[Haskell-cafe] Research Position, Leeds: Parallel functional visualization

[David Duke]

I have an open position in the Visualization group at Leeds for a researcher to work on one of two projects, the second of which is specific to Haskell:


We are seeking to appoint an ambitious researcher within the Visualization and VR group at Leeds to work on one of two projects, (i) time-varying and higher-dimensional graphs, or (ii) high-performance visualization in Haskell. The post is available immediately, and lasts until 31 July 2016.

1. Time-varying and higher-dimensional graphs.  We are developing powerful tools for the topological analysis of (time-varying) volumetric datasets.  These produce large graphs capturing combinatorial structures and events, and visual analysis of these graphs provides significant insight into the data.  Similar structures may be found in graphs resulting from optimisation problems in multi-parameter spaces.  The research will investigate (i) the link between layout aesthetics and specific tasks over time-varying data; (ii) layout techniques that respect these aesthetics, for example embedding of the graph into higher-dimensional manifolds; (iii) techniques for eliding structural detail, and linking visual abstractions to other models of the data; and (iv) efficient implementation of layout algorithms, using established toolkits (e.g. VTK), and parallelisation on cluster-based hardware and/or many-core GPU architecture. You will be working with researchers on the "Multifield Extension of Topological Analysis" project, funded recently by the EPSRC.


2. Parallel functional visualization:  Scientific visualization involves the application of compute-intensive algorithms to large volumes of data, often larger than can fit into core memory.  Systems are still largely based on the "pipeline" model, extended to exploit different processing technologies (multi-core CPUs, GPUs,  clusters), and strategies (streaming, tiling, out-of-core processing, etc).  This research concerns an innovative approach that exploits laziness and parallelism in the pure functional language Haskell. Scivis "applications" will be constructed from expressions in a declarative domain-specific language, and transformed into imagery via lower-level DSLs and strategies that coordinate use of heterogeneous resources. 

Both projects involve fundamental computer science and interdisciplinary work alongside experts from other disciplines providing data and contributing to its analysis. The appointee will be expected to contribute to the programme described above through all stages of the research process, including basic research, programming, testing, evaluation, publication of results at the highest level, and assistance in developing proposals for funding. Work will involve collaboration with other senior members of the Leeds visualization group (Dr. Hamish Carr and Dr. Roy Ruddle), and there will be opportunities to assist with academic supervision of undergraduate and postgraduate students.

CLOSING DATE: 21st September.

For further information including person specification, and/or to submit an application, please see the following URL:

http://jobs.leeds.ac.uk/fe/tpl_universityofleeds01.asp?s=FoQnTYvIgXJoLlXgd&jobid=87341,0258234882&key=83785920&c=566912236914&pagestamp=sewzhxglqcfnsctopm

For informal inquiries, please feel free to contact me directly.

David Duke
Head, School of Computing, and Reader in Visualization
E: D.J.Duke at leeds.ac.uk
W: www.comp.leeds.ac.uk/djd/