This repo contains the GPU and TPU code of parallelized Approximate-Bayesian-Computation(ABC) simulation-based inference for a stochastic epidemiology model for COVID-19. The model and parallelized ABC algorithm are described in the following publication:
"Accelerating Simulation-based Inference with Emerging AI Hardware", S Kulkarni, A Tsyplikhin, MM Krell, and CA Moritz, IEEE International Conference on Rebooting Computing (ICRC), 2020.
GPU code can run on any nvidia GPU. To run the TPU version, please import the ABC_TPU.ipynb file in Google Colaboratory, and select the TPU runtime.
The data is obtained from JHU CSSE COVID-19. It contains the COVID-19 case data (confirmed active cases, confirmed recovered cases, and confirmed deaths) for all countries. In this example we utilize case data of Italy.
There is an accompanying repo in Graphcore Demos in which the same model is implemented in Graphcore MK1 IPUs as part of the comparative analysis performed in the paper.