Dr. Oliver Pajonk

Institute of Scientific Computing
Hans-Sommer-Strasse 65
38092 Braunschweig

Room:
104b (1st floor)


Phone:
+49 (0) 531 391 - 3014


Fax:
+49 (0) 531 391 - 3003


E-Mail:
o.pajonk(at)tu-bs.de

Elektrobit Automotive GmbH

Hannoversche Straße 60
38116 Braunschweig, Germany

• Research and development of numerical methods for uncertainty quantification, high-dimensional inverse problems and optimisation in the geosciences
• Main application areas are currently history matching and optimisation for hydrocarbon reservoir applications of the E&P industry
• Parallel software and specific computer architectures
• Component-based software architecture
• Connected car applications

• Solution of dynamic inverse problems by sequential data assimilation, e.g. using sequential Monte Carlo methods like the ensemble Kalman filter or other particle filters, but especially of stochastic spectral methods / functional analytic methods such as the (generalized) Polynomial Chaos Expansion.
• There, I am mainly focussed on methods for improving efficiency like data compression and reduction techniques, parallelisation and efficient implementation - but code maintainability and efficiency in software development are also a topic.
• My main application is the history matching problem in hydrocarbon reservoir engineering. There, numerical models of such a reservoir are conditioned to production data, seismic data and other measurements such that they can reproduce the measurement history and predictions (together with uncertainty quantifications) can be made. It is a high dimensional, non-linear problem where sometimes millions of unknows have to be estimated (of which many are more or less correlated, but not in a simple way).

• The PLATON Project

• See my diploma thesis below.

• 2008-07-10: Update: There were some small notation errors that I removed.
• 2007-09-07: The bibliography of the PDF contained errors. Please download it again.