Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Computing Granular Dynamics on the GPU
Granular materials are widely discovered in nature and can be modeled for
example with lots of rigid bodies
interacting with each other. This works deals with the parallelization of many multi-body systems within a non-smooth dynamics framework
Granular materials are widely discovered in nature and can be modeled for
example with lots of rigid bodies
interacting with each other. The
interaction between two bodies is given by set-valued force laws such as the
2D Coulomb-friction and unilateral (one-sided) contacts. Integration of these
granular systems is done by a time-stepping method which involves typically
a collision detection to find all interacting bodies and an additional inclusion
solver
which determines the set-valued forces. Simulating such large
granular systems, consisting of several thousand bodies, is performed on a
cluster with a MPI (Message Passing Interface) implementation.
Granular materials are widely discovered in nature and can be modeled for example with lots of rigid bodies interacting with each other. The interaction between two bodies is given by set-valued force laws such as the 2D Coulomb-friction and unilateral (one-sided) contacts. Integration of these granular systems is done by a time-stepping method which involves typically a collision detection to find all interacting bodies and an additional inclusion solver which determines the set-valued forces. Simulating such large granular systems, consisting of several thousand bodies, is performed on a cluster with a MPI (Message Passing Interface) implementation.
The aim of this project is to test and implement different strategies to
solve the inclusion problem on the cluster by a parallel GPU
implementation with CUDA. (Methods are available)
**Theoretical Skills:** Dynamik strukturvarianter Systeme, Multibody
Dynamics, Parallel programming background a plus!
**Prog. Skills:**
Intermediate Level in C++ (STL,Boost), (CUDA C,
MPI), Python or Matlab, Linux
**Language:**
German or English
**Type of Project:**
ca. 30% Theory, 70% Programming
The aim of this project is to test and implement different strategies to solve the inclusion problem on the cluster by a parallel GPU implementation with CUDA. (Methods are available)
MAArbeit3.pdf
