The Institute of Software Methods for Product Virtualization addresses the digital description of an aircraft with all its properties and components based on highly-accurate models covering all disciplines, the whole development process and the entire lifecycle of an aircraft.
Different computer codes of various fidelities are used in the simulation-based analysis and optimization of an aircraft. Among those are highly-accurate HPC methods such as multi-level parallel CFD or CSM codes as well as coupled processes for multi-disciplinary simulation and optimization. A consistent approach to sensitivity analysis is a key ingredient in gradient-based optimization and Newton-type solution techniques. The emphasis is on the evaluation, development and application of efficient methods for code-differentiation in forward and reverse mode. In addition to disciplinary codes for CFD and CSM, multidisciplinary simulation chains are to be differentiated. The work will be carried out in close collaboration with other DLR institutes providing the computer methods to be differentiated.
Among your challenges and tasks are:
identification and evaluation of appropriate methods for algorithmic differentiation (AD)
development of consistent and efficient code-differentiation techniques for simulation codes and coupled simulation processes in forward and reverse mode
efficient adjoint methods for iterative and unsteady algorithms by means of checkpointing techniques
strategies for the computation of higher derivatives
differentiation of multi-level parallel codes
development/deployment of verification and testing strategies
embedding the sensitivity analysis within existing optimization and solution algorithms
You are going to work in concert with scientists from other disciplines and institutes in order to contribute to the digital aircraft description. Among the software to be differentiated will be a next-generation CFD code that is currently being developed in a joint initiative of Airbus, Onera and DLR, and a coupled simulation scenario for fluid-structure interaction by means of the MDA/MDO HPC framework FlowSimulator.
Master’s degree in mathematics, computer sciences, numerics or engineering
wide expertise and practical skills in the field of algorithmic differentiation (AD) in adjoint mode – ideally in conjunction with iterative methods
wide knowledge in applied computer sciences and mathematics
extended programming skills in C++ and preferably in C, Fortran and Python
knowledge in parallel computing
you are able to work as part of a team and independently
good communication skills in English language
knowledge in gradient-based optimization desired
experience in numerical methods for fluid flow and structure mechanics optional
desireably you are familiar with national and international project work
Look forward to a fulfilling job with an employer who appreciates your commitment and supports your personal and professional development.Our unique infrastructure offers you a working environment in which you have unparalled scope to develop your creative ideas and accomplish your professional objectives.Our human resources policy puts great value on a healthy work-life balance as well as equal opportunities for men and women.Individuals with disabilities will be given preferential consideration in the event their qualifications are equivalent to those of other candidates.
Differentiation of Simulation Codes and Processes for Aircraft Optimization at Deutsches Zentrum für Luft- und Raumfahrt e.V. (D
Stack Overflow · 09.08.2018