Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
134 tokens/sec
GPT-4o
10 tokens/sec
Gemini 2.5 Pro Pro
47 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Neural Operators Learn the Local Physics of Magnetohydrodynamics (2404.16015v2)

Published 24 Apr 2024 in physics.comp-ph, cs.AI, cs.NA, and math.NA

Abstract: Magnetohydrodynamics (MHD) plays a pivotal role in describing the dynamics of plasma and conductive fluids, essential for understanding phenomena such as the structure and evolution of stars and galaxies, and in nuclear fusion for plasma motion through ideal MHD equations. Solving these hyperbolic PDEs requires sophisticated numerical methods, presenting computational challenges due to complex structures and high costs. Recent advances introduce neural operators like the Fourier Neural Operator (FNO) as surrogate models for traditional numerical analyses. This study explores a modified Flux Fourier neural operator model to approximate the numerical flux of ideal MHD, offering a novel approach that outperforms existing neural operator models by enabling continuous inference, generalization outside sampled distributions, and faster computation compared to classical numerical schemes.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (64)
  1. P. Abrahamsen. A review of gaussian random fields and correlation functions. Tech. Rep. Rapport 917,, SNorsk Regnesentral Norwegian Computing Center, Oslo, 1997.
  2. H. Alfven. Existence of electromagnetic-hydrodynamic waves. Nature, 150:405–406, 1942.
  3. L. Bar and N. Socher. Strong solutions for pde-based tomography by unsupervised learning. SIAM Journal on Imaging Sciences, 14:128–155, 2021.
  4. Out-of-distributional risk bounds for neural operators with applications to the helmholtz equation. arXiv, arXiv:2301.11509, 2023.
  5. J.A. Bittencourt. Fundamentals of Plasma Physics. Springer New York, NY, 2004.
  6. An improved weighted essentially non-oscillatory scheme for hyperbolic conservation laws. Journal of Computational Physics, 227:3191–3211, 2008.
  7. The effect of nonzero ∇⋅b⋅∇𝑏\nabla\cdot b∇ ⋅ italic_b on the numerical solution of the magnetohydrodynamic equations. Journal of Computational Physics, 35:426, 1980.
  8. M. Brio and C.C. Wu. An upwind differencing scheme for the equations of ideal magnetohydrodynamics. Journal of Computational Physics, 75:400–422, 1988.
  9. Designing neural networks for hyperbolic conservation laws. arXiv, arXiv:2211.14375, 2022.
  10. Finite difference weighted essentially non-oscillatory schemes with constrained transport for ideal magnetohydrodynamics. Journal of Computational Physics, 268:302–325, 2014.
  11. B. Costa and W.S. Don. High order hybrid central-weno finite difference scheme for conservation laws. Journal of Computational and Applied Mathematics, 204:209–218, 2007.
  12. On the solution of nonlinear hyperbolic differential equations by finite differences. Communications on Pure and Applied Mathematics, 5(3):243–255, 1952.
  13. W. Dai and P.R. Woodward. A simple finite difference scheme for multidimensional magnetohydrodynamical equations. Journal of Computational Physics, 142:331–369, 1998.
  14. Simulation of magnetohydrodynamic flows: A constrained transport method. The Astrophysical Journal, 332, 1988.
  15. L. Fu. A very-high-order teno scheme for all-speed gas dynamics and turbulence. Computer Physics Communications, 244:117–131, 2019.
  16. X. Xiao G. Gupta and P. Bogdan. Multiwavelet-based operator learning for differential equations. Advances in Neural Information Processing Systems, 2021.
  17. S.K. Godunov. A difference method for numerical calculation of discontinuous solutions of the equations of hydrodynamics. Matematicheskii Sbornik, 47(3):271–306, 1959.
  18. Capacity bounds for the deeponet method of solving differential equations. arXiv, arXiv:2205.11359, 2022.
  19. S. Gottlieb and C.W. Shu. Total variation diminishing runge-kutta schemes. Mathematics of Computation, 67:73–85, 1998.
  20. Uniformly high order essentially non- oscillatory schemes iii. Journal of Computational Physics, 71(2):231–303, 1987.
  21. Mappedweighted-essentially-non-oscillatoryschemes:achieving optimal order near critical points. Journal of Computational Physics, 207:542–567, 2005.
  22. Hybrid tuned center-difference-weno method for larged eddy simulations in the presence of strong shocks. Journal of Computational Physics, 194:435–450, 2004.
  23. Learning to control pdes with differentiable physics. ICLR 2020, 2020.
  24. Scale-invariant learning by physics inversion. Advances in Neural Information Processing Systems, 2022.
  25. Generalization Error in Deep Learning. Birkhäuser Cham, 2019.
  26. G. Jiang and C.W. Shu. Efficient implementation of weighted eno schemes. Journal of Computational Physics, 126:202–228, 1996.
  27. G. Jiang and C. Wu. A high-order weno finite difference scheme for the equations of ideal magnetohydrodynamics. Journal of Computational Physics, 150:561–594, 1999.
  28. S. Jin. Runge-kutta methods for hyperbolic conservation laws with stiff relaxation terms. Journal of Computational Physics, 122(1):51–67, 1995.
  29. Perspectives on space and astrophysical plasma physics. Unstable Current Systems and Plasma Instabilities in Astrophysics, 107:537–552, 1985.
  30. T. Kim and M. Kang. Approximating numerical fluxes using fourier neural operators for hyperbolic conservation laws. arXiv, arXiv:2401.01783, 2024a.
  31. T. Kim and M. Kang. Bounding the rademacher complexity of fourier neural operators. Machine Learning, 2024b.
  32. On universal approximation and error bounds for fourier neural operators. Journal of Machine Learning Research, 22(290), 2021a.
  33. Neural operator: Learning maps between function spaces. arXiv, arXiv:2108.08481, 2021b.
  34. P.D. Lax. Weak solutions of nonlinear hyperbolic equations and their numerical computation. Communications on Pure and Applied Mathematics, 7(1):159–193, 1954.
  35. Hyperdeeponet: learning operator with complex target function space using the limited resources via hypernetwork. ICLR 2023, 2023.
  36. B.V. Leer. Towards the ultimate conservative difference scheme. ii. monotonicity and conservation combined in a second-order scheme. Journal of Computational Physics, 14(4):361–370, 1974.
  37. R.J. LeVeque. Numerical Methods for Conservation Laws. Birkhäuser Basel, 1992.
  38. Neural operator: Graph kernel network for partial differential equations. ICLR 2020 Workshop ODE/PDE+DL, 2020.
  39. Fourier neural operator for parametric partial differential equations. ICLR 2021, 2021.
  40. Weighted essentially non-oscillatory schemes. Journal of Computational Physics, 115(1):200–212, 1994.
  41. Learning nonlinear operators via deeponet based on the universal approximation theorem of operators. Nature Machine Intelligence, 3(3):218–229, 2021.
  42. Constraint-aware neural networks for riemann problems. Journal of Computational Physics, 409:109345, 2020.
  43. Global magnetohydrodynamic simulations: Performance quantification of magnetopause distances and convection potential predictions. Frontiers in Astronomy and Space Sciences, 8(45), 2021.
  44. K. O’Shea and R. Nash. An introduction to convolutional neural networks. arXiv, arXiv:1511.08458, 2015.
  45. R. Pakmor and V. Springel. Simulations of magnetic fields in isolated disc galaxies. Monthly Notices of the Royal Astronomical Society, 432(1), 2013.
  46. Fourcastnet: A global data-driven high-resolution weather model using adaptive fourier neural operators. arXiv, arXiv:2202.11214, 2022.
  47. S. Pirozzoli. Conservative hybrid compact-weno schemes for shock–turbulence interaction. Journal of Computational Physics, 178:81–117, 2002.
  48. E.R. Priest. Solar Magnetohydrodynamics. Springer Dordrecht, 1982.
  49. D. Ray and J.S. Hesthaven. An artificial neural network as a troubled-cell indicator. Journal of Computational Physics, 367:166–191, 2018.
  50. F. Rincon. Dynamo theories. Journal of Plasma Physics, 85(4), 2019.
  51. Magnetohydrodynamics with physics informed neural operators. Machine Learning: Science and Technology, 4, 2023.
  52. J.A. Rossmanith. An unstaggered, high-resolution constrained transport method for magnetohydrodynamic flows. SIAM Journal on Scientific Computing, 28:1766–1797, 2006.
  53. Neural-network-based riemann solver for real fluids and high explosives; application to computational fluid dynamics. Physics of Fluids, 34(11):116121, 2022.
  54. G. Sebastien. Introduction to Modern Magnetohydrodynamics. Cambridge University Press, 2016.
  55. Understanding Machine Learning: From Theory to Algorithms. Cambridge University Press, San MateoShaftesbury Road, Cambridge, 2014.
  56. K. Shibata and T. Magara. Solar flares: Magnetohydrodynamic processes. Living Reviews in Solar Physics, 8(1), 2011.
  57. P.K. Sweby. High resolution schemes using flux limiters for hyperbolic conservation laws. SIAM journal on numerical analysis, 21(5):995–1011, 1984.
  58. G. Toth. The ∇⋅b⋅∇𝑏\nabla\cdot b∇ ⋅ italic_b=0 constraint in shock-capturing magnetohydrodynamics codes. Journal of Computational Physics, 161:605–652, 2000.
  59. L.G. Valiant. A theory of the learnable. Communications of the ACM, 27(11):1134–1142, 1984.
  60. V.N. Vapnik. An overview of statistical learning theory. IEEE Transactions on Neural Networks, 10(5):988–999, 1999.
  61. A. A. Vlasov. On vibration properties of electron gas. Journal of Experimental and Theoretical Physics, 8:291, 1938.
  62. Learning to discretize: Solving 1d scalar conservation laws via deep reinforcement learning. arXiv, arXiv:1905.11079, 2020.
  63. U-fno—an enhanced fourier neural operator-based deep-learning model for multiphase flow. Advances in Water Resources, 163, 2022.
  64. J.A. Wesson. Hydromagnetic stability of tokamaks. Nuclear Fusion, 18:87–132, 1978.

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now