Using HPC infrastructures for deep learning applications in fusion research

In the fusion community, use of high performance computing (HPC) has been mostly dominated by heavy-duty plasma simulations, such as those based on particle-in-cell and gyrokinetic codes. However, there a growing interest in applying machine learning for knowledge discovery top large amounts experimental data collected from devices. particular, deep models are especially hungry accelerated hardware, graphics processing units (GPUs), it is becoming more common to find competing same resources that used simulation codes, which can be either CPU- or GPU-bound. this paper, we give examples models—such convolutional neural networks, recurrent variational autoencoders—hat variety tasks, including image processing, disruption prediction, anomaly detection diagnostics data. context, discuss how go using single GPU node multiple GPUs across nodes large-scale HPC infrastructure.