On the eigenvector bias of Fourier feature networks: From regression to solving multi-scale PDEs with physics-informed neural networks

Physics-informed neural networks (PINNs) are demonstrating remarkable promise in integrating physical models with gappy and noisy observational data, but they still struggle in cases where the target functions to be approximated exhibit high-frequency or multi-scale features. In this work we investi...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Computer methods in applied mechanics and engineering Ročník 384; číslo C; s. 113938
Hlavní autori: Wang, Sifan, Wang, Hanwen, Perdikaris, Paris
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: Amsterdam Elsevier B.V 01.10.2021
Elsevier BV
Elsevier
Predmet:
ISSN:0045-7825, 1879-2138
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Abstract Physics-informed neural networks (PINNs) are demonstrating remarkable promise in integrating physical models with gappy and noisy observational data, but they still struggle in cases where the target functions to be approximated exhibit high-frequency or multi-scale features. In this work we investigate this limitation through the lens of Neural Tangent Kernel (NTK) theory and elucidate how PINNs are biased towards learning functions along the dominant eigen-directions of their limiting NTK. Using this observation, we construct novel architectures that employ spatio-temporal and multi-scale random Fourier features, and justify how such coordinate embedding layers can lead to robust and accurate PINN models. Numerical examples are presented for several challenging cases where conventional PINN models fail, including wave propagation and reaction–diffusion dynamics, illustrating how the proposed methods can be used to effectively tackle both forward and inverse problems involving partial differential equations with multi-scale behavior. All code an data accompanying this manuscript will be made publicly available at https://github.com/PredictiveIntelligenceLab/MultiscalePINNs. •We argue that spectral bias in deep neural networks in fact corresponds to “NTK eigenvector bias”.•We show that Fourier feature mappings can modulate the frequency of the NTK eigenvectors.•By analyzing the NTK eigenspace, we engineer new effective architectures for multi-scale problems.•We put forth a collection of challenging benchmarks for PINNs.
AbstractList Physics-informed neural networks (PINNs) are demonstrating remarkable promise in integrating physical models with gappy and noisy observational data, but they still struggle in cases where the target functions to be approximated exhibit high-frequency or multi-scale features. In this work we investigate this limitation through the lens of Neural Tangent Kernel (NTK) theory and elucidate how PINNs are biased towards learning functions along the dominant eigen-directions of their limiting NTK. Using this observation, we construct novel architectures that employ spatio-temporal and multi-scale random Fourier features, and justify how such coordinate embedding layers can lead to robust and accurate PINN models. Numerical examples are presented for several challenging cases where conventional PINN models fail, including wave propagation and reaction–diffusion dynamics, illustrating how the proposed methods can be used to effectively tackle both forward and inverse problems involving partial differential equations with multi-scale behavior. All code an data accompanying this manuscript will be made publicly available at https://github.com/PredictiveIntelligenceLab/MultiscalePINNs. •We argue that spectral bias in deep neural networks in fact corresponds to “NTK eigenvector bias”.•We show that Fourier feature mappings can modulate the frequency of the NTK eigenvectors.•By analyzing the NTK eigenspace, we engineer new effective architectures for multi-scale problems.•We put forth a collection of challenging benchmarks for PINNs.
Physics-informed neural networks (PINNs) are demonstrating remarkable promise in integrating physical models with gappy and noisy observational data, but they still struggle in cases where the target functions to be approximated exhibit high-frequency or multi-scale features. In this work we investigate this limitation through the lens of Neural Tangent Kernel (NTK) theory and elucidate how PINNs are biased towards learning functions along the dominant eigen-directions of their limiting NTK. Using this observation, we construct novel architectures that employ spatio-temporal and multi-scale random Fourier features, and justify how such coordinate embedding layers can lead to robust and accurate PINN models. Numerical examples are presented for several challenging cases where conventional PINN models fail, including wave propagation and reaction–diffusion dynamics, illustrating how the proposed methods can be used to effectively tackle both forward and inverse problems involving partial differential equations with multi-scale behavior. All code an data accompanying this manuscript will be made publicly available at https://github.com/PredictiveIntelligenceLab/MultiscalePINNs.
ArticleNumber 113938
Author Perdikaris, Paris
Wang, Hanwen
Wang, Sifan
Author_xml – sequence: 1
  givenname: Sifan
  surname: Wang
  fullname: Wang, Sifan
  email: sifanw@sas.upenn.edu
  organization: Graduate Group in Applied Mathematics and Computational Science, University of Pennsylvania, Philadelphia, PA 19104, United States of America
– sequence: 2
  givenname: Hanwen
  orcidid: 0000-0002-4990-6810
  surname: Wang
  fullname: Wang, Hanwen
  email: wangh19@sas.upenn.edu
  organization: Graduate Group in Applied Mathematics and Computational Science, University of Pennsylvania, Philadelphia, PA 19104, United States of America
– sequence: 3
  givenname: Paris
  orcidid: 0000-0002-2816-3229
  surname: Perdikaris
  fullname: Perdikaris, Paris
  email: pgp@seas.upenn.edu
  organization: Department of Mechanical Engineering and Applied Mechanics, University of Pennsylvania, Philadelphia, PA 19104, United States of America
BackLink https://www.osti.gov/biblio/1786215$$D View this record in Osti.gov
BookMark eNp9kcFu3CAYhFGVSt2kfYDeUHr2lh_bGKenKMk2lSKlh_aMMPt7l60NG8Ab5QX63GHlKIccgpC4zIyG-U7JifMOCfkKbAkMxPfd0ox6yRmHJUDZlvIDWYBs2oJDKU_IgrGqLhrJ60_kNMYdy0cCX5D_946mLVK0G3QHNMkH2lkdqe_pyk_BYqA96jQFpA7Tow__4gVdBT_SgJuAMVqfEzyNfjhYt6HjNCRbRKMHpL-vbyJ9tGlL99unaE0srOt9GHGds6agh9fIz-Rjr4eIX17eM_J3dfPn6ra4u__56-ryrjBlW6dCNxwYE2swphHQCUAJpq47KarO5NsIxFZ3WppKdFx3PUiuTYNVX6FEycszcj7n-pisisYmNFvjncs_V9BIwaHOom-zaB_8w4QxqV1ewuVeiteCt7yUDLKqmVUm-BgD9iqn6ZTnSEHbQQFTRzJqpzIZdSSjZjLZCW-c-2BHHZ7e9fyYPZjXOWQqx_LoDK5tOHZfe_uO-xltKapt
CitedBy_id crossref_primary_10_1016_j_cma_2023_116012
crossref_primary_10_1007_s10845_024_02427_x
crossref_primary_10_1016_j_camwa_2025_06_024
crossref_primary_10_1016_j_cpc_2025_109782
crossref_primary_10_1088_1361_6420_ace9d4
crossref_primary_10_1016_j_est_2025_117055
crossref_primary_10_1016_j_cmpb_2024_108081
crossref_primary_10_1007_s10409_024_24140_x
crossref_primary_10_1007_s41116_023_00038_x
crossref_primary_10_1016_j_cnsns_2025_109343
crossref_primary_10_1007_s11831_024_10145_z
crossref_primary_10_1016_j_jcp_2024_113090
crossref_primary_10_1016_j_cageo_2024_105611
crossref_primary_10_1007_s10915_022_01881_0
crossref_primary_10_1063_5_0285013
crossref_primary_10_1016_j_oceaneng_2024_118341
crossref_primary_10_1007_s11831_023_09890_4
crossref_primary_10_1007_s10915_022_01980_y
crossref_primary_10_1007_s00707_023_03691_3
crossref_primary_10_1038_s41598_024_76983_w
crossref_primary_10_1016_j_cma_2022_114909
crossref_primary_10_1016_j_apm_2025_116232
crossref_primary_10_1016_j_cma_2025_118308
crossref_primary_10_1007_s11431_024_2873_2
crossref_primary_10_1007_s10444_023_10065_9
crossref_primary_10_1016_j_aei_2025_103581
crossref_primary_10_1109_TFUZZ_2023_3291488
crossref_primary_10_1785_0220240377
crossref_primary_10_1088_1361_6463_acbec3
crossref_primary_10_3390_math10224186
crossref_primary_10_1007_s00466_024_02516_x
crossref_primary_10_1016_j_compbiomed_2024_109528
crossref_primary_10_1016_j_buildenv_2025_113616
crossref_primary_10_1016_j_jcp_2021_110668
crossref_primary_10_1109_TGRS_2024_3440471
crossref_primary_10_1016_j_ymssp_2024_111920
crossref_primary_10_1109_LAWP_2024_3355896
crossref_primary_10_1145_3706106
crossref_primary_10_1002_adom_202403304
crossref_primary_10_1038_s41598_025_99354_5
crossref_primary_10_1016_j_physd_2023_134023
crossref_primary_10_1016_j_jcp_2025_114226
crossref_primary_10_1016_j_camwa_2024_01_021
crossref_primary_10_1016_j_cma_2024_117042
crossref_primary_10_1016_j_cam_2024_115874
crossref_primary_10_1109_TGRS_2023_3295414
crossref_primary_10_1016_j_biosystemseng_2025_01_017
crossref_primary_10_1016_j_neunet_2024_106998
crossref_primary_10_1016_j_engappai_2025_112149
crossref_primary_10_1038_s44384_025_00021_w
crossref_primary_10_1063_5_0281347
crossref_primary_10_1007_s10994_023_06442_6
crossref_primary_10_1016_j_jcp_2025_113905
crossref_primary_10_1002_nme_7388
crossref_primary_10_1016_j_neunet_2024_106750
crossref_primary_10_1007_s10462_025_11322_7
crossref_primary_10_1029_2023MS004033
crossref_primary_10_1061__ASCE_EM_1943_7889_0002156
crossref_primary_10_1016_j_physd_2023_133987
crossref_primary_10_1016_j_cma_2024_117036
crossref_primary_10_1016_j_engstruct_2025_119884
crossref_primary_10_1016_j_engappai_2023_107324
crossref_primary_10_1016_j_engappai_2024_109804
crossref_primary_10_1063_5_0274928
crossref_primary_10_1016_j_camwa_2022_12_008
crossref_primary_10_1016_j_engappai_2023_107453
crossref_primary_10_1016_j_cma_2024_117706
crossref_primary_10_1029_2025WR040052
crossref_primary_10_11648_j_ajcbm_20250902_11
crossref_primary_10_1016_j_neunet_2024_106886
crossref_primary_10_1007_s10462_024_10784_5
crossref_primary_10_1002_smtd_202400620
crossref_primary_10_1007_s10489_024_06195_2
crossref_primary_10_1016_j_cam_2022_114963
crossref_primary_10_1016_j_jcp_2024_112760
crossref_primary_10_1016_j_cma_2024_117024
crossref_primary_10_1146_annurev_conmatphys_043024_114758
crossref_primary_10_1007_s10910_022_01400_4
crossref_primary_10_1016_j_apm_2025_116301
crossref_primary_10_1016_j_cma_2024_117268
crossref_primary_10_1016_j_cma_2025_117784
crossref_primary_10_1016_j_cma_2023_116290
crossref_primary_10_1007_s00521_024_09766_z
crossref_primary_10_1016_j_cnsns_2024_107911
crossref_primary_10_1016_j_finel_2024_104305
crossref_primary_10_1016_j_jmps_2022_105177
crossref_primary_10_1016_j_enggeo_2024_107886
crossref_primary_10_1016_j_physd_2025_134878
crossref_primary_10_1063_5_0251167
crossref_primary_10_1126_sciadv_ads5236
crossref_primary_10_1016_j_jcp_2023_112751
crossref_primary_10_1016_j_ymssp_2024_111297
crossref_primary_10_1016_j_compositesa_2025_108758
crossref_primary_10_1007_s00466_024_02475_3
crossref_primary_10_1016_j_cpc_2025_109572
crossref_primary_10_1063_5_0285282
crossref_primary_10_1016_j_camwa_2025_03_026
crossref_primary_10_1016_j_cnsns_2025_109274
crossref_primary_10_1016_j_spasta_2023_100746
crossref_primary_10_1039_D2CP01560A
crossref_primary_10_3390_polym15244711
crossref_primary_10_1016_j_euromechsol_2024_105551
crossref_primary_10_1016_j_cma_2023_116184
crossref_primary_10_1016_j_jocs_2025_102631
crossref_primary_10_1016_j_trb_2024_103064
crossref_primary_10_1007_s10589_024_00597_1
crossref_primary_10_1016_j_optlaseng_2025_108913
crossref_primary_10_1364_AO_571796
crossref_primary_10_1016_j_icheatmasstransfer_2024_108330
crossref_primary_10_1109_ACCESS_2023_3244681
crossref_primary_10_1016_j_jcp_2024_112865
crossref_primary_10_1016_j_jcp_2024_112986
crossref_primary_10_1016_j_cma_2022_115141
crossref_primary_10_1016_j_engappai_2023_106425
crossref_primary_10_1016_j_watres_2025_123779
crossref_primary_10_1016_j_cma_2024_117122
crossref_primary_10_1016_j_cma_2025_117764
crossref_primary_10_1016_j_cma_2024_117004
crossref_primary_10_1016_j_cma_2025_118059
crossref_primary_10_1016_j_jcp_2022_111173
crossref_primary_10_1007_s11207_023_02226_2
crossref_primary_10_1121_10_0025235
crossref_primary_10_1016_j_jhydrol_2025_133654
crossref_primary_10_1134_S1064562423701120
crossref_primary_10_3390_app13126892
crossref_primary_10_3390_jmse12122344
crossref_primary_10_1016_j_compbiomed_2023_106693
crossref_primary_10_1016_j_compgeo_2025_107652
crossref_primary_10_1016_j_compgeo_2023_105433
crossref_primary_10_1002_nme_7585
crossref_primary_10_1016_j_ymssp_2025_112561
crossref_primary_10_1109_JLT_2023_3322893
crossref_primary_10_1007_s00158_024_03834_7
crossref_primary_10_1016_j_fuel_2025_136475
crossref_primary_10_1016_j_cma_2024_117116
crossref_primary_10_1016_j_media_2023_102925
crossref_primary_10_3390_w15132320
crossref_primary_10_1140_epjp_s13360_025_06644_y
crossref_primary_10_1016_j_cma_2022_115491
crossref_primary_10_1080_10407790_2023_2264489
crossref_primary_10_1109_TAP_2024_3433389
crossref_primary_10_1016_j_ymssp_2024_112274
crossref_primary_10_1016_j_cma_2024_116813
crossref_primary_10_1016_j_neunet_2025_107831
crossref_primary_10_1109_JMMCT_2023_3345798
crossref_primary_10_1016_j_ultrasmedbio_2024_08_004
crossref_primary_10_1016_j_jcp_2024_113012
crossref_primary_10_3390_math12071079
crossref_primary_10_1016_j_jcp_2023_112683
crossref_primary_10_1016_j_compbiomed_2024_109476
crossref_primary_10_1016_j_camwa_2024_06_029
crossref_primary_10_1016_j_cma_2025_118279
crossref_primary_10_1088_1402_4896_adc346
crossref_primary_10_1029_2022WR033168
crossref_primary_10_3389_fphy_2022_1062554
crossref_primary_10_1016_j_neunet_2025_107983
crossref_primary_10_1016_j_cma_2024_116805
crossref_primary_10_1016_j_neucom_2025_130578
crossref_primary_10_3390_app14083204
crossref_primary_10_1007_s00521_024_09883_9
crossref_primary_10_1016_j_ijmecsci_2024_109766
crossref_primary_10_1016_j_cma_2022_115399
crossref_primary_10_1016_j_cma_2025_117751
crossref_primary_10_1016_j_cma_2024_117691
crossref_primary_10_1007_s00366_025_02158_4
crossref_primary_10_1016_j_cma_2025_118043
crossref_primary_10_1016_j_neunet_2024_106703
crossref_primary_10_1016_j_neunet_2025_107978
crossref_primary_10_1016_j_ymssp_2025_112914
crossref_primary_10_1016_j_neunet_2025_108027
crossref_primary_10_1016_j_neunet_2025_107179
crossref_primary_10_1177_14750902231166424
crossref_primary_10_1007_s11071_022_08161_4
crossref_primary_10_1016_j_jcp_2024_112944
crossref_primary_10_1016_j_engstruct_2025_121218
crossref_primary_10_1016_j_engappai_2023_106867
crossref_primary_10_1016_j_jcp_2023_112464
crossref_primary_10_1016_j_jcp_2024_113112
crossref_primary_10_1088_2632_2153_ad2626
crossref_primary_10_1016_j_taml_2024_100494
crossref_primary_10_1016_j_cnsns_2024_108229
crossref_primary_10_1016_j_jcp_2022_111768
crossref_primary_10_1016_j_ymssp_2023_110147
crossref_primary_10_1016_j_jcp_2025_114050
crossref_primary_10_1016_j_neucom_2023_126424
crossref_primary_10_1016_j_cpc_2024_109202
crossref_primary_10_1088_2632_2153_ad652d
crossref_primary_10_1016_j_cma_2022_115852
crossref_primary_10_1016_j_tws_2024_111693
crossref_primary_10_3390_math11051109
crossref_primary_10_1007_s12145_024_01448_7
crossref_primary_10_1016_j_jer_2024_02_011
crossref_primary_10_1103_PhysRevE_104_025205
crossref_primary_10_1016_j_cma_2025_117956
crossref_primary_10_1016_j_compgeo_2025_107612
crossref_primary_10_1007_s00366_022_01709_3
crossref_primary_10_1016_j_mfglet_2023_08_074
crossref_primary_10_1016_j_compind_2025_104304
crossref_primary_10_1038_s41598_022_19157_w
crossref_primary_10_1029_2024JH000162
crossref_primary_10_1016_j_cma_2025_117852
crossref_primary_10_1016_j_jcp_2024_113341
crossref_primary_10_1137_21M1460764
crossref_primary_10_1016_j_cpc_2025_109599
crossref_primary_10_1007_s00521_025_11554_2
crossref_primary_10_1007_s00707_022_03449_3
crossref_primary_10_1007_s10409_021_01144_5
crossref_primary_10_1109_TGRS_2025_3581638
crossref_primary_10_1016_j_ymssp_2023_111006
crossref_primary_10_1109_MITP_2023_3338026
crossref_primary_10_3390_s23218665
crossref_primary_10_1016_j_fusengdes_2025_115182
crossref_primary_10_1007_s10714_024_03322_9
crossref_primary_10_1016_j_tws_2023_111423
crossref_primary_10_1016_j_cma_2021_114333
crossref_primary_10_1016_j_neucom_2024_127240
crossref_primary_10_1016_j_advengsoft_2024_103758
crossref_primary_10_1016_j_chaos_2024_115727
crossref_primary_10_1371_journal_pone_0315762
crossref_primary_10_1016_j_jcp_2023_112242
crossref_primary_10_1016_j_anucene_2025_111607
crossref_primary_10_1016_j_ijmultiphaseflow_2024_104877
crossref_primary_10_3390_math11081787
crossref_primary_10_1016_j_ymssp_2024_111458
crossref_primary_10_1016_j_engappai_2025_112044
crossref_primary_10_1016_j_knosys_2024_111831
crossref_primary_10_1103_PhysRevResearch_4_043090
crossref_primary_10_1016_j_ijmecsci_2023_108575
crossref_primary_10_1016_j_jcp_2025_113843
crossref_primary_10_1016_j_cam_2025_116918
crossref_primary_10_1007_s13160_023_00617_3
crossref_primary_10_1109_ACCESS_2024_3457670
crossref_primary_10_1007_s11071_025_11700_4
crossref_primary_10_1016_j_engappai_2024_107887
crossref_primary_10_1109_TAI_2022_3192362
crossref_primary_10_1007_s44379_025_00015_1
crossref_primary_10_1007_s42967_024_00398_7
crossref_primary_10_1016_j_jcp_2022_111539
crossref_primary_10_1016_j_ijheatfluidflow_2025_109988
crossref_primary_10_1016_j_applthermaleng_2025_125534
crossref_primary_10_1063_5_0279565
crossref_primary_10_1088_3049_4753_adfe73
crossref_primary_10_1103_ytyy_pvys
crossref_primary_10_1016_j_jcp_2024_113676
crossref_primary_10_1063_5_0276518
crossref_primary_10_1140_epjs_s11734_025_01963_8
crossref_primary_10_1016_j_earscirev_2025_105276
crossref_primary_10_1016_j_cam_2024_116408
crossref_primary_10_1063_5_0286561
crossref_primary_10_1016_j_cageo_2023_105477
crossref_primary_10_7498_aps_74_20250129
crossref_primary_10_1080_00295639_2023_2184194
crossref_primary_10_1016_j_jcp_2025_114370
crossref_primary_10_1016_j_tws_2025_113159
crossref_primary_10_1080_17499518_2024_2315301
crossref_primary_10_1016_j_eng_2024_12_034
crossref_primary_10_1109_TNNLS_2025_3545967
crossref_primary_10_35940_ijies_G1106_12060625
crossref_primary_10_1016_j_cma_2024_117518
crossref_primary_10_1016_j_engappai_2025_110327
crossref_primary_10_3390_mca28020062
crossref_primary_10_1016_j_jcp_2023_112258
crossref_primary_10_1016_j_ymssp_2024_111111
crossref_primary_10_1016_j_ijmecsci_2025_110111
crossref_primary_10_1016_j_cma_2025_118346
crossref_primary_10_1029_2021JB023703
crossref_primary_10_1080_17486025_2025_2502029
crossref_primary_10_1016_j_aitf_2025_100005
crossref_primary_10_2118_218402_PA
crossref_primary_10_1137_23M1560574
crossref_primary_10_1016_j_jocs_2025_102696
crossref_primary_10_1007_s00366_025_02174_4
crossref_primary_10_1016_j_jtice_2023_104872
crossref_primary_10_1088_1402_4896_ad5053
crossref_primary_10_1007_s10851_025_01259_5
crossref_primary_10_3390_app15020939
crossref_primary_10_1016_j_engappai_2024_108400
crossref_primary_10_1038_s41598_025_02959_z
crossref_primary_10_1016_j_cma_2023_115944
crossref_primary_10_1002_msd2_70030
crossref_primary_10_1002_nme_7176
crossref_primary_10_1016_j_jcp_2024_113656
crossref_primary_10_1016_j_neunet_2025_107437
crossref_primary_10_1016_j_jcp_2021_110600
crossref_primary_10_1109_TPAMI_2023_3307688
Cites_doi 10.1038/s41586-020-2649-2
10.1109/TIT.2005.850052
10.1002/aic.690381003
10.1109/72.712178
10.3389/fphy.2020.00042
10.1016/j.jcp.2018.10.045
10.1016/j.cma.2019.112732
10.1098/rspa.2020.0290
10.1006/jcph.2002.6995
10.1016/j.jcp.2019.05.027
10.1073/pnas.1718942115
10.1126/science.aaw4741
10.1109/ACCESS.2019.2963375
10.1364/OE.384875
10.1016/j.jcp.2019.109120
10.1143/PTPS.64.346
10.1016/j.jcp.2019.05.024
10.1371/journal.pcbi.1007575
10.1016/j.cma.2019.112623
ContentType Journal Article
Copyright 2021 Elsevier B.V.
Copyright Elsevier BV Oct 1, 2021
Copyright_xml – notice: 2021 Elsevier B.V.
– notice: Copyright Elsevier BV Oct 1, 2021
DBID AAYXX
CITATION
7SC
7TB
8FD
FR3
JQ2
KR7
L7M
L~C
L~D
OTOTI
DOI 10.1016/j.cma.2021.113938
DatabaseName CrossRef
Computer and Information Systems Abstracts
Mechanical & Transportation Engineering Abstracts
Technology Research Database
Engineering Research Database
ProQuest Computer Science Collection
Civil Engineering Abstracts
Advanced Technologies Database with Aerospace
Computer and Information Systems Abstracts – Academic
Computer and Information Systems Abstracts Professional
OSTI.GOV
DatabaseTitle CrossRef
Civil Engineering Abstracts
Technology Research Database
Computer and Information Systems Abstracts – Academic
Mechanical & Transportation Engineering Abstracts
ProQuest Computer Science Collection
Computer and Information Systems Abstracts
Engineering Research Database
Advanced Technologies Database with Aerospace
Computer and Information Systems Abstracts Professional
DatabaseTitleList
Civil Engineering Abstracts
DeliveryMethod fulltext_linktorsrc
Discipline Applied Sciences
Engineering
EISSN 1879-2138
ExternalDocumentID 1786215
10_1016_j_cma_2021_113938
S0045782521002759
GroupedDBID --K
--M
-~X
.DC
.~1
0R~
1B1
1~.
1~5
4.4
457
4G.
5GY
5VS
7-5
71M
8P~
9JN
AABNK
AACTN
AAEDT
AAEDW
AAIAV
AAIKJ
AAKOC
AALRI
AAOAW
AAQFI
AAXUO
AAYFN
ABAOU
ABBOA
ABFNM
ABJNI
ABMAC
ABYKQ
ACAZW
ACDAQ
ACGFS
ACIWK
ACRLP
ACZNC
ADBBV
ADEZE
ADGUI
ADTZH
AEBSH
AECPX
AEKER
AENEX
AFKWA
AFTJW
AGHFR
AGUBO
AGYEJ
AHHHB
AHJVU
AHZHX
AIALX
AIEXJ
AIGVJ
AIKHN
AITUG
AJOXV
ALMA_UNASSIGNED_HOLDINGS
AMFUW
AMRAJ
AOUOD
ARUGR
AXJTR
BJAXD
BKOJK
BLXMC
CS3
DU5
EBS
EFJIC
EFLBG
EO8
EO9
EP2
EP3
F5P
FDB
FIRID
FNPLU
FYGXN
G-Q
GBLVA
GBOLZ
IHE
J1W
JJJVA
KOM
LG9
LY7
M41
MHUIS
MO0
N9A
O-L
O9-
OAUVE
OZT
P-8
P-9
P2P
PC.
PQQKQ
Q38
RNS
ROL
RPZ
SDF
SDG
SDP
SES
SPC
SPCBC
SST
SSV
SSW
SSZ
T5K
TN5
WH7
XPP
ZMT
~02
~G-
29F
9DU
AAQXK
AATTM
AAXKI
AAYWO
AAYXX
ABEFU
ABWVN
ABXDB
ACLOT
ACNNM
ACRPL
ACVFH
ADCNI
ADIYS
ADJOM
ADMUD
ADNMO
AEIPS
AEUPX
AFJKZ
AFPUW
AGQPQ
AI.
AIGII
AIIUN
AKBMS
AKRWK
AKYEP
ANKPU
APXCP
ASPBG
AVWKF
AZFZN
CITATION
EFKBS
EJD
FEDTE
FGOYB
G-2
HLZ
HVGLF
HZ~
R2-
SBC
SET
SEW
VH1
VOH
WUQ
ZY4
~HD
7SC
7TB
8FD
FR3
JQ2
KR7
L7M
L~C
L~D
AALMO
AAPBV
ABPIF
ABPTK
OTOTI
PQEST
ID FETCH-LOGICAL-c395t-a721006d1cc761b61e81c55b864bc4bc76ee9aba8c46b2abf182ac7e4f4e8e823
ISICitedReferencesCount 350
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000687712900004&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 0045-7825
IngestDate Thu May 18 22:27:08 EDT 2023
Sun Nov 09 07:47:57 EST 2025
Sat Nov 29 07:30:00 EST 2025
Tue Nov 18 22:38:41 EST 2025
Fri Feb 23 02:47:35 EST 2024
IsDoiOpenAccess false
IsOpenAccess true
IsPeerReviewed true
IsScholarly true
Issue C
Keywords Spectral bias
Deep learning
Neural Tangent Kernel
Scientific machine learning
Partial differential equations
Language English
LinkModel OpenURL
MergedId FETCHMERGED-LOGICAL-c395t-a721006d1cc761b61e81c55b864bc4bc76ee9aba8c46b2abf182ac7e4f4e8e823
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
USDOE Advanced Research Projects Agency - Energy (ARPA-E)
ORCID 0000-0002-2816-3229
0000-0002-4990-6810
0000000228163229
0000000249906810
OpenAccessLink https://doi.org/10.1016/j.cma.2021.113938
PQID 2562923801
PQPubID 2045269
ParticipantIDs osti_scitechconnect_1786215
proquest_journals_2562923801
crossref_citationtrail_10_1016_j_cma_2021_113938
crossref_primary_10_1016_j_cma_2021_113938
elsevier_sciencedirect_doi_10_1016_j_cma_2021_113938
PublicationCentury 2000
PublicationDate 2021-10-01
2021-10-00
20211001
PublicationDateYYYYMMDD 2021-10-01
PublicationDate_xml – month: 10
  year: 2021
  text: 2021-10-01
  day: 01
PublicationDecade 2020
PublicationPlace Amsterdam
PublicationPlace_xml – name: Amsterdam
– name: Netherlands
PublicationTitle Computer methods in applied mechanics and engineering
PublicationYear 2021
Publisher Elsevier B.V
Elsevier BV
Elsevier
Publisher_xml – name: Elsevier B.V
– name: Elsevier BV
– name: Elsevier
References Hunter (b54) 2007; 9
Liu, Cai, Xu (b40) 2020
Zhu, Zabaras, Koutsourelakis, Perdikaris (b10) 2019; 394
Rahaman, Baratin, Arpit, Draxler, Lin, Hamprecht, Bengio, Courville (b31) 2019
Kuramoto (b52) 1978; 64
Raissi, Yazdani, Karniadakis (b12) 2020; 367
Driscoll, Hale, Trefethen (b49) 2014
Li, Xu, Zhang (b39) 2020
Sun, Gao, Pan, Wang (b11) 2020; 361
Yazdani, Lu, Raissi, Karniadakis (b17) 2020; 16
Kingma, Ba (b43) 2014
Harris, Millman, van der Walt, Gommers, Virtanen, Cournapeau, Wieser, Taylor, Berg, Smith (b55) 2020; 585
Wang, Yu, Perdikaris (b6) 2020
Fang, Zhan (b18) 2019; 8
Han, Jentzen, Weinan (b23) 2018; 115
Lagaris, Likas, Fotiadis (b3) 1998; 9
Gao, Sun, Wang (b28) 2020
Shin, Zhang, Karniadakis (b8) 2020
Jin, Cai, Li, Karniadakis (b13) 2020
Ronen, Jacobs, Kasten, Kritchman (b33) 2019
Reyes, Howard, Perdikaris, Tartakovsky (b14) 2020
Wang, Zhang, Cai (b38) 2020
Bradbury, Frostig, Hawkins, Johnson, Leary, Maclaurin, Necula, Paszke, VanderPlas, Wanderman-Milne, Zhang (b53) 2018
Li, Kovachki, Azizzadenesheli, Liu, Bhattacharya, Stuart, Anandkumar (b9) 2020
Wang, Teng, Perdikaris (b29) 2020
Raissi (b24) 2018
Shin, Darbon, Karniadakis (b5) 2020
Costabal, Yang, Perdikaris, Hurtado, Kuhl (b15) 2020; 8
Luo, Yang (b7) 2020
Cox, Matthews (b50) 2002; 176
Lu, Pestourie, Yao, Wang, Verdugo, Johnson (b42) 2021
Karumuri, Tripathy, Bilionis, Panchal (b25) 2020; 404
Xavier Glorot, Yoshua Bengio, Understanding the difficulty of training deep feedforward neural networks, in: Proceedings of the thirteenth international conference on artificial intelligence and statistics, 2010, pp. 249–256.
(b45) 1998
Lee, Xiao, Schoenholz, Bahri, Novak, Sohl-Dickstein, Pennington (b36) 2019
Chen, Lu, Karniadakis, Negro (b19) 2020; 28
Sivashinsky (b51) 1977; 4
Yang, Perdikaris (b26) 2019; 394
Tancik, Srinivasan, Mildenhall, Fridovich-Keil, Raghavan, Singhal, Ramamoorthi, Barron, Ng (b41) 2020
Patel, Manickam, Trask, Wood, Lee, Tomas, Cyr (b30) 2020
Hesthaven, Gottlieb, Gottlieb (b46) 2007
Arora, Du, Hu, Li, Salakhutdinov, Wang (b35) 2019
Shawe-Taylor, Williams, Cristianini, Kandola (b44) 2005; 51
Psichogios, Ungar (b2) 1992; 38
Kissas, Yang, Hwuang, Witschey, Detre, Perdikaris (b16) 2020; 358
Raissi, Perdikaris, Karniadakis (b4) 2019; 378
Wang, Perdikaris (b20) 2020
Jacot, Gabriel, Hongler (b34) 2018
Yang, Meng, Karniadakis (b21) 2020
Williams, Trager, Silva, Panozzo, Zorin, Bruna (b37) 2019
Baydin, Pearlmutter, Radul, Siskind (b1) 2017; 18
Cao, Fang, Wu, Zhou, Gu (b32) 2019
Gray, Scott (b48) 1990
Yang, Bhouri, Perdikaris (b22) 2020; 476
Lu, Jin, Karniadakis (b27) 2019
Wang (10.1016/j.cma.2021.113938_b29) 2020
Reyes (10.1016/j.cma.2021.113938_b14) 2020
Cao (10.1016/j.cma.2021.113938_b32) 2019
Jin (10.1016/j.cma.2021.113938_b13) 2020
Kingma (10.1016/j.cma.2021.113938_b43) 2014
Liu (10.1016/j.cma.2021.113938_b40) 2020
Hesthaven (10.1016/j.cma.2021.113938_b46) 2007
Li (10.1016/j.cma.2021.113938_b39) 2020
Han (10.1016/j.cma.2021.113938_b23) 2018; 115
Hunter (10.1016/j.cma.2021.113938_b54) 2007; 9
Harris (10.1016/j.cma.2021.113938_b55) 2020; 585
Wang (10.1016/j.cma.2021.113938_b38) 2020
Shawe-Taylor (10.1016/j.cma.2021.113938_b44) 2005; 51
Baydin (10.1016/j.cma.2021.113938_b1) 2017; 18
Karumuri (10.1016/j.cma.2021.113938_b25) 2020; 404
Cox (10.1016/j.cma.2021.113938_b50) 2002; 176
Costabal (10.1016/j.cma.2021.113938_b15) 2020; 8
Kissas (10.1016/j.cma.2021.113938_b16) 2020; 358
Chen (10.1016/j.cma.2021.113938_b19) 2020; 28
Tancik (10.1016/j.cma.2021.113938_b41) 2020
Yang (10.1016/j.cma.2021.113938_b21) 2020
Wang (10.1016/j.cma.2021.113938_b20) 2020
Gao (10.1016/j.cma.2021.113938_b28) 2020
Yazdani (10.1016/j.cma.2021.113938_b17) 2020; 16
Zhu (10.1016/j.cma.2021.113938_b10) 2019; 394
Williams (10.1016/j.cma.2021.113938_b37) 2019
Shin (10.1016/j.cma.2021.113938_b5) 2020
Jacot (10.1016/j.cma.2021.113938_b34) 2018
Lee (10.1016/j.cma.2021.113938_b36) 2019
Patel (10.1016/j.cma.2021.113938_b30) 2020
Rahaman (10.1016/j.cma.2021.113938_b31) 2019
Yang (10.1016/j.cma.2021.113938_b26) 2019; 394
Psichogios (10.1016/j.cma.2021.113938_b2) 1992; 38
Luo (10.1016/j.cma.2021.113938_b7) 2020
Li (10.1016/j.cma.2021.113938_b9) 2020
Raissi (10.1016/j.cma.2021.113938_b12) 2020; 367
Gray (10.1016/j.cma.2021.113938_b48) 1990
Lu (10.1016/j.cma.2021.113938_b27) 2019
Lu (10.1016/j.cma.2021.113938_b42) 2021
Raissi (10.1016/j.cma.2021.113938_b24) 2018
Driscoll (10.1016/j.cma.2021.113938_b49) 2014
Kuramoto (10.1016/j.cma.2021.113938_b52) 1978; 64
Bradbury (10.1016/j.cma.2021.113938_b53) 2018
(10.1016/j.cma.2021.113938_b45) 1998
Lagaris (10.1016/j.cma.2021.113938_b3) 1998; 9
Wang (10.1016/j.cma.2021.113938_b6) 2020
Shin (10.1016/j.cma.2021.113938_b8) 2020
Raissi (10.1016/j.cma.2021.113938_b4) 2019; 378
Fang (10.1016/j.cma.2021.113938_b18) 2019; 8
Yang (10.1016/j.cma.2021.113938_b22) 2020; 476
Sivashinsky (10.1016/j.cma.2021.113938_b51) 1977; 4
10.1016/j.cma.2021.113938_b47
Arora (10.1016/j.cma.2021.113938_b35) 2019
Sun (10.1016/j.cma.2021.113938_b11) 2020; 361
Ronen (10.1016/j.cma.2021.113938_b33) 2019
References_xml – year: 2020
  ident: b9
  article-title: Fourier neural operator for parametric partial differential equations
– volume: 394
  start-page: 136
  year: 2019
  end-page: 152
  ident: b26
  article-title: Adversarial uncertainty quantification in physics-informed neural networks
  publication-title: J. Comput. Phys.
– year: 2020
  ident: b41
  article-title: Fourier features let networks learn high frequency functions in low dimensional domains
– volume: 4
  start-page: 1177
  year: 1977
  end-page: 1206
  ident: b51
  article-title: Nonlinear analysis of hydrodynamic instability in laminar flames—i. derivation of basic equations
  publication-title: AcAau
– year: 2018
  ident: b24
  article-title: Forward-backward stochastic neural networks: Deep learning of high-dimensional partial differential equations
– volume: 361
  year: 2020
  ident: b11
  article-title: Surrogate modeling for fluid flows based on physics-constrained deep learning without simulation data
  publication-title: Comput. Methods Appl. Mech. Engrg.
– year: 2020
  ident: b13
  article-title: Nsfnets (Navier-Stokes flow nets): Physics-informed neural networks for the incompressible Navier-Stokes equations
– year: 2020
  ident: b6
  article-title: When and why PINNs fail to train: A neural tangent kernel perspective
– volume: 18
  start-page: 5595
  year: 2017
  end-page: 5637
  ident: b1
  article-title: Automatic differentiation in machine learning: a survey
  publication-title: J. Mach. Learn. Res.
– year: 2020
  ident: b40
  article-title: Multi-scale deep neural network (MscaleDNN) for solving Poisson-Boltzmann equation in complex domains
– year: 2020
  ident: b28
  article-title: PhyGeoNet: Physics-informed geometry-adaptive convolutional neural networks for solving parametric PDEs on irregular domain
– volume: 38
  start-page: 1499
  year: 1992
  end-page: 1511
  ident: b2
  article-title: A hybrid neural network-first principles approach to process modeling
  publication-title: AIChE J.
– volume: 16
  year: 2020
  ident: b17
  article-title: Systems biology informed deep learning for inferring parameters and hidden dynamics
  publication-title: PLoS Comput. Biol.
– year: 1990
  ident: b48
  article-title: Chemical oscillations and instabilities: non-linear chemical kinetics
– volume: 8
  start-page: 42
  year: 2020
  ident: b15
  article-title: Physics-informed neural networks for cardiac activation mapping
  publication-title: Front. Phys.
– volume: 9
  start-page: 90
  year: 2007
  end-page: 95
  ident: b54
  article-title: Matplotlib: A 2d graphics environment
  publication-title: IEEE Ann. Hist. Comput.
– volume: 394
  start-page: 56
  year: 2019
  end-page: 81
  ident: b10
  article-title: Physics-constrained deep learning for high-dimensional surrogate modeling and uncertainty quantification without labeled data
  publication-title: J. Comput. Phys.
– volume: 51
  start-page: 2510
  year: 2005
  end-page: 2522
  ident: b44
  article-title: On the eigenspectrum of the gram matrix and the generalization error of kernel-PCA
  publication-title: IEEE Trans. Inform. Theory
– year: 2020
  ident: b39
  article-title: A DNN-based algorithm for multi-scale elliptic problems
– start-page: 4761
  year: 2019
  end-page: 4771
  ident: b33
  article-title: The convergence rate of neural networks for learned functions of different frequencies
  publication-title: Advances in Neural Information Processing Systems
– year: 2020
  ident: b38
  article-title: Multi-scale deep neural network (MscaleDNN) methods for oscillatory stokes flows in complex domains
– year: 2020
  ident: b29
  article-title: Understanding and mitigating gradient pathologies in physics-informed neural networks
– volume: 404
  year: 2020
  ident: b25
  article-title: Simulator-free solution of high-dimensional stochastic elliptic partial differential equations using deep neural networks
  publication-title: J. Comput. Phys.
– year: 2020
  ident: b14
  article-title: Learning unknown physics of non-Newtonian fluids
– volume: 8
  start-page: 24506
  year: 2019
  end-page: 24513
  ident: b18
  article-title: Deep physical informed neural networks for metamaterial design
  publication-title: IEEE Access
– year: 2014
  ident: b43
  article-title: Adam: A method for stochastic optimization
– volume: 378
  start-page: 686
  year: 2019
  end-page: 707
  ident: b4
  article-title: Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
  publication-title: J. Comput. Phys.
– year: 2007
  ident: b46
  article-title: Spectral Methods for Time-Dependent Problems, Vol. 21
– reference: Xavier Glorot, Yoshua Bengio, Understanding the difficulty of training deep feedforward neural networks, in: Proceedings of the thirteenth international conference on artificial intelligence and statistics, 2010, pp. 249–256.
– volume: 64
  start-page: 346
  year: 1978
  end-page: 367
  ident: b52
  article-title: Diffusion-induced chaos in reaction systems
  publication-title: Progr. Theoret. Phys. Suppl.
– start-page: 5301
  year: 2019
  end-page: 5310
  ident: b31
  article-title: On the spectral bias of neural networks
  publication-title: International Conference on Machine Learning
– volume: 9
  start-page: 987
  year: 1998
  end-page: 1000
  ident: b3
  article-title: Artificial neural networks for solving ordinary and partial differential equations
  publication-title: IEEE Trans. Neural Netw.
– year: 2020
  ident: b8
  article-title: Error estimates of residual minimization using neural networks for linear PDEs
– start-page: 8141
  year: 2019
  end-page: 8150
  ident: b35
  article-title: On exact computation with an infinitely wide neural net
  publication-title: Advances in Neural Information Processing Systems
– volume: 115
  start-page: 8505
  year: 2018
  end-page: 8510
  ident: b23
  article-title: Solving high-dimensional partial differential equations using deep learning
  publication-title: Proc. Natl. Acad. Sci.
– year: 2019
  ident: b32
  article-title: Towards understanding the spectral bias of deep learning
– year: 2021
  ident: b42
  article-title: Physics-informed neural networks with hard constraints for inverse design
– year: 2019
  ident: b37
  article-title: Gradient dynamics of shallow univariate relu networks
– start-page: 8572
  year: 2019
  end-page: 8583
  ident: b36
  article-title: Wide neural networks of any depth evolve as linear models under gradient descent
  publication-title: Advances in Neural Information Processing Systems
– volume: 358
  year: 2020
  ident: b16
  article-title: Machine learning in cardiovascular flows modeling: Predicting arterial blood pressure from non-invasive 4D flow MRI data using physics-informed neural networks
  publication-title: Comput. Methods Appl. Mech. Engrg.
– volume: 28
  start-page: 11618
  year: 2020
  end-page: 11633
  ident: b19
  article-title: Physics-informed neural networks for inverse problems in nano-optics and metamaterials
  publication-title: Opt. Express
– year: 2020
  ident: b30
  article-title: Thermodynamically consistent physics-informed neural networks for hyperbolic systems
– year: 1998
  ident: b45
  publication-title: Partial Differential Equations
– volume: 585
  start-page: 357
  year: 2020
  end-page: 362
  ident: b55
  article-title: Array programming with numpy
  publication-title: Nature
– volume: 476
  year: 2020
  ident: b22
  article-title: Bayesian differential programming for robust systems identification under uncertainty
  publication-title: Proc. R. Soc. A: Math. Phys. Eng. Sci.
– start-page: 8571
  year: 2018
  end-page: 8580
  ident: b34
  article-title: Neural tangent kernel: Convergence and generalization in neural networks
  publication-title: Advances in Neural Information Processing Systems
– volume: 367
  start-page: 1026
  year: 2020
  end-page: 1030
  ident: b12
  article-title: Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations
  publication-title: Science
– year: 2020
  ident: b20
  article-title: Deep learning of free boundary and Stefan problems
– year: 2020
  ident: b5
  article-title: On the convergence of physics informed neural networks for linear second-order elliptic and parabolic type PDEs
– year: 2018
  ident: b53
  article-title: JAX: composable transformations of Python+NumPy programs
– year: 2020
  ident: b21
  article-title: B-pinns: Bayesian physics-informed neural networks for forward and inverse pde problems with noisy data
– year: 2014
  ident: b49
  article-title: Chebfun guide
– year: 2020
  ident: b7
  article-title: Two-layer neural networks for partial differential equations: Optimization and generalization theory
– volume: 176
  start-page: 430
  year: 2002
  end-page: 455
  ident: b50
  article-title: Exponential time differencing for stiff systems
  publication-title: J. Comput. Phys.
– year: 2019
  ident: b27
  article-title: DeepONet: Learning nonlinear operators for identifying differential equations based on the universal approximation theorem of operators
– year: 2020
  ident: 10.1016/j.cma.2021.113938_b21
– volume: 585
  start-page: 357
  issue: 7825
  year: 2020
  ident: 10.1016/j.cma.2021.113938_b55
  article-title: Array programming with numpy
  publication-title: Nature
  doi: 10.1038/s41586-020-2649-2
– volume: 51
  start-page: 2510
  issue: 7
  year: 2005
  ident: 10.1016/j.cma.2021.113938_b44
  article-title: On the eigenspectrum of the gram matrix and the generalization error of kernel-PCA
  publication-title: IEEE Trans. Inform. Theory
  doi: 10.1109/TIT.2005.850052
– year: 2020
  ident: 10.1016/j.cma.2021.113938_b13
– volume: 38
  start-page: 1499
  issue: 10
  year: 1992
  ident: 10.1016/j.cma.2021.113938_b2
  article-title: A hybrid neural network-first principles approach to process modeling
  publication-title: AIChE J.
  doi: 10.1002/aic.690381003
– start-page: 8141
  year: 2019
  ident: 10.1016/j.cma.2021.113938_b35
  article-title: On exact computation with an infinitely wide neural net
– volume: 4
  start-page: 1177
  issue: 11
  year: 1977
  ident: 10.1016/j.cma.2021.113938_b51
  article-title: Nonlinear analysis of hydrodynamic instability in laminar flames—i. derivation of basic equations
  publication-title: AcAau
– year: 2020
  ident: 10.1016/j.cma.2021.113938_b41
– volume: 9
  start-page: 987
  issue: 5
  year: 1998
  ident: 10.1016/j.cma.2021.113938_b3
  article-title: Artificial neural networks for solving ordinary and partial differential equations
  publication-title: IEEE Trans. Neural Netw.
  doi: 10.1109/72.712178
– year: 2007
  ident: 10.1016/j.cma.2021.113938_b46
– volume: 8
  start-page: 42
  year: 2020
  ident: 10.1016/j.cma.2021.113938_b15
  article-title: Physics-informed neural networks for cardiac activation mapping
  publication-title: Front. Phys.
  doi: 10.3389/fphy.2020.00042
– volume: 378
  start-page: 686
  year: 2019
  ident: 10.1016/j.cma.2021.113938_b4
  article-title: Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
  publication-title: J. Comput. Phys.
  doi: 10.1016/j.jcp.2018.10.045
– year: 2020
  ident: 10.1016/j.cma.2021.113938_b9
– year: 2014
  ident: 10.1016/j.cma.2021.113938_b49
– year: 2020
  ident: 10.1016/j.cma.2021.113938_b6
– volume: 361
  year: 2020
  ident: 10.1016/j.cma.2021.113938_b11
  article-title: Surrogate modeling for fluid flows based on physics-constrained deep learning without simulation data
  publication-title: Comput. Methods Appl. Mech. Engrg.
  doi: 10.1016/j.cma.2019.112732
– year: 1990
  ident: 10.1016/j.cma.2021.113938_b48
– volume: 476
  issue: 2243
  year: 2020
  ident: 10.1016/j.cma.2021.113938_b22
  article-title: Bayesian differential programming for robust systems identification under uncertainty
  publication-title: Proc. R. Soc. A: Math. Phys. Eng. Sci.
  doi: 10.1098/rspa.2020.0290
– volume: 176
  start-page: 430
  issue: 2
  year: 2002
  ident: 10.1016/j.cma.2021.113938_b50
  article-title: Exponential time differencing for stiff systems
  publication-title: J. Comput. Phys.
  doi: 10.1006/jcph.2002.6995
– year: 2019
  ident: 10.1016/j.cma.2021.113938_b27
– volume: 394
  start-page: 136
  year: 2019
  ident: 10.1016/j.cma.2021.113938_b26
  article-title: Adversarial uncertainty quantification in physics-informed neural networks
  publication-title: J. Comput. Phys.
  doi: 10.1016/j.jcp.2019.05.027
– volume: 115
  start-page: 8505
  issue: 34
  year: 2018
  ident: 10.1016/j.cma.2021.113938_b23
  article-title: Solving high-dimensional partial differential equations using deep learning
  publication-title: Proc. Natl. Acad. Sci.
  doi: 10.1073/pnas.1718942115
– volume: 367
  start-page: 1026
  issue: 6481
  year: 2020
  ident: 10.1016/j.cma.2021.113938_b12
  article-title: Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations
  publication-title: Science
  doi: 10.1126/science.aaw4741
– start-page: 5301
  year: 2019
  ident: 10.1016/j.cma.2021.113938_b31
  article-title: On the spectral bias of neural networks
– year: 1998
  ident: 10.1016/j.cma.2021.113938_b45
– volume: 8
  start-page: 24506
  year: 2019
  ident: 10.1016/j.cma.2021.113938_b18
  article-title: Deep physical informed neural networks for metamaterial design
  publication-title: IEEE Access
  doi: 10.1109/ACCESS.2019.2963375
– start-page: 4761
  year: 2019
  ident: 10.1016/j.cma.2021.113938_b33
  article-title: The convergence rate of neural networks for learned functions of different frequencies
– year: 2019
  ident: 10.1016/j.cma.2021.113938_b37
– volume: 28
  start-page: 11618
  issue: 8
  year: 2020
  ident: 10.1016/j.cma.2021.113938_b19
  article-title: Physics-informed neural networks for inverse problems in nano-optics and metamaterials
  publication-title: Opt. Express
  doi: 10.1364/OE.384875
– year: 2020
  ident: 10.1016/j.cma.2021.113938_b40
– year: 2020
  ident: 10.1016/j.cma.2021.113938_b14
– year: 2020
  ident: 10.1016/j.cma.2021.113938_b20
– start-page: 8572
  year: 2019
  ident: 10.1016/j.cma.2021.113938_b36
  article-title: Wide neural networks of any depth evolve as linear models under gradient descent
– year: 2020
  ident: 10.1016/j.cma.2021.113938_b28
– volume: 404
  year: 2020
  ident: 10.1016/j.cma.2021.113938_b25
  article-title: Simulator-free solution of high-dimensional stochastic elliptic partial differential equations using deep neural networks
  publication-title: J. Comput. Phys.
  doi: 10.1016/j.jcp.2019.109120
– volume: 64
  start-page: 346
  year: 1978
  ident: 10.1016/j.cma.2021.113938_b52
  article-title: Diffusion-induced chaos in reaction systems
  publication-title: Progr. Theoret. Phys. Suppl.
  doi: 10.1143/PTPS.64.346
– volume: 394
  start-page: 56
  year: 2019
  ident: 10.1016/j.cma.2021.113938_b10
  article-title: Physics-constrained deep learning for high-dimensional surrogate modeling and uncertainty quantification without labeled data
  publication-title: J. Comput. Phys.
  doi: 10.1016/j.jcp.2019.05.024
– year: 2019
  ident: 10.1016/j.cma.2021.113938_b32
– volume: 16
  issue: 11
  year: 2020
  ident: 10.1016/j.cma.2021.113938_b17
  article-title: Systems biology informed deep learning for inferring parameters and hidden dynamics
  publication-title: PLoS Comput. Biol.
  doi: 10.1371/journal.pcbi.1007575
– ident: 10.1016/j.cma.2021.113938_b47
– year: 2018
  ident: 10.1016/j.cma.2021.113938_b24
– year: 2021
  ident: 10.1016/j.cma.2021.113938_b42
– volume: 358
  year: 2020
  ident: 10.1016/j.cma.2021.113938_b16
  article-title: Machine learning in cardiovascular flows modeling: Predicting arterial blood pressure from non-invasive 4D flow MRI data using physics-informed neural networks
  publication-title: Comput. Methods Appl. Mech. Engrg.
  doi: 10.1016/j.cma.2019.112623
– volume: 9
  start-page: 90
  issue: 03
  year: 2007
  ident: 10.1016/j.cma.2021.113938_b54
  article-title: Matplotlib: A 2d graphics environment
  publication-title: IEEE Ann. Hist. Comput.
– start-page: 8571
  year: 2018
  ident: 10.1016/j.cma.2021.113938_b34
  article-title: Neural tangent kernel: Convergence and generalization in neural networks
– year: 2020
  ident: 10.1016/j.cma.2021.113938_b5
– volume: 18
  start-page: 5595
  issue: 1
  year: 2017
  ident: 10.1016/j.cma.2021.113938_b1
  article-title: Automatic differentiation in machine learning: a survey
  publication-title: J. Mach. Learn. Res.
– year: 2014
  ident: 10.1016/j.cma.2021.113938_b43
– year: 2018
  ident: 10.1016/j.cma.2021.113938_b53
– year: 2020
  ident: 10.1016/j.cma.2021.113938_b7
– year: 2020
  ident: 10.1016/j.cma.2021.113938_b30
– year: 2020
  ident: 10.1016/j.cma.2021.113938_b29
– year: 2020
  ident: 10.1016/j.cma.2021.113938_b38
– year: 2020
  ident: 10.1016/j.cma.2021.113938_b39
– year: 2020
  ident: 10.1016/j.cma.2021.113938_b8
SSID ssj0000812
Score 2.736374
Snippet Physics-informed neural networks (PINNs) are demonstrating remarkable promise in integrating physical models with gappy and noisy observational data, but they...
SourceID osti
proquest
crossref
elsevier
SourceType Open Access Repository
Aggregation Database
Enrichment Source
Index Database
Publisher
StartPage 113938
SubjectTerms Deep learning
Eigenvectors
Inverse problems
Mathematical analysis
Neural networks
Neural Tangent Kernel
Partial differential equations
Robustness (mathematics)
Scientific machine learning
Spectral bias
Wave propagation
Title On the eigenvector bias of Fourier feature networks: From regression to solving multi-scale PDEs with physics-informed neural networks
URI https://dx.doi.org/10.1016/j.cma.2021.113938
https://www.proquest.com/docview/2562923801
https://www.osti.gov/biblio/1786215
Volume 384
WOSCitedRecordID wos000687712900004&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
journalDatabaseRights – providerCode: PRVESC
  databaseName: Elsevier SD Freedom Collection Journals 2021
  customDbUrl:
  eissn: 1879-2138
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0000812
  issn: 0045-7825
  databaseCode: AIEXJ
  dateStart: 19950101
  isFulltext: true
  titleUrlDefault: https://www.sciencedirect.com
  providerName: Elsevier
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwtV1NbxMxELWilAMc-CggQgvygRPRouxX7OVWlUSFQ5tDkXJb2c4sSimbKgmhv4DfwU9lbI-z2wIRHJCiVWTF1irzdvxmdvyGsVeAJL2CqsJHvNJRlioR6YFBZ6jSfDAzMBOJcc0mxOmpnE6LSafzI5yF2VyKupbX18XVfzU1jqGx7dHZfzD3dlEcwO9odLyi2fH6V4Y_84WLYGU2Ny4n39dz5So2xtSgrgIn59mvfQ24q4ob23MmS_jk62Jry0nxPl26wRUdRiu0JvQn70Z0IM7nRFaRl161VQTgJDzCom3aG3pHUMNqV4OriP9-AXv4OIhFQ6OP2OT6KaU9rxooh8ETVX9rTrNNAPH-2fZV9Ox4Ob-R1kjibYEc5dqIGLR9d5ZHyGf8O3Dw7lqKIkpirw8T_Hnqe84RcI9_u0_4lMXFG-O0p5LYtrYpaJmb8tuxwJjPqhjsJSIv0G_uHb0fTT80u72MvSI93V14c-5qCG8t_yfu012gO_-FDDiGc_6Q3afQhB95SD1iHaj32QMKUzhtAqt9dq-lYfmYfT-rOeKNt_DGLd74ouKEN0544wEab7lFG2_QxtcLTmjjLbRxizZu0cZvo417tG2XfMI-jkfnxycRNfeITFrk60iJJEaPP4uNEcNYD2OQsclzLYeZNvgRQ4BCaSVNNtSJ0hUGwsoIyKoMJMgkfcq69aKGZ4wbU4k0NXpWDGSWSi1BDeQswZ1TSNzRoMcG4W8vDSnf2wYsl2Uocbwo0VKltVTpLdVjr7dTrrzsy64fZ8GWJfFWz0dLhNuuaQfW7naK1Ws2trAN5xDgeuwwwKEk77IqMT5JMCJDVvl85-QDdrd5pg5Zd738Ci_YHbNZz1fLlwThn5lNz_g
linkProvider Elsevier
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=On+the+eigenvector+bias+of+Fourier+feature+networks%3A+From+regression+to+solving+multi-scale+PDEs+with+physics-informed+neural+networks&rft.jtitle=Computer+methods+in+applied+mechanics+and+engineering&rft.au=Wang%2C+Sifan&rft.au=Wang%2C+Hanwen&rft.au=Perdikaris%2C+Paris&rft.date=2021-10-01&rft.pub=Elsevier&rft.issn=0045-7825&rft.eissn=1879-2138&rft.volume=384&rft.issue=C&rft_id=info:doi/10.1016%2Fj.cma.2021.113938&rft.externalDocID=1786215
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=0045-7825&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=0045-7825&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=0045-7825&client=summon