2019 Fiscal Year Annual Research Report
Development of an HPC enhanced system for finding fine-grained lifeline recovery plans and evaluating their long term economic performance
| Project/Area Number |
18H01675
|
|---|
| Research Institution | The University of Tokyo |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2018-04-01 – 2022-03-31
|
| Keywords | Agent based simulations / HPC / natural disasters / economy |
|---|
| Outline of Annual Research Achievements |
We made a major achievement by extending the HPC extension to simulate more than 300 million agents using low computational resources. The largest simulations reported in literature is only with 10 million agents. We developed efficient algorithms, making it possible to simulate a one period of Euro-zone with 331 million agents in 152 seconds, which is an unprecedented performance. Further, we modify the code to simulate economies following a natural disaster and conducted extensive test to ensure the implemented complex model produces accurate results. We integrated the developed ABEM with physics based earthquake disaster simulator, simulated a virtual earthquake disaster in Osaka area, and simulated the economic progress following the virtual disaster.
|
| Current Status of Research Progress |
Current Status of Research Progress
1: Research has progressed more than it was originally planned.
Reason
For FY2019, three major tasks were planned: 1) modify the developed ABEM to simulate economies hit by a natural disaster; 2) enhance the HPC extension to simulate large economies; 3) and integrate with physics based simulators. All these three tasks were successfully complete. Further, the enhancement of the HPC extension produced unprecedented level of computational performance, exceeding our original expectation. The high level of computational performance made it possible to conduct large scale ensemble simulations of large economies. Especially, the high computational performance enable us to conduce very large scale ensemble simulations to include not only the uncertainties related to economic modeling, but also the uncertainties involved in disaster simulations.
|
| Strategy for Future Research Activity |
The COVID-19 pandemic shows that non-physical disasters like pandemic also can severely impact the economy. If the developed system is extended to simulate the conditions due to pandemic, it will be a useful for predicting the economic progress and evaluate recovery strategies. Considering recent calls for COVID-19 related researches, a slight change in research plan is made. Specifically, the developed system is to be tuned to simulate 1:1 scale model of Japanese economy by analyzing the micro-level big-data from Teikoku DataBank Ltd., and validate the model by simulating quarterly economic progress. Further, large-scale simulation of the national economy after a major earthquake in Osaka region will be simulated.
|
