{"id":"https://openalex.org/W7160905822","doi":"https://doi.org/10.48550/arxiv.2605.08475","title":"Transformers Can Implement Preconditioned Richardson Iteration for In-Context Gaussian Kernel Regression","display_name":"Transformers Can Implement Preconditioned Richardson Iteration for In-Context Gaussian Kernel Regression","publication_year":2026,"publication_date":"2026-05-08","ids":{"openalex":"https://openalex.org/W7160905822","doi":"https://doi.org/10.48550/arxiv.2605.08475"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.08475","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.08475","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,"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"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.08475","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5002115537","display_name":"Mingsong Yan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yan, Mingsong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135915383","display_name":"Dongyang Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Dongyang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5087819865","display_name":"Charles Kulick","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kulick, Charles","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5135995428","display_name":"Sui Tang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tang, Sui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"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/T12814","display_name":"Gaussian Processes and Bayesian Inference","score":0.8604000210762024,"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/T12814","display_name":"Gaussian Processes and Bayesian Inference","score":0.8604000210762024,"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":0.032099999487400055,"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/T11948","display_name":"Machine Learning in Materials Science","score":0.025800000876188278,"subfield":{"id":"https://openalex.org/subfields/2505","display_name":"Materials Chemistry"},"field":{"id":"https://openalex.org/fields/25","display_name":"Materials Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/softmax-function","display_name":"Softmax function","score":0.7078999876976013},{"id":"https://openalex.org/keywords/gaussian","display_name":"Gaussian","score":0.4239000082015991},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.41920000314712524},{"id":"https://openalex.org/keywords/polynomial-regression","display_name":"Polynomial regression","score":0.41029998660087585},{"id":"https://openalex.org/keywords/nonlinear-system","display_name":"Nonlinear system","score":0.4090999960899353},{"id":"https://openalex.org/keywords/linear-regression","display_name":"Linear regression","score":0.3783999979496002},{"id":"https://openalex.org/keywords/linear-model","display_name":"Linear model","score":0.36649999022483826},{"id":"https://openalex.org/keywords/nonlinear-regression","display_name":"Nonlinear regression","score":0.3652999997138977},{"id":"https://openalex.org/keywords/gaussian-quadrature","display_name":"Gaussian quadrature","score":0.365200012922287}],"concepts":[{"id":"https://openalex.org/C188441871","wikidata":"https://www.wikidata.org/wiki/Q7554146","display_name":"Softmax function","level":3,"score":0.7078999876976013},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.6323000192642212},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.5388000011444092},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.45100000500679016},{"id":"https://openalex.org/C163716315","wikidata":"https://www.wikidata.org/wiki/Q901177","display_name":"Gaussian","level":2,"score":0.4239000082015991},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.41920000314712524},{"id":"https://openalex.org/C120068334","wikidata":"https://www.wikidata.org/wiki/Q45343","display_name":"Polynomial regression","level":3,"score":0.41029998660087585},{"id":"https://openalex.org/C158622935","wikidata":"https://www.wikidata.org/wiki/Q660848","display_name":"Nonlinear system","level":2,"score":0.4090999960899353},{"id":"https://openalex.org/C48921125","wikidata":"https://www.wikidata.org/wiki/Q10861030","display_name":"Linear regression","level":2,"score":0.3783999979496002},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.367000013589859},{"id":"https://openalex.org/C163175372","wikidata":"https://www.wikidata.org/wiki/Q3339222","display_name":"Linear model","level":2,"score":0.36649999022483826},{"id":"https://openalex.org/C46889948","wikidata":"https://www.wikidata.org/wiki/Q2755024","display_name":"Nonlinear regression","level":3,"score":0.3652999997138977},{"id":"https://openalex.org/C167196314","wikidata":"https://www.wikidata.org/wiki/Q767680","display_name":"Gaussian quadrature","level":4,"score":0.365200012922287},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.3427000045776367},{"id":"https://openalex.org/C159694833","wikidata":"https://www.wikidata.org/wiki/Q2321565","display_name":"Iterative method","level":2,"score":0.3416999876499176},{"id":"https://openalex.org/C83546350","wikidata":"https://www.wikidata.org/wiki/Q1139051","display_name":"Regression","level":2,"score":0.33799999952316284},{"id":"https://openalex.org/C61326573","wikidata":"https://www.wikidata.org/wiki/Q1496376","display_name":"Gaussian process","level":3,"score":0.32409998774528503},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.31949999928474426},{"id":"https://openalex.org/C12267149","wikidata":"https://www.wikidata.org/wiki/Q282453","display_name":"Support vector machine","level":2,"score":0.3068999946117401},{"id":"https://openalex.org/C122280245","wikidata":"https://www.wikidata.org/wiki/Q620622","display_name":"Kernel method","level":3,"score":0.29899999499320984},{"id":"https://openalex.org/C2778770139","wikidata":"https://www.wikidata.org/wiki/Q1966904","display_name":"Solver","level":2,"score":0.2978000044822693},{"id":"https://openalex.org/C200695384","wikidata":"https://www.wikidata.org/wiki/Q1739319","display_name":"Kernel regression","level":3,"score":0.2921000123023987},{"id":"https://openalex.org/C167431342","wikidata":"https://www.wikidata.org/wiki/Q1754327","display_name":"Preconditioner","level":3,"score":0.2800000011920929},{"id":"https://openalex.org/C6802819","wikidata":"https://www.wikidata.org/wiki/Q1072174","display_name":"Linear system","level":2,"score":0.27309998869895935},{"id":"https://openalex.org/C49766605","wikidata":"https://www.wikidata.org/wiki/Q207643","display_name":"Linear map","level":2,"score":0.2624000012874603},{"id":"https://openalex.org/C145242015","wikidata":"https://www.wikidata.org/wiki/Q774123","display_name":"Approximation theory","level":2,"score":0.2621000111103058},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2549999952316284},{"id":"https://openalex.org/C152877465","wikidata":"https://www.wikidata.org/wiki/Q208042","display_name":"Regression analysis","level":2,"score":0.25290000438690186},{"id":"https://openalex.org/C7218915","wikidata":"https://www.wikidata.org/wiki/Q1054475","display_name":"Gaussian function","level":3,"score":0.2515999972820282}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.08475","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.08475","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,"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":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.08475","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.08475","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,"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"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Mechanistic":[0],"accounts":[1],"of":[2,121,190],"in-context":[3,65,233],"learning":[4],"(ICL)":[5],"have":[6],"identified":[7],"iterative":[8],"algorithms":[9],"for":[10,119,142,231],"linear":[11,15,20,98,178],"regression":[12,68,168],"and":[13,33,73,111,194,216],"related":[14,31],"prediction":[16,118],"tasks,":[17],"often":[18],"using":[19],"or":[21],"ReLU":[22,146],"attention":[23,35,49,135],"variants.":[24],"For":[25],"nonlinear":[26,232],"ICL,":[27],"prior":[28],"work":[29],"has":[30],"softmax":[32,48,134],"kernelized":[34],"to":[36,151,170],"functional-gradient-type":[37],"dynamics,":[38],"but":[39],"it":[40],"remains":[41],"unclear":[42],"whether":[43],"a":[44,52,76,105,127,137,223],"standard":[45,77],"transformer":[46,79,107,132],"with":[47,55,70,108,186,203],"can":[50,80,229],"implement":[51],"convergent":[53],"solver":[54],"an":[56],"end-to-end":[57],"prediction-error":[58],"guarantee.":[59],"In":[60],"this":[61,211],"paper,":[62],"we":[63,103,162,180],"study":[64],"kernel":[66,97],"ridge":[67],"(KRR)":[69],"Gaussian":[71],"kernels":[72],"show":[74],"that":[75,115,196,226],"softmax-attention":[78,227],"approximate":[81,152],"the":[82,95,131,153,159,173,182,187],"KRR":[83,192],"predictor":[84],"during":[85],"its":[86,197],"forward":[87],"pass":[88],"by":[89,158],"implementing":[90],"preconditioned":[91,174,204,219],"Richardson":[92,175,205,220],"iteration":[93,221],"on":[94,166],"associated":[96],"system.":[99],"Under":[100],"bounded-data":[101],"assumptions,":[102],"construct":[104],"single-head":[106],"$O(\\log(1/\u03b5))$":[109],"blocks":[110],"MLP":[112,147],"width":[113],"$O(\\sqrt{N/\u03b5})$":[114],"achieves":[116],"$\u03b5$-accurate":[117],"prompts":[120],"length":[122],"$N$.":[123],"Our":[124],"construction":[125],"reveals":[126],"functional":[128],"decomposition":[129],"within":[130],"architecture:":[133],"produces":[136],"row-normalized":[138],"Gaussian-kernel":[139,234],"operator":[140],"needed":[141],"cross-token":[143],"interactions,":[144],"while":[145],"layers":[148],"act":[149],"locally":[150],"intra-token":[154],"scalar":[155],"arithmetic":[156],"required":[157],"update.":[160],"Empirically,":[161],"train":[163],"GPT-2-style":[164],"transformers":[165,228],"Gaussian-process":[167],"tasks":[169],"further":[171,209],"test":[172],"interpretation.":[176,212],"Through":[177],"probing,":[179],"compare":[181],"transformer's":[183],"layer-wise":[184],"predictions":[185],"step-wise":[188],"outputs":[189],"classical":[191],"solvers":[193],"find":[195],"error":[198],"profiles":[199],"align":[200],"most":[201],"consistently":[202],"iteration.":[206],"Ablation":[207],"studies":[208],"support":[210],"Together,":[213],"our":[214],"theory":[215],"experiments":[217],"identify":[218],"as":[222],"concrete":[224],"mechanism":[225],"realize":[230],"regression.":[235]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-05-13T00:00:00"}