Core Concepts
Juno mission data used to reconstruct Jupiter's magnetic field with physics-informed neural networks.
Abstract
Key Points:
Two reconstructions of Jupiter’s magnetic field using physics informed neural networks: PINN33 and PINN50.
Clearer images at depth compared to spherical harmonic methods.
Dynamo inferred at a fractional radius of 0.8.
Abstract:
Magnetic sounding from Juno mission data provides insights into Jupiter's interior.
New reconstructions based on physics-informed neural networks offer clearer images of the internal magnetic field.
Introduction:
Juno mission revolutionizes understanding of Jupiter's interior through gravity and magnetic measurements.
New reconstructions highlight local features like the Great Blue Spot.
Data:
Vector magnetic field data from Juno's first 50 perijoves used for reconstructions.
Measurements within planetocentric spherical radius r ≤ 4.0RJ utilized.
Method:
Physics informed neural networks (PINNs) used for spatial representation constrained by data and physical laws.
Models based on first 33 and 50 orbits created, showing stability in downwards continuation.
Results and Discussion:
Comparison of models with Juno data shows similar fit but clearer images at depth with PINNs.
Dynamo boundary estimated at a fractional radius of 0.8RJ.
Concluding Remarks:
Meshless method offers noise reduction in reconstructed field at depth, aiding in understanding secular changes near the dynamo region.
Stats
PINN33とPINN50に基づくモデルは、ジュノのデータと類似した再構築を提供します。
ダイナモ境界は、約0.8RJの分率で推定されます。