HIRING OF A DOCTORATE UNDER THE LEGAL REGIME OF SCIENTIFIC EMPLOYMENT (Under reference CTTI-37/20-CFUM(1))
Topics: new efficient numerical tools. We seek a post-doctoral resource to develop new numerical methods that tackle the following issues:
Curved boundary domain is responsible for dramatic high-order approximation accuracy reduction. A new approach to compute approximations for 2D curved boundary with a very simple mechanism is proposed leading to a smaller computational effort, saving power and time consumption. Application to fluid dynamic problem (compressible Navier-Stokes equations) is a crucial issue.
High-order schemes (even implicit) for non-stationary problems suffer from a major drawback since the time step is controlled with the smallest space parameter to ensure the stability leading to important computational cost. We propose the design of a time-space discretization using the Finite Difference/Finite Volume g framework to provide an unconditionally stable scheme that enables large time step and efficiently reduces the computational consumption.
We want to take advantage of all the hardware resources to increase the computational efficiency. Recent computers have developed strategies to increase the efficiency, and we design numerical schemes fully adapted to such specification by revisiting the ADI methods. The algorithm gives rise to a decomposition of the problem for many-core systems and turns the method to be very efficient.