Ongeveer 12 uur geleden - Universiteit Twente (UT) - Enschede
PhD in Mathematical Aspects of Number Representation in Embedded MRI Software
A four-year PhD position is available at Eindhoven University of Technology (TU/e) in Mathematical Aspects of Number Representation in Embedded MRI Software
- de Rondom, Eindhoven, Noord-Brabant
- Tijdelijk contract / Tijdelijke opdracht
- Uren per week:
- 38 uur
FunctieomschrijvingMathematical Aspects of Number Representation in Embedded MRI Software
A four-year PhD position is available at Eindhoven University of Technology (TU/e), starting date: April 1 or May 1, 2019. Research concerns analysis, improvement, and automation of the conversion of floating-point computing software into fixed-point embedded-system software, at prescribed precision.
In this position research and innovation activities will be performed in the context of the three-year ECSEL 2018 project Arrowhead Tools. ECSEL is an industry-driven program for the Electronic Components and Systems in Europe, partly funded by the European Community and partly by national authorities.
Computing processors embedded in Magnetic Resonance Imaging (MRI) systems through field-programmable gate arrays need to be energy efficient and hence computationally efficient. For this purpose, they are coded in Very High-Integrated Circuits Description Language (VHDL), with fixed-point number representation at relatively low precision. The VHDL codes are derived from models coded in high-precision floating-point number representation. The necessary conversion of these floating-point codes into fixed-point VHDL codes is a source of performance degradation of the algorithms involved.
Currently, converting floating-point computing software into fixed-point computing software is an empirical process, which consumes much costly development time and may easily lead to non-optimal implementations. The goal of the planned research is: to analyze, improve and automate the conversion of floating-point computing software into fixed-point embedded-system software, at prescribed precision. Direct contacts exist with dr.ir. Mark van Helvoort, manager of the Philips Public-Private Partnership Program Magnetic Resonance Imaging.
The accuracy of the conversion of the floating-point software into fixed-point software will be assessed through numerical analysis. This accuracy analysis of algorithms is expected to be of both scientific and engineering interest. The outcomes of the analysis will enable improvement of the accuracy and efficiency of scientific computing tools embedded in MRI systems, through techniques such as variable arithmetic algebra and interval arithmetic.
You will be a member of the group Scientific Computing within the Centre for Analysis, Scientific Computing and Applications of the Department of Mathematics and Computer Science at TU/e. Your advisors will be:
Keywords: finite precision arithmetic, variable precision arithmetic, precise computing, efficient computations, interval arithmetic, numerical analysis, numerical algebra, floating-point numbers, fixed-point numbers, automated computing, MRI systems.
Additional informationTo apply:
For informal inquiries please contact dr. Michiel Hochstenbach (m.e.hochstenbach[at]tue.nl). To apply, please use the blue button 'Apply for this job' or visit the following website: https://jobs.tue.nl/en/vacancies.html.
See also www.win.tue.nl/~hochsten/arrowhead.html.
Please note: you can only upload 5 documents of 2 MB each. If necessary please combine documents.
For further information about employment conditions, you may contact Ms. Marjolein von Reth, personnel advisor TU/e, pzwin[at]tue.nl.