Sort by
Refine Your Search
-
efficient GPU-supported implementation. Your tasks: Development of physics-aware ML models for 3D blood-flow prediction Integration of domain decomposition methods into the learning framework to enable
-
-aware learning methods with domain decomposition techniques, enabling parallel training and efficient GPU-supported implementation. Your tasks: Development of physics-aware ML models for 3D blood-flow
-
managing supercomputer resources Strong skills in algorithm development for large sparse matrices Excellency in programming GPU accelerators from all major vendors Very good command of written and spoken
-
of spikes by a model Develop proxy apps representing the different processing stages of spiking network simulation code (targeting CPU and accelerators such as GPU or IPU) Systematic benchmarking of proxy
-
is of advantage: Knowledge of parallel programming and HPC architectures, including accelerators (e.g., GPUs) Experience in modelling and simulation, ideally in the field of energy systems Experience
-
commonly used on Unix systems. Additional languages or experience with libraries for utilizing GPU hardware efficiently, e.g., CUDA, are a plus. Experience in AI programming with, e.g., PyTorch(-DDP
-
GPU-capable, parallelized simulation frameworks. Work closely with experts in HPC and power systems to enhance scalability and computational performance. Disseminate your findings through scientific
-
benchmark them with a realistic case study. The main focus of the project can develop either more in the mathematical theory of MCMC, the implementation of code for the Jülich supercomputers (GPU/CPU
-
-following inverters. Implementing and optimizing scalable algorithms for transient and stability analyses on HPC architectures (CPU, GPU, hybrid). Enhancing the numerical robustness and efficiency of existing
-
. Additional languages or experience with libraries for utilizing GPU hardware efficiently, e.g., CUDA, are a plus. Experience in AI programming with, e.g., PyTorch(-DDP), Horovod, or DeepSpeed, and in