ABOUT EVENT

Workshop Description

This workshop will cover how deep generative models can be used to learn and generate complex geometric structures, with a focus on subsurface applications such as fracture networks and heterogeneous geological fields. The workshop will introduce the motivation for using generative models in scientific computing, explain how complex geometry can be represented by deep learning models, and demonstrate how the Koa HPC system can support model training, data processing, and large-scale numerical experiments. Attendees will learn the basic workflow of preparing geometry data, training or running a generative model, and interpreting generated results. They will also see how these tasks can be connected to the Koa HPC system.

Prerequisites

Basic Python knowledge is recommended. Familiarity with machine learning is helpful. Example scripts or notebooks may be provided before the workshop. No required readings are needed. Optional references: Koa HPC information: https://datascience.hawaii.edu/hpc/

Learning Objectives

Participants will learn how deep generative models can represent and generate complex subsurface geometries, and how the Koa HPC system can support data processing, model training, and scientific computing workflows.

Tools Used

• Koa HPC System
• Python
• Jupyter notebooks
• PyTorch
• NumPy
• Matplotlib
• GitHub

Registration: https://ci.its.hawaii.edu/portal/workshops/hidsi-workshop-series/deep-gen-models/