{"id":"https://openalex.org/W7213517318","doi":"https://doi.org/10.48550/arxiv.2609.18314","title":"Beyond Quadratic Loss: The Stability Phase Diagram of Adam","display_name":"Beyond Quadratic Loss: The Stability Phase Diagram of Adam","publication_year":2026,"publication_date":"2026-09-16","ids":{"openalex":"https://openalex.org/W7213517318","doi":"https://doi.org/10.48550/arxiv.2609.18314"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2609.18314","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.18314","pdf_url":"https://arxiv.org/pdf/2609.18314","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/2609.18314","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5133481282","display_name":"Gaoxiang Tang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Tang, Gaoxiang","raw_affiliation_strings":["IIIS , Tsinghua University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IIIS , Tsinghua University","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5074300516","display_name":"Huanran Chen","orcid":"https://orcid.org/0000-0003-0847-9485","observed_orcids":["https://orcid.org/0000-0003-0847-9485"]},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chen, Huanran","raw_affiliation_strings":["College AI , Tsinghua University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"College AI , Tsinghua University","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5101924541","display_name":"Ziming Liu","orcid":"https://orcid.org/0000-0003-3581-7494","observed_orcids":["https://orcid.org/0000-0003-3581-7494","https://orcid.org/0000-0002-7090-8853"]},"institutions":[{"id":"https://openalex.org/I4405259049","display_name":"Shanghai Qi Zhi Institute","ror":"https://ror.org/01n14yn69","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4405259049"]},{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Liu, Ziming","raw_affiliation_strings":["College AI , Tsinghua University Shanghai Qizhi Institute MetaCircle"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"College AI , Tsinghua University Shanghai Qizhi Institute MetaCircle","institution_ids":["https://openalex.org/I4405259049","https://openalex.org/I99065089"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9984999895095825,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9984999895095825,"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/T10320","display_name":"Neural Networks and Applications","score":0.000699999975040555,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.00019999999494757503,"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/adam-optimizer","display_name":"Adam optimizer","score":0.996999979019165},{"id":"https://openalex.org/keywords/stability-phase-diagram","display_name":"stability phase diagram","score":0.8450000286102295},{"id":"https://openalex.org/keywords/loss-landscape","display_name":"loss landscape","score":0.5490000247955322},{"id":"https://openalex.org/keywords/cross-entropy-loss","display_name":"cross-entropy loss","score":0.8569999933242798},{"id":"https://openalex.org/keywords/squared-error-loss","display_name":"squared error loss","score":0.6679999828338623},{"id":"https://openalex.org/keywords/neural-network-training-dynamics","display_name":"neural network training dynamics","score":0.6919999718666077}],"concepts":[{"id":"https://openalex.org/C129844170","wikidata":"https://www.wikidata.org/wiki/Q41299","display_name":"Quadratic equation","level":2,"score":0.7980999946594238},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.6651999950408936},{"id":"https://openalex.org/C2780388253","wikidata":"https://www.wikidata.org/wiki/Q5421508","display_name":"Exponent","level":2,"score":0.6226999759674072},{"id":"https://openalex.org/C60718061","wikidata":"https://www.wikidata.org/wiki/Q1414747","display_name":"Momentum (technical analysis)","level":2,"score":0.6062999963760376},{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.5608000159263611},{"id":"https://openalex.org/C62354387","wikidata":"https://www.wikidata.org/wiki/Q875399","display_name":"Boundary (topology)","level":2,"score":0.5044000148773193},{"id":"https://openalex.org/C121864883","wikidata":"https://www.wikidata.org/wiki/Q677916","display_name":"Statistical physics","level":1,"score":0.5031999945640564},{"id":"https://openalex.org/C85906118","wikidata":"https://www.wikidata.org/wiki/Q186693","display_name":"Phase diagram","level":3,"score":0.4950999915599823},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.46230000257492065},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.42899999022483826},{"id":"https://openalex.org/C207821765","wikidata":"https://www.wikidata.org/wiki/Q405372","display_name":"Instability","level":2,"score":0.4278999865055084},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.42570000886917114},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.4226999878883362},{"id":"https://openalex.org/C44280652","wikidata":"https://www.wikidata.org/wiki/Q104837","display_name":"Phase (matter)","level":2,"score":0.39730000495910645},{"id":"https://openalex.org/C182310444","wikidata":"https://www.wikidata.org/wiki/Q1332643","display_name":"Boundary value problem","level":2,"score":0.35269999504089355},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.35010001063346863},{"id":"https://openalex.org/C186399060","wikidata":"https://www.wikidata.org/wiki/Q959962","display_name":"Diagram","level":2,"score":0.3402999937534332},{"id":"https://openalex.org/C135889238","wikidata":"https://www.wikidata.org/wiki/Q46276","display_name":"Kinetic energy","level":2,"score":0.3230000138282776},{"id":"https://openalex.org/C2985627981","wikidata":"https://www.wikidata.org/wiki/Q828827","display_name":"Power loss","level":3,"score":0.3010999858379364},{"id":"https://openalex.org/C2164484","wikidata":"https://www.wikidata.org/wiki/Q5170150","display_name":"Core (optical fiber)","level":2,"score":0.2946999967098236},{"id":"https://openalex.org/C75438885","wikidata":"https://www.wikidata.org/wiki/Q3403615","display_name":"Large deviations theory","level":2,"score":0.28040000796318054},{"id":"https://openalex.org/C2776799497","wikidata":"https://www.wikidata.org/wiki/Q484298","display_name":"Surface (topology)","level":2,"score":0.27730000019073486},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.2743000090122223},{"id":"https://openalex.org/C188324986","wikidata":"https://www.wikidata.org/wiki/Q7180938","display_name":"Phase boundary","level":3,"score":0.2597000002861023},{"id":"https://openalex.org/C166437778","wikidata":"https://www.wikidata.org/wiki/Q50695","display_name":"Quadratic function","level":3,"score":0.2549999952316284}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2609.18314","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.18314","pdf_url":"https://arxiv.org/pdf/2609.18314","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.2609.18314","doi":"https://doi.org/10.48550/arxiv.2609.18314","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2609.18314","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2609.18314","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.18314","pdf_url":"https://arxiv.org/pdf/2609.18314","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":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7213517318.pdf","grobid_xml":"https://content.openalex.org/works/W7213517318.grobid-xml"},"referenced_works_count":23,"referenced_works":["https://openalex.org/W1686810756","https://openalex.org/W2000483484","https://openalex.org/W2112796928","https://openalex.org/W2130984546","https://openalex.org/W2626325961","https://openalex.org/W2909489335","https://openalex.org/W2912811302","https://openalex.org/W2962933129","https://openalex.org/W2964121744","https://openalex.org/W2972960568","https://openalex.org/W4226434736","https://openalex.org/W4283382874","https://openalex.org/W4289446290","https://openalex.org/W4366731970","https://openalex.org/W4403964581","https://openalex.org/W4404348772","https://openalex.org/W4406337656","https://openalex.org/W4415096760","https://openalex.org/W4415796246","https://openalex.org/W7134918563","https://openalex.org/W7160633226","https://openalex.org/W7170714020","https://openalex.org/W7204101404"],"related_works":["https://openalex.org/W4416133862","https://openalex.org/W7212715725","https://openalex.org/W4410932063","https://openalex.org/W3086677109","https://openalex.org/W4283382874","https://openalex.org/W7214910994","https://openalex.org/W7204775134","https://openalex.org/W4416102054","https://openalex.org/W7079467925","https://openalex.org/W7204101404"],"abstract_inverted_index":{"Loss":[0],"spikes":[1,20,133],"are":[2],"recurrent":[3],"instabilities":[4],"in":[5,16],"neural-network":[6],"training":[7,43],"and":[8,83,141],"can":[9],"arise":[10],"from":[11,62],"multiple":[12],"mechanisms.":[13],"For":[14],"Adam":[15,131],"particular,":[17],"macroscopic":[18],"loss":[19,69,77,91,132,144],"have":[21],"been":[22],"linked":[23],"to":[24,88,134],"optimizer":[25,125],"dynamics,":[26,64],"yet":[27],"how":[28],"its":[29],"two":[30],"momentum":[31,139],"timescales":[32,140],"govern":[33],"them":[34],"remains":[35],"unclear.":[36],"We":[37,94],"investigate":[38],"this":[39],"dependence":[40],"by":[41,111],"mapping":[42],"dynamics":[44],"across":[45],"the":[46,80,85,89,121,136,147],"$(\u03b2_1,\u03b2_2)$":[47],"plane.":[48],"Across":[49],"a":[50,66,102,106,112],"range":[51],"of":[52,123],"model--task":[53],"settings,":[54],"an":[55,124],"approximately":[56,71],"linear":[57],"boundary,":[58],"$1-\u03b2_2=C(1-\u03b2_1)$,":[59],"separates":[60],"spiky":[61],"non-spiky":[63],"whereas":[65],"one-dimensional":[67,75],"quadratic":[68,108],"produces":[70,116],"cubic":[72],"slope.":[73],"A":[74],"superquadratic":[76,118,143],"$L(x)\\propto|x|^n$":[78],"recovers":[79],"near-linear":[81],"scaling":[82],"links":[84],"boundary":[86],"coefficient":[87],"effective":[90,117],"exponent":[92],"$n$.":[93],"further":[95],"show":[96],"that":[97],"confident":[98],"cross-entropy":[99],"losses":[100],"develop":[101],"core--wall":[103],"landscape":[104],"comprising":[105],"narrow":[107],"core":[109],"followed":[110],"steep":[113],"wall,":[114],"which":[115],"behavior":[119],"at":[120],"scale":[122],"update.":[126],"Together,":[127],"these":[128],"results":[129],"connect":[130],"both":[135],"mismatch":[137],"between":[138],"finite-scale":[142],"geometry":[145],"beyond":[146],"Hessian.":[148]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-09-18T00:00:00"}
