Formation and Hydrodynamics of Fish Schools
Announcement date: Oct. 21, 2011
DESCRIPTION: The formation and dynamics of fish schools will be studied by performing hydrodynamically accurate simulations of flows past self-propelled 3D, fish-like swimmers. We will examine the interplay between hydrodynamics, learning and optimization in this complex system We will investigate the capability and mechanisms by which such swimmers exploit vortical structures in a cooperative manner. The combination of the accurate flow simulations and learning/optimization algorithms for fish schools is a novel approach that would help distinguish between hydrodynamic interactions and learned or optimized patterns of behavior. The focus of this work entails two interlinked components:
the simulation of fish schooling hydrodynamics by resolving accurately the flow-structure interactions and hydrodynamic interactions of multiple swimming bodies.
the investigation of fish schooling behavior as determined by various learning scenarios and resulting in energetic efficiency, defense against predation, etc.
These goals require the development of state of the art numerical methods for multiple, complex, deforming bodies immersed in a fluid. In the context of optimization we will develop novel algorithms that combine reinforcement learning and stochastic optimization to help quantifying member and group traits such as selfish or altruistic behaviors as well as leadership in the context of the fish school. We expect that the results of these simulations will provide state of the art understanding for the governing mechanisms of collective behavior of swimmers and will have relevance for realistic engineering and robotic devices.
the simulation of fish schooling hydrodynamics by resolving accurately the flow-structure interactions and hydrodynamic interactions of multiple swimming bodies.
the investigation of fish schooling behavior as determined by various learning scenarios and resulting in energetic efficiency, defense against predation, etc.
These goals require the development of state of the art numerical methods for multiple, complex, deforming bodies immersed in a fluid. In the context of optimization we will develop novel algorithms that combine reinforcement learning and stochastic optimization to help quantifying member and group traits such as selfish or altruistic behaviors as well as leadership in the context of the fish school. We expect that the results of these simulations will provide state of the art understanding for the governing mechanisms of collective behavior of swimmers and will have relevance for realistic engineering and robotic devices.
PREREQUISITES: University degree in any of the following disciplines: Mechanical/Electrical Engineering, Physics, Computational and Applied Mathematics, Computer Science, Applied Mechanics. PhD studies at ETHZ are conducted in English.
DEADLINE FOR RECEIPT OF APPLICATIONS: August 30, 2013
DURATION OF APPOINTMENT: 3 years (+an eventual fourth year).
FELLOWSHIP: Stipend of up to CHF 60,000 per year pending on qualifications.
DURATION OF APPOINTMENT: 3 years (+an eventual fourth year).
FELLOWSHIP: Stipend of up to CHF 60,000 per year pending on qualifications.
APPLYING: Please e-mail to petros@ethz.ch including:
- Curriculum Vitae (including contact information of two references)
- Grades of all University Classes
- A one page statement of your background and research interests
- GRE and TOEFL Scores (if available)
CONTACT: Prof. Petros Koumoutsakos
Chair of Computational Science
Universitätstrasse 6, ETH Zurich
CH-8092, Switzerland
www.cse-lab.ethz.ch
Chair of Computational Science
Universitätstrasse 6, ETH Zurich
CH-8092, Switzerland
www.cse-lab.ethz.ch