Awarded to Peter Deuflhard (Zuse Institute Berlin, Germany).
Professor Peter Deuflhard's contributions to applied mathematics have a breadth, depth and originality that is almost without parallel. His contributions to algorithm-oriented numerical analysis are fundamental and range from highly nonlinear algebraic systems through large-scale ordinary and partial differential equations to Markov chains. Within these fields they cover direct and inverse problems, optimization aspects and optimal control. Characteristic of his work is that he always lays a firm, often innovative, mathematical basis on which he constructs highly efficient algorithms for hard real-life problems in science and technology. His style of research has revolutionized scientific computing, a large number of highly reputed scholars follow his tracks.
The range of application areas in which Peter Deuflhard has contributed is stunning. Among them are (just in recent years):
- Chemical engineering (chemical combustion, hydrogen motors, car engine catalysators, pollution reduction in coal power stations, ...)
- Microwave technology up to nano-optics (numerical treatment of high-frequency Maxwell equations, Schrödinger-type equations, discrete transparent boundary conditions, design of nano-photonic devices, ...)
- Medicine (optimal therapy planning in the cancer treatment hyperthermia, modelling and simulation of human motion for osteotomic surgery, thermoregulation in the human vascular system, computer-assisted facial surgery, 3D image segmentation, ...)
- Biotechnology (molecular conformation dynamics, computational drug design, virtual screening, understanding of prion diseases, ...)
The efficiency of his algorithms typically originates from new mathematical and algorithmic concepts that Peter Deuflhard has both invented and designed. Let me mention a few of them:affine invariant Newton and Gauss-Newton techniques, from small nonlinear algebraic systems (e.g., in multiple shooting or collocation methods for boundary value problems for ODEs) to adaptive multilevel finite-element methods for PDEs; extrapolation methods for ordinary differential equations (order and step-size control for non-stiff, stiff, and differential-algebraic equations, linearly implicit methods for stiff and differential equations); discrete Galerkin methods for countable differential equations (important in polymer chemistry); cascadic multigrid methods; and, most recently, Perron cluster analysis.
Professor Deuflhard collaborates intensively with engineers, physicians, practitioners, and scientists in many different fields. He was quintessential in forming modern scientific computing as a field integrating a wide range of applied mathematicians, computer and other scientists aiming at a fundamental understanding of phenomena and processes by combining mathematics and computing technology.
Subcommittee for the 2007 ICIAM Maxwell Prize
- Mario Primicerio (Florence), chair;
- Michael Berry (Bristol);
- Stephen Davis (North-Western);
- Leah Edelstein-Keshet (UBC);
- Etienne Guyon (Paris);
- Masayasu Mimura (Hiroshima).