{"id":"https://openalex.org/W7128399456","doi":"https://doi.org/10.48550/arxiv.2602.06079","title":"Canzona: A Unified, Asynchronous, and Load-Balanced Framework for Distributed Matrix-based Optimizers","display_name":"Canzona: A Unified, Asynchronous, and Load-Balanced Framework for Distributed Matrix-based Optimizers","publication_year":2026,"publication_date":"2026-02-04","ids":{"openalex":"https://openalex.org/W7128399456","doi":"https://doi.org/10.48550/arxiv.2602.06079"},"language":null,"primary_location":{"id":"pmh:oai:arXiv.org:2602.06079","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2602.06079","pdf_url":"https://arxiv.org/pdf/2602.06079","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","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2602.06079","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5125422939","display_name":"Liangyu Wang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Liangyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125423157","display_name":"Siqi Zhang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Siqi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125383736","display_name":"Junjie Wang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Junjie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125398854","display_name":"Yiming Dong","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dong, Yiming","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125410547","display_name":"Bo Zheng","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zheng, Bo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125427170","display_name":"Zihan Qiu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qiu, Zihan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5041102869","display_name":"Shengkun Tang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tang, Shengkun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125388048","display_name":"Di Wang","orcid":"https://orcid.org/0000-0002-0050-612X","observed_orcids":["https://orcid.org/0000-0002-0050-612X"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Di","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5004626105","display_name":"Rui Men","orcid":"https://orcid.org/0000-0002-4429-3461","observed_orcids":["https://orcid.org/0000-0002-4429-3461"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Men, Rui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5125378317","display_name":"Dayiheng Liu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Dayiheng","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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.8657000064849854,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.8657000064849854,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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":0.09200000017881393,"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/T10028","display_name":"Topic Modeling","score":0.0038999998942017555,"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/distributed-optimization","display_name":"distributed optimization","score":0.4790000021457672},{"id":"https://openalex.org/keywords/asynchronous-optimization","display_name":"asynchronous optimization","score":0.4650000035762787},{"id":"https://openalex.org/keywords/load-balancing","display_name":"load balancing","score":0.8560000061988831},{"id":"https://openalex.org/keywords/tensor-parallelism","display_name":"tensor parallelism","score":0.8109999895095825},{"id":"https://openalex.org/keywords/data-parallelism","display_name":"data parallelism","score":0.9940000176429749},{"id":"https://openalex.org/keywords/muon-optimizer","display_name":"Muon optimizer","score":0.7210000157356262},{"id":"https://openalex.org/keywords/large-language-model-training","display_name":"large language model training","score":0.6069999933242798},{"id":"https://openalex.org/keywords/qwen3","display_name":"Qwen3","score":0.7860000133514404}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8202999830245972},{"id":"https://openalex.org/C151319957","wikidata":"https://www.wikidata.org/wiki/Q752739","display_name":"Asynchronous communication","level":2,"score":0.734000027179718},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.6407999992370605},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.6111999750137329},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.5828999876976013},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.5788999795913696},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.5200999975204468},{"id":"https://openalex.org/C193702766","wikidata":"https://www.wikidata.org/wiki/Q1414548","display_name":"Concurrency","level":2,"score":0.49320000410079956},{"id":"https://openalex.org/C174348530","wikidata":"https://www.wikidata.org/wiki/Q188635","display_name":"Bridging (networking)","level":2,"score":0.47189998626708984},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.4156999886035919},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.4106000065803528},{"id":"https://openalex.org/C2989134064","wikidata":"https://www.wikidata.org/wiki/Q288510","display_name":"Execution time","level":2,"score":0.3889000117778778},{"id":"https://openalex.org/C2777303404","wikidata":"https://www.wikidata.org/wiki/Q759757","display_name":"Convergence (economics)","level":2,"score":0.37709999084472656},{"id":"https://openalex.org/C130120984","wikidata":"https://www.wikidata.org/wiki/Q2835898","display_name":"Distributed algorithm","level":2,"score":0.352400004863739},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.3411000072956085},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.3296999931335449},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.32109999656677246},{"id":"https://openalex.org/C79974875","wikidata":"https://www.wikidata.org/wiki/Q483639","display_name":"Cloud computing","level":2,"score":0.31619998812675476},{"id":"https://openalex.org/C2776834041","wikidata":"https://www.wikidata.org/wiki/Q25346349","display_name":"Execution model","level":2,"score":0.3050999939441681},{"id":"https://openalex.org/C188045909","wikidata":"https://www.wikidata.org/wiki/Q3306359","display_name":"Atomicity","level":3,"score":0.30230000615119934},{"id":"https://openalex.org/C150495011","wikidata":"https://www.wikidata.org/wiki/Q128392","display_name":"Concurrent computing","level":2,"score":0.2921999990940094},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.287200003862381},{"id":"https://openalex.org/C155281189","wikidata":"https://www.wikidata.org/wiki/Q3518150","display_name":"Tensor (intrinsic definition)","level":2,"score":0.27970001101493835},{"id":"https://openalex.org/C100776233","wikidata":"https://www.wikidata.org/wiki/Q2532492","display_name":"Bridge (graph theory)","level":2,"score":0.271699994802475},{"id":"https://openalex.org/C107568181","wikidata":"https://www.wikidata.org/wiki/Q5319000","display_name":"Dynamic priority scheduling","level":3,"score":0.2648000121116638},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.258899986743927},{"id":"https://openalex.org/C156891508","wikidata":"https://www.wikidata.org/wiki/Q1004114","display_name":"Bulk synchronous parallel","level":3,"score":0.2581999897956848},{"id":"https://openalex.org/C2984822820","wikidata":"https://www.wikidata.org/wiki/Q1123036","display_name":"Processor scheduling","level":3,"score":0.25769999623298645},{"id":"https://openalex.org/C162319229","wikidata":"https://www.wikidata.org/wiki/Q175263","display_name":"Data structure","level":2,"score":0.2563000023365021},{"id":"https://openalex.org/C60891933","wikidata":"https://www.wikidata.org/wiki/Q796575","display_name":"Ant colony","level":3,"score":0.2538999915122986},{"id":"https://openalex.org/C152124472","wikidata":"https://www.wikidata.org/wiki/Q1204361","display_name":"Redundancy (engineering)","level":2,"score":0.25110000371932983}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2602.06079","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2602.06079","pdf_url":"https://arxiv.org/pdf/2602.06079","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.2602.06079","doi":"https://doi.org/10.48550/arxiv.2602.06079","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2602.06079","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":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2602.06079","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2602.06079","pdf_url":"https://arxiv.org/pdf/2602.06079","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":[],"sustainable_development_goals_aurora":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"The":[0],"scaling":[1],"of":[2,59,144],"Large":[3],"Language":[4],"Models":[5],"(LLMs)":[6],"drives":[7],"interest":[8],"in":[9,31,152],"matrix-based":[10],"optimizers":[11],"(e.g.,":[12],"Shampoo,":[13],"Muon,":[14],"SOAP)":[15],"for":[16,23],"their":[17,21],"convergence":[18],"efficiency;":[19],"yet":[20],"requirement":[22],"holistic":[24],"updates":[25,118],"conflicts":[26],"with":[27],"the":[28,56,100,126,142,165],"tensor":[29],"fragmentation":[30],"distributed":[32],"frameworks":[33],"like":[34],"Megatron.":[35],"Existing":[36],"solutions":[37],"are":[38],"suboptimal:":[39],"synchronous":[40],"approaches":[41],"suffer":[42],"from":[43,81],"computational":[44],"redundancy,":[45],"while":[46,98],"layer-wise":[47],"partitioning":[48],"fails":[49],"to":[50,115,131,164],"reconcile":[51],"this":[52,65],"conflict":[53],"without":[54],"violating":[55],"geometric":[57],"constraints":[58],"efficient":[60],"communication":[61],"primitives.":[62],"To":[63],"bridge":[64],"gap,":[66],"we":[67,88,106],"propose":[68],"Canzona,":[69],"a":[70,149],"Unified,":[71],"Asynchronous,":[72],"and":[73,119,156],"Load-Balanced":[74],"framework":[75],"that":[76,95,138],"decouples":[77],"logical":[78],"optimizer":[79,158],"assignment":[80],"physical":[82],"parameter":[83],"distribution.":[84],"For":[85,103],"Data":[86],"Parallelism,":[87,105],"introduce":[89],"an":[90,108],"alpha-Balanced":[91],"Static":[92],"Partitioning":[93],"strategy":[94],"respects":[96],"atomicity":[97],"neutralizing":[99],"load":[101],"imbalance.":[102],"Tensor":[104],"design":[107],"Asynchronous":[109],"Compute":[110],"pipeline":[111],"utilizing":[112],"Micro-Group":[113],"Scheduling":[114],"batch":[116],"fragmented":[117],"hide":[120],"reconstruction":[121],"overhead.":[122],"Extensive":[123],"evaluations":[124],"on":[125,134],"Qwen3":[127],"model":[128],"family":[129],"(up":[130],"32B":[132],"parameters)":[133],"256":[135],"GPUs":[136],"demonstrate":[137],"our":[139],"approach":[140],"preserves":[141],"efficiency":[143],"established":[145],"parallel":[146],"architectures,":[147],"achieving":[148],"1.57x":[150],"speedup":[151],"end-to-end":[153],"iteration":[154],"time":[155],"reducing":[157],"step":[159],"latency":[160],"by":[161],"5.8x":[162],"compared":[163],"baseline.":[166]},"counts_by_year":[],"updated_date":"2026-09-25T08:18:29.962427","created_date":"2026-02-10T00:00:00"}