Jasmin Fisher and Byron Cook from Microsoft Research and Nir Piterman from Imperial College to prove liveness properties of cells. Specifically, we target proving stabilization for the mammalian epidermis model (Schaub, BMC Sys.Bio.'06). The website, tool and papers: BioCheck.

Executable Biology: C.elegans

In the ConCell project, we want to gain insight into a biological developmental process by building and analysing its executable model. We targeted C.elegans vulval development, as a typical and well-studied example of organogenesis. Our in-silico model bases on "parallel" Petri nets and correctly predicts the results of 46 out of 48 known biological in vivo experiments. Using this model we made detailed statistic predictions and contributed to the biological knowledge of the process. Find more in the project's website: ConCell.

Deobjectifier: Low memory-overhead arrays of objects in Java

A common structure in scientific programming are large arrays of small objects on which computation is being performed - for example in a graph application you compute on a big graph represented as a list of nodes and a list of edges. Unfortunately, in Java, such arrays lead to huge memory overhead, slow allocation and unnecessary and slow garbage collection on the array. The Deobjectifier is a library that offers arrays of objects without the memory-overhead, with fast allocation and typically faster access, with little effort from the programmer. Deobjectifier automatically rewrites given array of objects and its accesses into arrays of primitive types. Deobjectifier is open source. Learn more about the project in the Deobjectify website.

Publications

Teaching

For master projects check the wiki. Some availble master projects:

Check out our short-track master program.

Links