{"id":"https://openalex.org/W7212171925","doi":"https://doi.org/10.48550/arxiv.2609.09589","title":"A Function-Space Approach to the Statistical Mechanics of Learning Dynamics","display_name":"A Function-Space Approach to the Statistical Mechanics of Learning Dynamics","publication_year":2026,"publication_date":"2026-09-09","ids":{"openalex":"https://openalex.org/W7212171925","doi":"https://doi.org/10.48550/arxiv.2609.09589"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2609.09589","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.09589","pdf_url":"https://arxiv.org/pdf/2609.09589","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2609.09589","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5151167524","display_name":"Yizhou Zhang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Yizhou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5072239003","display_name":"Weichen Wu","orcid":"https://orcid.org/0009-0005-1477-6779","observed_orcids":["https://orcid.org/0009-0005-1477-6779","https://orcid.org/0000-0002-5955-3289"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Weichen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5008387608","display_name":"Lun Du","orcid":"https://orcid.org/0000-0002-7625-0650","observed_orcids":["https://orcid.org/0000-0002-7625-0650"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Du, Lun","raw_affiliation_strings":["Independent Researcher ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Independent Researcher ,","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5087858861","display_name":"Zhengjie Miao","orcid":"https://orcid.org/0009-0008-2371-1186","observed_orcids":["https://orcid.org/0009-0008-2371-1186"]},"institutions":[{"id":"https://openalex.org/I18014758","display_name":"Simon Fraser University","ror":"https://ror.org/0213rcc28","country_code":"CA","type":"education","lineage":["https://openalex.org/I18014758"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Miao, Zhengjie","raw_affiliation_strings":["Simon Fraser University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Simon Fraser University","institution_ids":["https://openalex.org/I18014758"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10320","display_name":"Neural Networks and Applications","score":0.7720000147819519,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10320","display_name":"Neural Networks and Applications","score":0.7720000147819519,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.16539999842643738,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T12814","display_name":"Gaussian Processes and Bayesian Inference","score":0.010200000368058681,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/statistical-mechanics-of-learning","display_name":"statistical mechanics of learning","score":0.9010000228881836},{"id":"https://openalex.org/keywords/function-spaces","display_name":"function spaces","score":0.8029999732971191},{"id":"https://openalex.org/keywords/learning-dynamics","display_name":"learning dynamics","score":0.7080000042915344},{"id":"https://openalex.org/keywords/learning-operators","display_name":"learning operators","score":0.7680000066757202},{"id":"https://openalex.org/keywords/density-of-states","display_name":"density of states","score":0.7649999856948853},{"id":"https://openalex.org/keywords/boltzmann-weights","display_name":"Boltzmann weights","score":0.6790000200271606},{"id":"https://openalex.org/keywords/density-matrix","display_name":"density matrix","score":0.5889999866485596},{"id":"https://openalex.org/keywords/relu-neural-networks","display_name":"ReLU networks","score":0.6489999890327454},{"id":"https://openalex.org/keywords/kernel-methods","display_name":"kernel methods","score":0.5720000267028809},{"id":"https://openalex.org/keywords/neural-network-training-stability","display_name":"neural network training stability","score":0.37700000405311584}],"concepts":[{"id":"https://openalex.org/C121864883","wikidata":"https://www.wikidata.org/wiki/Q677916","display_name":"Statistical physics","level":1,"score":0.6801000237464905},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.5637999773025513},{"id":"https://openalex.org/C17020691","wikidata":"https://www.wikidata.org/wiki/Q139677","display_name":"Operator (biology)","level":5,"score":0.5626000165939331},{"id":"https://openalex.org/C99874945","wikidata":"https://www.wikidata.org/wiki/Q188715","display_name":"Statistical mechanics","level":2,"score":0.5616999864578247},{"id":"https://openalex.org/C158693339","wikidata":"https://www.wikidata.org/wiki/Q190524","display_name":"Eigenvalues and eigenvectors","level":2,"score":0.5471000075340271},{"id":"https://openalex.org/C195065555","wikidata":"https://www.wikidata.org/wiki/Q214881","display_name":"Curvature","level":2,"score":0.4569999873638153},{"id":"https://openalex.org/C34388435","wikidata":"https://www.wikidata.org/wiki/Q2267362","display_name":"Bounded function","level":2,"score":0.4472000002861023},{"id":"https://openalex.org/C2777952282","wikidata":"https://www.wikidata.org/wiki/Q3354606","display_name":"Transfer operator","level":2,"score":0.42089998722076416},{"id":"https://openalex.org/C79379906","wikidata":"https://www.wikidata.org/wiki/Q3174497","display_name":"Dynamical systems theory","level":2,"score":0.40880000591278076},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.38769999146461487},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.3646000027656555},{"id":"https://openalex.org/C14103023","wikidata":"https://www.wikidata.org/wiki/Q11681459","display_name":"Pairing","level":3,"score":0.35929998755455017},{"id":"https://openalex.org/C2780719617","wikidata":"https://www.wikidata.org/wiki/Q1030752","display_name":"Salient","level":2,"score":0.3483999967575073},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.3452000021934509},{"id":"https://openalex.org/C158622935","wikidata":"https://www.wikidata.org/wiki/Q660848","display_name":"Nonlinear system","level":2,"score":0.34060001373291016},{"id":"https://openalex.org/C2776029896","wikidata":"https://www.wikidata.org/wiki/Q3935810","display_name":"Relaxation (psychology)","level":2,"score":0.3366999924182892},{"id":"https://openalex.org/C197055811","wikidata":"https://www.wikidata.org/wiki/Q207522","display_name":"Probability density function","level":2,"score":0.3160000145435333},{"id":"https://openalex.org/C33962884","wikidata":"https://www.wikidata.org/wiki/Q378637","display_name":"Dynamical system (definition)","level":3,"score":0.31290000677108765},{"id":"https://openalex.org/C8272713","wikidata":"https://www.wikidata.org/wiki/Q176737","display_name":"Stochastic process","level":2,"score":0.29910001158714294},{"id":"https://openalex.org/C85725439","wikidata":"https://www.wikidata.org/wiki/Q466686","display_name":"Anisotropy","level":2,"score":0.29600000381469727},{"id":"https://openalex.org/C114289077","wikidata":"https://www.wikidata.org/wiki/Q3284399","display_name":"Statistical model","level":2,"score":0.2955000102519989},{"id":"https://openalex.org/C35304006","wikidata":"https://www.wikidata.org/wiki/Q5962","display_name":"Boltzmann constant","level":2,"score":0.29409998655319214},{"id":"https://openalex.org/C145912823","wikidata":"https://www.wikidata.org/wiki/Q113558","display_name":"Dynamics (music)","level":2,"score":0.2922999858856201},{"id":"https://openalex.org/C192576344","wikidata":"https://www.wikidata.org/wiki/Q194706","display_name":"Boltzmann machine","level":3,"score":0.288100004196167},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.28450000286102295},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.2824999988079071},{"id":"https://openalex.org/C74650414","wikidata":"https://www.wikidata.org/wiki/Q11397","display_name":"Classical mechanics","level":1,"score":0.2752000093460083},{"id":"https://openalex.org/C2778572836","wikidata":"https://www.wikidata.org/wiki/Q380933","display_name":"Space (punctuation)","level":2,"score":0.2718999981880188},{"id":"https://openalex.org/C159886148","wikidata":"https://www.wikidata.org/wiki/Q176645","display_name":"Markov process","level":2,"score":0.26809999346733093},{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.2662999927997589},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.26100000739097595},{"id":"https://openalex.org/C156778621","wikidata":"https://www.wikidata.org/wiki/Q1365748","display_name":"Spectrum (functional analysis)","level":2,"score":0.259799987077713},{"id":"https://openalex.org/C2982736386","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Statistical learning","level":2,"score":0.25200000405311584}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2609.09589","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.09589","pdf_url":"https://arxiv.org/pdf/2609.09589","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2609.09589","doi":"https://doi.org/10.48550/arxiv.2609.09589","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2609.09589","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2609.09589","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.09589","pdf_url":"https://arxiv.org/pdf/2609.09589","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7212171925.pdf","grobid_xml":"https://content.openalex.org/works/W7212171925.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4414451553","https://openalex.org/W4415873422","https://openalex.org/W3046277353","https://openalex.org/W7221798567","https://openalex.org/W2034776141","https://openalex.org/W2991552293","https://openalex.org/W7140341621","https://openalex.org/W7203504123","https://openalex.org/W2970239942","https://openalex.org/W7161737598"],"abstract_inverted_index":{"In":[0],"the":[1,35,53,68,92,103,167,184,216],"kernel":[2],"regime,":[3],"neural-network":[4],"learning":[5,44,70,217],"inherits":[6],"its":[7,95],"preferences":[8],"from":[9],"a":[10,30,76,141,192,211],"frozen":[11],"spectrum.":[12],"During":[13],"feature":[14],"learning,":[15],"this":[16,117,197],"spectrum":[17],"evolves,":[18],"yet":[19],"networks":[20],"retain":[21],"systematic":[22],"biases":[23],"toward":[24,151],"simple,":[25],"smooth":[26,204],"directions.":[27,157],"We":[28],"develop":[29],"function-space":[31,80],"statistical":[32,115],"framework":[33],"explaining":[34],"origin":[36],"of":[37,55,94,132,137,173,187],"these":[38],"preferences,":[39],"treating":[40],"functions":[41],"and":[42,139,164,195],"their":[43],"operators":[45],"as":[46,215],"macroscopic":[47],"variables,":[48],"with":[49,134,161],"parameterization":[50],"entering":[51],"through":[52],"multiplicity":[54,84],"parameter":[56,83],"configurations":[57],"realizing":[58],"each":[59],"function.":[60],"For":[61,176],"mean-squared":[62],"loss,":[63],"error":[64],"relaxes":[65],"exactly":[66,123],"under":[67],"evolving":[69],"operator":[71,88,174,218],"$M=JJ^\\ast$.":[72],"Training":[73],"stochasticity":[74],"induces":[75],"Gaussian":[77],"weight":[78],"over":[79],"states,":[81],"while":[82],"contributes":[85],"an":[86,110],"entropic":[87,181],"$B$,":[89,138],"defined":[90],"by":[91,128,200],"curvature":[93,182],"log":[96],"multiplicity.":[97],"A":[98],"local":[99,142],"Laplace":[100],"expansion":[101],"yields":[102],"fluctuation":[104],"free":[105,118],"energy":[106,119],"$\u03a6_{\\mathrm{fluc}}(M;B)=\\frac{\u03c3_\u03be^2}{2}\\log\\det(M^{-1}+B)+\\mathrm{const}$,":[107],"analogous":[108],"to":[109,183],"Occam":[111],"factor.":[112],"Under":[113],"mild":[114],"conditions,":[116],"is":[120,126,148],"rotationally":[121],"stationary":[122],"when":[124],"$[M,B]=0$,":[125],"minimized":[127],"pairing":[129],"large":[130],"eigenvalues":[131,136],"$M$":[133],"small":[135],"generates":[140],"restoring":[143],"force":[144],"against":[145],"mismatch.":[146],"Learning":[147],"therefore":[149],"biased":[150],"faster":[152],"relaxation":[153],"along":[154],"entropically":[155],"cheaper":[156],"This":[158],"preference":[159],"strengthens":[160],"training":[162],"noise":[163],"vanishes":[165],"in":[166,210],"deterministic":[168],"limit,":[169],"beyond":[170],"gradient-flow":[171],"accounts":[172],"alignment.":[175],"ReLU":[177],"networks,":[178],"we":[179],"relate":[180],"minimal":[185],"rearrangement":[186],"activation":[188],"boundaries":[189],"required":[190],"for":[191],"functional":[193],"change":[194],"bound":[196],"structural":[198],"cost":[199],"directional":[201],"smoothness.":[202],"Consequently,":[203],"directions":[205],"are":[206],"preferentially":[207],"learned":[208],"faster,":[209],"data-adaptive":[212],"manner,":[213],"even":[214],"evolves.":[219]},"counts_by_year":[],"updated_date":"2026-11-06T12:05:46.655746","created_date":"2026-09-11T00:00:00"}