High-performance computing racks with weather forecast visualization displays

Virtual NAIRR hands-on training

Virtual Hands-On Training on Anvil, Containers, and AI Workflows

Connect AI forecast workflows, WRF weather modeling, containers, and MPI-aware HPC practice using Purdue's Anvil system as the instructional platform.

Purdue University Indiana University Mississippi State University
Workshop collaboration CI-PIVOT professionals from three institutions
Purdue
Indiana University
Mississippi State
Date Wednesday, July 15, 2026
Time 1:00-5:00 PM Eastern Time
Location Online via Zoom
Registration Status Capacity reached

Workshop overview

AI-forward HPC workflows for weather and scientific modeling

You are invited to participate in a virtual, hands-on NAIRR training workshop focused on AI-forward HPC workflows that connect classical physics-based models with emerging AI-driven weather modeling workflows.

The workshop will combine cluster onboarding, MPI-aware HPC usage, comparative Weather Research and Forecasting (WRF) and Pangu-Weather demonstrations, and portable container and Conda-based AI workflow design.

Blending short lectures, guided demonstrations, and hands-on activities on Anvil, this four-hour session is intended for graduate students, postdoctoral researchers, faculty, research software engineers, and research support personnel.

While the examples used will center on weather modeling, the workflow lessons are designed to generalize across a broader range of computational science domains.

Workshop format

A practical path from cluster access to model comparison

Three CI-PIVOT Cyberinfrastructure Professionals from Purdue, Indiana University, and Mississippi State University will lead the program. About half of the workshop is devoted to a structured WRF and Pangu-Weather demonstration.

Anvil onboarding

Open OnDemand, filesystems, Slurm, modules, and a light introduction to MPI as the model for distributed-memory execution on clusters.

Containers and AI sidecars

Apptainer recipe and image logic, build-move-run workflows across systems, the host-versus-container boundary, and Conda for GPU-enabled AI tooling.

Hands-on workflow

Prepared WRF and Pangu-Weather scripts, staged input data, selected execution steps, output processing, visualization, and model comparison.

Hands-on component

Hands-on Anvil activities with coordinated workshop access

Hands-on participation requires an ACCESS ID. Organizers are using the workshop allocation to coordinate access for registered participants in advance.

Hands-on Anvil access is being coordinated for registered participants who provided an ACCESS ID.

Inspect relevant modules, filesystems, and selected scheduler workflow steps.

Observe Apptainer build and run concepts and the transition to AI sidecar tooling.

Work through a prepared WRF and Pangu-Weather case using provided scripts and staged outputs.

Review and compare representative outputs instead of rebuilding every large dependency during the session.

Proposed agenda

Wednesday, July 15, 2026, 1:00-5:00 PM Eastern Time

Agenda items marked Hands-on are designed for participants with Anvil access. Participants without access can still observe the demonstrations and join the discussion.

Time Presenter Topic
1:00-1:15 NAIRR Welcome and Anvil access & allocation check
1:15-2:00 Bithi De, Purdue University Hands-on Anvil as the host HPC environment: access pathways, filesystems, Slurm, modules, and an MPI introduction
2:00-2:10 Break Short break
2:10-3:05 Cong Gian, Indiana University Demo Containerization and AI sidecar workflow: Apptainer recipe and image logic, portability across clusters, and Conda/GPU workflow design
3:05-3:15 Break Short break
3:15-4:00 Adonte Knight, Mississippi State University Demo WRF and Pangu, Part I: Pangu overview, Conda environment setup, prepared data/model layout, inference workflow, and output processing
4:00-4:10 Break Short break
4:10-4:40 Adonte Knight, Mississippi State University Hands-on WRF and Pangu, Part II: WRF-ARW/WPS workflow, selected execution steps, visualization, and AI-versus-dynamic-model comparison
4:40-4:55 All Q&A
4:55-5:00 All Closing remarks

Participant preparation

Complete account setup before the workshop

Registered participants who want to complete the hands-on Anvil activities should complete account setup before the workshop whenever possible.

ACCESS and Anvil

Create an ACCESS ID and enroll in ACCESS Duo multi-factor authentication. Organizers will use submitted ACCESS IDs to coordinate Anvil access through the workshop allocation where possible.

Open OnDemand

Test login to Anvil in advance. Open OnDemand is recommended because it runs in a web browser and reduces local software requirements.

Local readiness

A stable internet connection, a modern web browser, and basic comfort with terminal copy-and-paste are strongly recommended.

Command cheat sheet

A lightweight command cheat sheet is available in the workshop materials repository. It covers Linux basics such as changing directories, listing files, checking the current path, creating directories, copying files, and copying/pasting commands in a terminal. Participants do not need advanced Linux or prior HPC experience to attend.

Registration

Registration has reached capacity

We apologize for the inconvenience. Registration was originally advertised to close on July 6, 2026, but the workshop has reached capacity and we are no longer accepting new registrations. Thank you for your interest in the training.

View Capacity Notice