{"id":"https://openalex.org/W7225640385","doi":"https://doi.org/10.5281/zenodo.24074235","title":"Exact Second-Moment Dynamics, Critical Batch Scaling, Vocabulary Phase Transitions, and the Feshbach\u2013Guyan\u2013Southwell Counterdiabatic Replacement for Adam","display_name":"Exact Second-Moment Dynamics, Critical Batch Scaling, Vocabulary Phase Transitions, and the Feshbach\u2013Guyan\u2013Southwell Counterdiabatic Replacement for Adam","publication_year":2026,"publication_date":"2026-11-09","ids":{"openalex":"https://openalex.org/W7225640385","doi":"https://doi.org/10.5281/zenodo.24074235"},"language":"en","primary_location":{"id":"doi:10.5281/zenodo.24074235","doi":"https://doi.org/10.5281/zenodo.24074235","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24074235","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.5281/zenodo.24074235","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5160983350","display_name":"Gabriel Paiva","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Paiva, Gabriel","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5160983350"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"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/T10963","display_name":"Advanced Optimization Algorithms Research","score":0.00019999999494757503,"subfield":{"id":"https://openalex.org/subfields/2612","display_name":"Numerical Analysis"},"field":{"id":"https://openalex.org/fields/26","display_name":"Mathematics"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10320","display_name":"Neural Networks and Applications","score":9.999999747378752e-05,"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/secular-equation","display_name":"secular equation","score":0.6060000061988831},{"id":"https://openalex.org/keywords/renewal-equation","display_name":"renewal equation","score":0.7730000019073486},{"id":"https://openalex.org/keywords/convergence-rate","display_name":"convergence rate","score":0.36399999260902405},{"id":"https://openalex.org/keywords/phase-transitions","display_name":"phase transitions","score":0.7080000042915344},{"id":"https://openalex.org/keywords/adam-optimizer","display_name":"Adam optimizer","score":0.800000011920929},{"id":"https://openalex.org/keywords/squared-error-loss","display_name":"squared error loss","score":0.4869999885559082}],"concepts":[{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.7921000123023987},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.5263000130653381},{"id":"https://openalex.org/C130367717","wikidata":"https://www.wikidata.org/wiki/Q189791","display_name":"Diagonal","level":2,"score":0.4690999984741211},{"id":"https://openalex.org/C129844170","wikidata":"https://www.wikidata.org/wiki/Q41299","display_name":"Quadratic equation","level":2,"score":0.44690001010894775},{"id":"https://openalex.org/C17020691","wikidata":"https://www.wikidata.org/wiki/Q139677","display_name":"Operator (biology)","level":5,"score":0.4345000088214874},{"id":"https://openalex.org/C179254644","wikidata":"https://www.wikidata.org/wiki/Q13222844","display_name":"Moment (physics)","level":2,"score":0.41200000047683716},{"id":"https://openalex.org/C151376022","wikidata":"https://www.wikidata.org/wiki/Q168698","display_name":"Exponential function","level":2,"score":0.39890000224113464},{"id":"https://openalex.org/C195065555","wikidata":"https://www.wikidata.org/wiki/Q214881","display_name":"Curvature","level":2,"score":0.38690000772476196},{"id":"https://openalex.org/C45347329","wikidata":"https://www.wikidata.org/wiki/Q5166604","display_name":"Convolution (computer science)","level":3,"score":0.3856000006198883},{"id":"https://openalex.org/C33676613","wikidata":"https://www.wikidata.org/wiki/Q13415176","display_name":"Dimension (graph theory)","level":2,"score":0.38429999351501465},{"id":"https://openalex.org/C180188523","wikidata":"https://www.wikidata.org/wiki/Q574576","display_name":"Exponential decay","level":2,"score":0.3580000102519989},{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.3546999990940094},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.3472999930381775},{"id":"https://openalex.org/C41949839","wikidata":"https://www.wikidata.org/wiki/Q1756677","display_name":"Stability theory","level":3,"score":0.3433000147342682},{"id":"https://openalex.org/C60718061","wikidata":"https://www.wikidata.org/wiki/Q1414747","display_name":"Momentum (technical analysis)","level":2,"score":0.3249000012874603},{"id":"https://openalex.org/C140532419","wikidata":"https://www.wikidata.org/wiki/Q249748","display_name":"Spectral radius","level":3,"score":0.3248000144958496},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.3248000144958496},{"id":"https://openalex.org/C11577676","wikidata":"https://www.wikidata.org/wiki/Q134237","display_name":"Square root","level":2,"score":0.320499986410141},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.3109000027179718},{"id":"https://openalex.org/C64812099","wikidata":"https://www.wikidata.org/wiki/Q176604","display_name":"Random matrix","level":3,"score":0.30979999899864197},{"id":"https://openalex.org/C158693339","wikidata":"https://www.wikidata.org/wiki/Q190524","display_name":"Eigenvalues and eigenvectors","level":2,"score":0.3057999908924103},{"id":"https://openalex.org/C61062188","wikidata":"https://www.wikidata.org/wiki/Q835065","display_name":"Second moment of area","level":2,"score":0.30559998750686646},{"id":"https://openalex.org/C202444582","wikidata":"https://www.wikidata.org/wiki/Q837863","display_name":"Pure mathematics","level":1,"score":0.30079999566078186},{"id":"https://openalex.org/C13355873","wikidata":"https://www.wikidata.org/wiki/Q2920850","display_name":"Connection (principal bundle)","level":2,"score":0.29420000314712524},{"id":"https://openalex.org/C114614502","wikidata":"https://www.wikidata.org/wiki/Q76592","display_name":"Combinatorics","level":1,"score":0.28999999165534973},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.289000004529953},{"id":"https://openalex.org/C204575570","wikidata":"https://www.wikidata.org/wiki/Q207476","display_name":"Holomorphic function","level":2,"score":0.2842999994754791},{"id":"https://openalex.org/C39927690","wikidata":"https://www.wikidata.org/wiki/Q11197","display_name":"Logarithm","level":2,"score":0.2793999910354614},{"id":"https://openalex.org/C44280652","wikidata":"https://www.wikidata.org/wiki/Q104837","display_name":"Phase (matter)","level":2,"score":0.27489998936653137},{"id":"https://openalex.org/C151342819","wikidata":"https://www.wikidata.org/wiki/Q62542","display_name":"Phase space","level":2,"score":0.2720000147819519},{"id":"https://openalex.org/C156778621","wikidata":"https://www.wikidata.org/wiki/Q1365748","display_name":"Spectrum (functional analysis)","level":2,"score":0.27000001072883606},{"id":"https://openalex.org/C202787564","wikidata":"https://www.wikidata.org/wiki/Q6510488","display_name":"Heat equation","level":2,"score":0.2574000060558319},{"id":"https://openalex.org/C118615104","wikidata":"https://www.wikidata.org/wiki/Q121416","display_name":"Discrete mathematics","level":1,"score":0.2565999925136566},{"id":"https://openalex.org/C162392398","wikidata":"https://www.wikidata.org/wiki/Q272404","display_name":"Finite set","level":2,"score":0.2522999942302704}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.5281/zenodo.24074235","doi":"https://doi.org/10.5281/zenodo.24074235","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24074235","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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"},{"id":"doi:10.5281/zenodo.24074237","doi":"https://doi.org/10.5281/zenodo.24074237","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24074237","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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":"doi:10.5281/zenodo.24074235","doi":"https://doi.org/10.5281/zenodo.24074235","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24074235","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7225290145","https://openalex.org/W7225655361","https://openalex.org/W7133499639","https://openalex.org/W7203725111","https://openalex.org/W4283378363","https://openalex.org/W4380715436","https://openalex.org/W7162773681","https://openalex.org/W7133190866","https://openalex.org/W4367553283","https://openalex.org/W7079467925"],"abstract_inverted_index":{"For":[0],"heavy-ball":[1],"SGD":[2,471],"on":[3],"a":[4,51,99,156,222,262,395,585],"quadratic":[5,569],"loss":[6,108,151],"in":[7,38],"R^P":[8],"with":[9,43,144,161,176,209,335,384,423],"minibatch":[10],"Hessians":[11],"H_{B_t}":[12],"of":[13,41,55,81,98,135,235,464,490,552,587],"mean":[14],"H":[15,42],"\u227b":[16],"0":[17,344,373,412],"at":[18,87,173,245,294,363,472],"finite":[19,23,246],"dimension":[20],"P":[21],"and":[22,50,124,167,216,237,251,269,321,341,405,413,434,450,478,502,513,523,528,532,543,579],"batch":[24,499],"size":[25,90],"B,":[26,248],"we":[27,192,240,273,368],"analyze":[28],"the":[29,39,65,69,73,78,92,96,106,131,141,150,174,229,232,253,257,275,299,325,346,458,465,488,495,511,577],"exact":[30,111,276,358],"second-moment":[31,74],"dynamics":[32],"when":[33,118,218,399],"features":[34],"have":[35],"independent":[36,548],"coordinates":[37],"eigenbasis":[40],"mode-dependent":[44,62],"fourth":[45,56],"moments":[46],"3":[47,353],"+":[48,120,146,315,329],"\u03ba_k":[49],"common":[52],"random":[53],"scale":[54],"moment":[57],"\u03b3":[58],"\u2265":[59,122,148,352],"1.":[60,139],"Although":[61],"kurtoses":[63],"place":[64],"curvature":[66],"noise":[67,184,409],"outside":[68],"spectrally":[70],"expressible":[71],"class,":[72],"operator":[75],"T":[76],"has":[77],"spectral":[79],"radius":[80],"its":[82,153],"restriction":[83],"to":[84,231,255,311,394,526,547],"diagonal":[85],"blocks":[86],"every":[88],"step":[89],"\u03b7,":[91,249],"stability":[93],"threshold":[94,175],"is":[95,130,457,583],"root":[97,134],"rank-one":[100],"secular":[101],"equation":[102,114,117],"S(\u03b7)":[103,186],"=":[104,126,138,187,285,296,313,333,339,343,372,417,430],"1,":[105,149],"expected":[107],"obeys":[109],"an":[110,162,177,210,357],"discrete":[112],"convolution":[113],"(a":[115],"renewal":[116,179],"\u03b3(2":[119,145],"\u03ba_k)":[121,147],"1),":[123,191],"\u03c1(T)":[125,170],"1/z\u22c6":[127],"where":[128],"z\u22c6":[129],"smallest":[132],"positive":[133],"z":[136],"K\u0302(z)":[137],"On":[140],"stable":[142],"side":[143],"approaches":[152],"floor":[154,185,398],"as":[155,205,519],"single":[157],"exponential":[158],"C":[159,164],"\u03c1(T)^t":[160],"explicit":[163,178,211],">":[165,318,411],"0,":[166],"1":[168,201,208,334,402,427],"\u2212":[169,196,198,202,213,403,428],"vanishes":[171],"linearly":[172],"slope.":[180],"Along":[181],"any":[182],"fixed":[183],"s":[188],"\u2208":[189,289],"(0,":[190],"prove":[193,369],"that":[194,306,323,370,491],"(1":[195],"\u03c1(T))/(1":[197],"\u03b2)":[199,214,250],"\u2192":[200],"\u03b7_EMA(1;":[203],"s)/\u03b7\u22c6_EMA(1)":[204],"\u03b2":[206,342,371],"\u2191":[207],"O(1":[212],"correction,":[215],"characterize":[217],"anisotropic":[219,414,508],"spectra":[220],"yield":[221],"momentum":[223,310],"speedup":[224],"strictly":[225],"above":[226],"2.":[227],"Applying":[228],"theory":[230,254,497],"vocabulary":[233],"model":[234],"Everett":[236],"Paquette":[238],"(2026),":[239],"settle":[241],"their":[242,580],"open":[243],"problems":[244],"(d,":[247],"extend":[252],"construct":[256],"Feshbach\u2013Guyan\u2013Southwell":[258],"(FGS)":[259],"Counterdiabatic":[260],"Optimizer,":[261],"provably":[263],"optimal":[264],"replacement":[265],"for":[266,280,521,567],"Adam":[267,293,307,522],"(Kingma":[268],"Ba,":[270],"2015):":[271],"(i)":[272],"solve":[274],"minimax":[277],"envelope":[278],"W_\u03b8(B\u0306)":[279],"all":[281],"power-law":[282],"preconditioners":[283],"\u03b7_j":[284],"\u03b1":[286],"q_j^(\u2212\u03b8)":[287],"(\u03b8":[288],"[0,":[290],"1],":[291],"including":[292],"\u03b8":[295,332],"1/2),":[297],"proving":[298,322],"Momentum\u2013Preconditioning":[300],"Duality":[301],"S_{\u03b8,mom}(\u221e)":[302],"\u2261":[303],"S_{(1+\u03b8)/2,SGD}(\u221e),":[304],"showing":[305],"delays":[308],"beneficial":[309],"B\u00d7(1/2)":[312],"2(\u221a\u03ba_V":[314],"1)":[316],"B\u00d7(0)":[317],"4":[319,431],"B\u00d7(0),":[320],"within":[324],"damped":[326],"family":[327],"\u03b1(q_j":[328],"\u03b5\u00b2)^(\u2212\u03b8)":[330],"only":[331],"mobility-edge":[336],"damping":[337,426],"\u03b5\u22c6\u00b2(B)":[338],"(d+2)/(B\u22121)":[340],"achieves":[345],"Mandelstam\u2013Tamm":[347],"speed":[348],"limit":[349],"across":[350],"V":[351],"tokens;":[354],"(ii)":[355],"via":[356],"Feshbach\u2013Poincar\u00e9":[359],"stroboscopic":[360],"return":[361],"map":[362],"active":[364],"token":[365,509],"arrival":[366],"times,":[367],"(Guyan":[374],"static":[375],"condensation)":[376],"contracts":[377],"rare":[378],"rows":[379],"by":[380,443],"(d+1)/(d+2)":[381],"per":[382],"hit":[383],"zero":[385],"silent-step":[386],"memory":[387],"traffic,":[388],"whereas":[389],"Adam's":[390],"finite-\u03b2\u2082":[391],"accumulator":[392],"collapses":[393],"non-contracting":[396],"SignSGD":[397],"p_q":[400],"\u226a":[401],"\u03b2\u2082;":[404],"(iii)":[406],"under":[407],"label":[408],"\u03c3\u00b2":[410],"embeddings":[415],"H_j":[416],"q_j":[418,432],"H\u0303_j,":[419],"Equal-Rate":[420],"FGS":[421,512,515],"water-filling":[422],"co-scaled":[424],"critical":[425],"\u03b2\u22c6_j":[429],"\u03b7_{EMA,j}":[433],"HVP-free":[435,535],"online":[436],"Southwell":[437],"trace":[438],"estimation":[439],"reduces":[440],"sample":[441],"complexity":[442],"1/(V":[444],"q_min)":[445,452],"over":[446,453,555],"global":[447,454],"Heavy":[448],"Ball":[449],"2/(V":[451],"SGD.":[455],"This":[456,573],"full,":[459],"extended":[460],"version":[461],"(54":[462],"pages)":[463],"author's":[466],"earlier":[467],"preprint":[468],"\"Minibatch":[469],"Heavy-Ball":[470],"Finite":[473],"Dimension:":[474],"Exact":[475],"Threshold,":[476],"Rate":[477],"Critical":[479],"Slowing":[480],"Down\"":[481],"(Zenodo,":[482],"5":[483],"November":[484],"2026,":[485],"https://zenodo.org/records/23972672).":[486],"Beyond":[487],"results":[489],"note,":[492],"it":[493,582],"adds":[494],"vocabulary-model":[496],"(critical":[498],"size,":[500],"bifurcation":[501],"bend":[503],"scale,":[504],"label-noise":[505],"phase":[506],"transition,":[507],"embeddings),":[510],"Universal":[514],"(UFGS)":[516],"counterdiabatic":[517],"optimizers":[518],"replacements":[520],"AdamW,":[524],"extensions":[525],"logistic":[527],"softmax":[529],"cross-entropy":[530],"losses,":[531],"implementable":[533],"optimum-free,":[534],"estimators.":[536],"Preliminary":[537],"release,":[538],"not":[539,561,584],"peer":[540],"reviewed.":[541],"Results":[542],"proofs":[544],"remain":[545],"subject":[546],"mathematical":[549],"verification.":[550],"Claims":[551],"general":[553],"superiority":[554],"practical":[556],"AdamW":[557],"or":[558,570,591],"Muon":[559],"should":[560],"be":[562],"inferred":[563],"from":[564],"guarantees":[565],"proved":[566],"restricted":[568],"predictable-preconditioner":[571],"models.":[572],"public":[574],"draft":[575],"documents":[576],"ideas":[578],"development;":[581],"certificate":[586],"theorem":[588],"correctness,":[589],"novelty,":[590],"conference":[592],"acceptance.":[593]},"counts_by_year":[],"updated_date":"2026-11-11T12:16:26.678958","created_date":"2026-11-10T00:00:00"}
