Papers
Topics
Authors
Recent
Detailed Answer
Quick Answer
Concise responses based on abstracts only
Detailed Answer
Well-researched responses based on abstracts and relevant paper content.
Custom Instructions Pro
Preferences or requirements that you'd like Emergent Mind to consider when generating responses
Gemini 2.5 Flash
Gemini 2.5 Flash 45 tok/s
Gemini 2.5 Pro 49 tok/s Pro
GPT-5 Medium 11 tok/s Pro
GPT-5 High 19 tok/s Pro
GPT-4o 88 tok/s Pro
Kimi K2 214 tok/s Pro
GPT OSS 120B 460 tok/s Pro
Claude Sonnet 4 38 tok/s Pro
2000 character limit reached

Scientific machine learning for closure models in multiscale problems: a review (2403.02913v2)

Published 5 Mar 2024 in math.NA and cs.NA

Abstract: Closure problems are omnipresent when simulating multiscale systems, where some quantities and processes cannot be fully prescribed despite their effects on the simulation's accuracy. Recently, scientific machine learning approaches have been proposed as a way to tackle the closure problem, combining traditional (physics-based) modeling with data-driven (machine-learned) techniques, typically through enriching differential equations with neural networks. This paper reviews the different reduced model forms, distinguished by the degree to which they include known physics, and the different objectives of a priori and a posteriori learning. The importance of adhering to physical laws (such as symmetries and conservation laws) in choosing the reduced model form and choosing the learning method is discussed. The effect of spatial and temporal discretization and recent trends toward discretization-invariant models are reviewed. In addition, we make the connections between closure problems and several other research disciplines: inverse problems, Mori-Zwanzig theory, and multi-fidelity methods. In conclusion, much progress has been made with scientific machine learning approaches for solving closure problems, but many challenges remain. In particular, the generalizability and interpretability of learned models is a major issue that needs to be addressed further.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (138)
  1. doi:10.1093/acrefore/9780190228620.013.826.
  2. doi:10.1007/BF02179552.
  3. doi:10.1073/pnas.1909854116.
  4. doi:10.1016/j.jcp.2022.110941.
  5. arXiv:2310.19385, doi:10.48550/arXiv.2310.19385.
  6. doi:10.1088/1748-9326/ac0eb0.
  7. doi:10.1029/2018MS001472.
  8. doi:10.2118/37324-JPT.
  9. doi:10.1146/annurev-biophys-083012-130348.
  10. doi:10.1088/1361-651X/ab7150.
  11. doi:10.1146/annurev-fluid-010518-040547.
  12. arXiv:2303.17078, doi:10.48550/arXiv.2303.17078.
  13. doi:10.1098/rspa.2023.0058.
  14. doi:10.1063/5.0128446.
  15. doi:10.1007/978-3-319-22470-1.
  16. doi:10.1126/sciadv.1602614.
  17. doi:10.1016/j.jcp.2018.10.045.
  18. doi:10.1007/s40304-018-0127-z.
  19. doi:10.1016/j.camwa.2011.09.028.
  20. doi:10.1007/s10409-021-01148-1.
  21. arXiv:2112.15275, doi:10.48550/arXiv.2112.15275.
  22. arXiv:2202.11214, doi:10.48550/arXiv.2202.11214.
  23. arXiv:2212.12794, doi:10.48550/arXiv.2212.12794.
  24. doi:10.1038/s42256-021-00302-5.
  25. arXiv:2010.08895, doi:10.48550/arXiv.2010.08895.
  26. arXiv:2301.13770, doi:10.48550/arXiv.2301.13770.
  27. arXiv:1905.09883, doi:10.48550/arXiv.1905.09883.
  28. arXiv:2306.01174, doi:10.48550/arXiv.2306.01174.
  29. arXiv:2105.01030.
  30. doi:10.5194/gmd-13-2185-2020.
  31. doi:10.1016/j.cma.2016.03.025.
  32. doi:10.1103/PhysRevFluids.2.054604.
  33. doi:10.1017/jfm.2016.615.
  34. doi:10.1016/j.jcp.2016.05.003.
  35. doi:10.1016/j.jcp.2016.07.038.
  36. arXiv:1606.07987, doi:10.1103/PhysRevFluids.2.034603.
  37. doi:10.2514/6.2018-2900.
  38. doi:10.1016/j.jcp.2019.108910.
  39. doi:10.1103/PhysRevFluids.4.034602.
  40. doi:10.1017/jfm.2020.931.
  41. doi:10.1017/jfm.2018.770.
  42. doi:10.1553/etna_vol56s117.
  43. doi:10.1016/j.physd.2022.133568.
  44. doi:10.1063/5.0149861.
  45. arXiv:2207.11417, doi:10.48550/arXiv.2207.11417.
  46. doi:10.1073/pnas.1810286115.
  47. arXiv:2009.10675, doi:10.1103/PhysRevFluids.6.050504.
  48. doi:10.23967/wccm-eccomas.2020.115.
  49. doi:10.1016/j.jcp.2022.111090.
  50. doi:10.1103/PhysRevFluids.6.050501.
  51. doi:10.1038/s41467-020-17142-3.
  52. doi:10.1103/PhysRevFluids.7.024305.
  53. arXiv:2306.05014, doi:10.48550/arXiv.2306.05014.
  54. arXiv:2307.13144, doi:10.48550/arXiv.2307.13144.
  55. doi:10.1016/j.camwa.2023.04.030.
  56. doi:10.1017/jfm.2022.738.
  57. doi:10.1002/gamm.202100002.
  58. doi:10.1016/j.cma.2022.115457.
  59. doi:10.1007/978-1-4020-8839-1.
  60. doi:10.2514/6.2019-1884.
  61. doi:10.1016/j.jcp.2020.109811.
  62. arXiv:2209.11614.
  63. arXiv:2307.03683, doi:10.48550/arXiv.2307.03683.
  64. doi:10.1029/2022MS003124.
  65. arXiv:2308.05732, doi:10.48550/arXiv.2308.05732.
  66. doi:10.1016/j.cma.2023.116161.
  67. arXiv:2307.13517, doi:10.48550/arXiv.2307.13517.
  68. arXiv:2107.02093, doi:10.48550/arXiv.2107.02093.
  69. doi:10.1016/j.camwa.2023.06.012.
  70. doi:10.1063/1.869867.
  71. doi:10.1007/s00162-019-00512-z.
  72. doi:10.1017/jfm.2017.637.
  73. doi:10.1080/23311940.2018.1464368.
  74. doi:10.1063/5.0027146.
  75. doi:10.1063/5.0059643.
  76. doi:10.1017/jfm.2020.948.
  77. arXiv:2303.02338.
  78. arXiv:2006.02619, doi:10.48550/arXiv.2006.02619.
  79. doi:10.1098/rspa.2021.0904.
  80. arXiv:2308.06675, doi:10.48550/arXiv.2308.06675.
  81. doi:10.1016/j.jcp.2018.02.039.
  82. doi:10.1038/s41467-022-28957-7.
  83. doi:10.1063/5.0106940.
  84. arXiv:1908.04127, doi:10.48550/arXiv.1908.04127.
  85. doi:10.1017/S0022112085001987.
  86. doi:10.1063/1.4974093.
  87. doi:10.1016/j.engappai.2023.107483.
  88. arXiv:physics/0605106, doi:10.3842/SIGMA.2006.052.
  89. doi:10.1007/BF00271419.
  90. arXiv:1906.06622, doi:10.48550/arXiv.1906.06622.
  91. doi:10.1002/fld.4684.
  92. doi:10.1017/jfm.2021.994.
  93. arXiv:2002.03061, doi:10.48550/arXiv.2002.03061.
  94. doi:10.1063/5.0066049.
  95. doi:10.1016/j.physa.2022.128327.
  96. arXiv:2310.18897, doi:10.48550/arXiv.2310.18897.
  97. doi:10.1098/rspa.2020.1004.
  98. arXiv:1806.07366.
  99. P. Kidger, On Neural Differential Equations (Feb. 2022). arXiv:2202.02435, doi:10.48550/arXiv.2202.02435.
  100. arXiv:2001.04385, doi:10.48550/arXiv.2001.04385.
  101. doi:10.1201/9780203749319.
  102. arXiv:2005.13420, doi:10.48550/arXiv.2005.13420.
  103. doi:10.1073/pnas.2101784118.
  104. doi:10.1016/j.taml.2022.100389.
  105. arXiv:2202.03376, doi:10.48550/arXiv.2202.03376.
  106. doi:10.1073/pnas.1814058116.
  107. arXiv:2207.00556, doi:10.48550/arXiv.2207.00556.
  108. doi:10.1016/S0898-1221(03)90019-8.
  109. doi:10.1017/S0022112003006268.
  110. arXiv:2206.01178, doi:10.48550/arXiv.2206.01178.
  111. doi:10.1017/S0962492910000061.
  112. doi:10.1002/wics.1427.
  113. doi:10.1137/1.9781611974546.
  114. doi:10.1016/j.jcp.2013.10.027.
  115. doi:10.3934/fods.2020004.
  116. doi:10.1007/s10409-021-01152-5.
  117. doi:10.2140/camcos.2006.1.1.
  118. arXiv:1611.03311, doi:10.1103/PhysRevFluids.2.014604.
  119. doi:10.1016/j.jcp.2017.07.053.
  120. arXiv:1803.09318, doi:10.1137/18M1177263.
  121. doi:10.1175/BAMS-D-15-00268.1.
  122. doi:10.1016/j.jcp.2018.06.038.
  123. doi:10.1016/j.physd.2021.132894.
  124. doi:10.1063/5.0131929.
  125. arXiv:2204.09157, doi:10.1016/j.jcp.2023.112462.
  126. doi:10.3934/fods.2020019.
  127. doi:10.1007/978-0-387-73829-1.
  128. doi:10.1017/CBO9780511613203.
  129. doi:10.1017/jfm.2023.510.
  130. P. Spalart, An Old-Fashioned Framework for Machine Learning in Turbulence Modeling (Aug. 2023). arXiv:2308.00837, doi:10.48550/arXiv.2308.00837.
  131. doi:10.1007/s10494-019-00089-x.
  132. doi:10.1029/2020GL088376.
  133. doi:10.1063/5.0091282.
  134. arXiv:1803.04779, doi:10.1063/1.5028373.
  135. doi:10.1103/PhysRevResearch.2.012080.
  136. arXiv:2402.04467, doi:10.48550/arXiv.2402.04467.
  137. doi:10.1063/5.0070890.
  138. arXiv:2210.07182, doi:10.48550/arXiv.2210.07182.
Citations (14)
List To Do Tasks Checklist Streamline Icon: https://streamlinehq.com

Collections

Sign up for free to add this paper to one or more collections.

Summary

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

Dice Question Streamline Icon: https://streamlinehq.com

Follow-Up Questions

We haven't generated follow-up questions for this paper yet.

X Twitter Logo Streamline Icon: https://streamlinehq.com

Don't miss out on important new AI/ML research

See which papers are being discussed right now on X, Reddit, and more:

“Emergent Mind helps me see which AI papers have caught fire online.”

Philip

Philip

Creator, AI Explained on YouTube