2023 Fiscal Year Final Research Report
Modeling of mantle dynamics based on neural networks
| Project/Area Number |
21K03714
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Review Section |
Basic Section 17040:Solid earth sciences-related
|
|---|
| Research Institution | The University of Tokyo |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2021-04-01 – 2024-03-31
|
| Keywords | ニューラルネットワーク / マントルダイナミクス / 熱対流 / 熱伝導 |
|---|
| Outline of Final Research Achievements |
This study investigates the applicability of Physics-Informed Neural Networks (PINN), a recently proposed method which is based on neural networks, to the prediction of spatiotemporal variations in temperature and rock velocity inside the Earth. I have tested various formulations and ways to incorporate boundary conditions, and found it difficult to predict the results even for relatively simple problems using PINN at this stage.
|
| Free Research Field |
地球ダイナミクス
|
| Academic Significance and Societal Importance of the Research Achievements |
近年発表された偏微分方程式を解くための新たな手法 Physics-Informed Neural Networks (以後PINNと呼ぶ)は、従来の数値計算手法と比較して、解が不連続になる問題や時間逆行問題に対しても適用が容易であるなどの理由で注目されてきた。しかし本研究では、PINNを用いて地球内部における温度や岩石流動の時空間変化を精度良く予測することはまだ現段階では難しいということを明らかにした。
|
