Collaborative Research: Petascale Simulation of Viral Infection Propagation Through Air Travel

Ashok Srinivasan, Florida State University

Usage Details

Ashok Srinivasan, Robert Pahle, Matthew Scotch, Sirish Namilae, Sudheer C D, Mehran Sadeghi Lahijani, Anuj Mubayi, Audrey Gbaguidi, Peirrot Derjany, Meysam Ghaffari

Air travel has been identified as a leading factor in the spread of infections, such as influenza, and this has motivated calls for limitations on air travel in order to prevent the spread of disease outbreak. However, such limitations carry considerable economic and human costs. Consequently, it is necessary to evaluate the extent of impact of air travel on spread of viral diseases and to also identify policy options that can mitigate its spread without major disruption to air travel. Computer simulations play a crucial role in evaluating viable policy options and exploring potential consequences of decisions taken by policy makers. The goal of Project VIPRA is to use the Blue Waters leadership computing system to create a massively parallel simulation infrastructure that will provide useful insight to decision makers dealing with viral infection propagation through air travel. This infrastructure will enable formulation of new policies that will reduce the likelihood of a pandemic, contributing to NSF's mission to advance the health, prosperity and welfare of our nation.

Project VIPRA uses a fine-scale model for human movement, which can capture detailed dynamics, and so can evaluate the impact of subtle changes in procedures that have the potential to bring a substantial reduction in the likelihood of an epidemic spreading through air travel. This fine-scale approach comes with the bottleneck of high computational cost. In preliminary work on Blue Waters, the time required for the primary component of the computation, related to spread of Ebola Virus Disease in an airplane, has been reduced from several hours to around 20 minutes per run on approximately 70,000 cores. These optimizations make it easy to analyze anticipated policies ahead of an emergency. However, in the course of a decision meeting, results for potential new policies need to be produced within a couple of minutes. In addition, air-borne viral diseases, in the more complex situations seen in airports, will require even greater computational effort. This can be handled using GPUs on Blue Waters and by further scaling the computation to greater parallelism. This Blue Waters allocation will enable deployment of an optimized decision support system that can analyze policy or procedural options in the time constraints of a new emergency.