{"id":"https://openalex.org/W7148554124","doi":"https://doi.org/10.48550/arxiv.2604.00344","title":"Agent Q-Mix: Selecting the Right Action for LLM Multi-Agent Systems through Reinforcement Learning","display_name":"Agent Q-Mix: Selecting the Right Action for LLM Multi-Agent Systems through Reinforcement Learning","publication_year":2026,"publication_date":"2026-04-01","ids":{"openalex":"https://openalex.org/W7148554124","doi":"https://doi.org/10.48550/arxiv.2604.00344"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2604.00344","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2604.00344","pdf_url":"https://arxiv.org/pdf/2604.00344","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":"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/2604.00344","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5111089095","display_name":"Eric Hanchen Jiang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I161318765","display_name":"University of California, Los Angeles","ror":"https://ror.org/046rm7j60","country_code":"US","type":"education","lineage":["https://openalex.org/I161318765"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jiang, Eric Hanchen","raw_affiliation_strings":["University of California Los Angeles ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California Los Angeles ,","institution_ids":["https://openalex.org/I161318765"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132755230","display_name":"Levina Li","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I161318765","display_name":"University of California, Los Angeles","ror":"https://ror.org/046rm7j60","country_code":"US","type":"education","lineage":["https://openalex.org/I161318765"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Li, Levina","raw_affiliation_strings":["University of California Los Angeles ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California Los Angeles ,","institution_ids":["https://openalex.org/I161318765"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5123522690","display_name":"Rui Sun","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I161318765","display_name":"University of California, Los Angeles","ror":"https://ror.org/046rm7j60","country_code":"US","type":"education","lineage":["https://openalex.org/I161318765"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Sun, Rui","raw_affiliation_strings":["University of California Los Angeles ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California Los Angeles ,","institution_ids":["https://openalex.org/I161318765"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132801931","display_name":"Xiao Liang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I161318765","display_name":"University of California, Los Angeles","ror":"https://ror.org/046rm7j60","country_code":"US","type":"education","lineage":["https://openalex.org/I161318765"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Liang, Xiao","raw_affiliation_strings":["University of California Los Angeles ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California Los Angeles ,","institution_ids":["https://openalex.org/I161318765"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132786940","display_name":"Yubei Li","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I161318765","display_name":"University of California, Los Angeles","ror":"https://ror.org/046rm7j60","country_code":"US","type":"education","lineage":["https://openalex.org/I161318765"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Li, Yubei","raw_affiliation_strings":["University of California Los Angeles ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California Los Angeles ,","institution_ids":["https://openalex.org/I161318765"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125727862","display_name":"Yuchen Wu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I161318765","display_name":"University of California, Los Angeles","ror":"https://ror.org/046rm7j60","country_code":"US","type":"education","lineage":["https://openalex.org/I161318765"]},{"id":"https://openalex.org/I201448701","display_name":"University of Washington","ror":"https://ror.org/00cvxb145","country_code":"US","type":"education","lineage":["https://openalex.org/I201448701"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Wu, Yuchen","raw_affiliation_strings":["University of California Los Angeles ,","University of Washington ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California Los Angeles ,","institution_ids":["https://openalex.org/I161318765"]},{"raw_affiliation_string":"University of Washington ,","institution_ids":["https://openalex.org/I201448701"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5085445569","display_name":"Haozheng Luo","orcid":"https://orcid.org/0000-0001-9826-2874","observed_orcids":["https://orcid.org/0000-0001-9826-2874"]},"institutions":[{"id":"https://openalex.org/I111979921","display_name":"Northwestern University","ror":"https://ror.org/000e0be47","country_code":"US","type":"education","lineage":["https://openalex.org/I111979921"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Luo, Haozheng","raw_affiliation_strings":["Northwestern University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Northwestern University","institution_ids":["https://openalex.org/I111979921"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132811780","display_name":"Hengli Li","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Hengli","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132793734","display_name":"Zhi Zhang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I161318765","display_name":"University of California, Los Angeles","ror":"https://ror.org/046rm7j60","country_code":"US","type":"education","lineage":["https://openalex.org/I161318765"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Zhang, Zhi","raw_affiliation_strings":["University of California Los Angeles ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California Los Angeles ,","institution_ids":["https://openalex.org/I161318765"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5111134948","display_name":"Zhaolu Kang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kang, Zhaolu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132789862","display_name":"Kai-Wei Chang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I161318765","display_name":"University of California, Los Angeles","ror":"https://ror.org/046rm7j60","country_code":"US","type":"education","lineage":["https://openalex.org/I161318765"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Chang, Kai-Wei","raw_affiliation_strings":["University of California Los Angeles ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California Los Angeles ,","institution_ids":["https://openalex.org/I161318765"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5132789391","display_name":"Ying Nian Wu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I161318765","display_name":"University of California, Los Angeles","ror":"https://ror.org/046rm7j60","country_code":"US","type":"education","lineage":["https://openalex.org/I161318765"]},{"id":"https://openalex.org/I201448701","display_name":"University of Washington","ror":"https://ror.org/00cvxb145","country_code":"US","type":"education","lineage":["https://openalex.org/I201448701"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Wu, Ying Nian","raw_affiliation_strings":["University of California Los Angeles ,","University of Washington ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California Los Angeles ,","institution_ids":["https://openalex.org/I161318765"]},{"raw_affiliation_string":"University of Washington ,","institution_ids":["https://openalex.org/I201448701"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"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/T10456","display_name":"Multi-Agent Systems and Negotiation","score":0.4154999852180481,"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/T10456","display_name":"Multi-Agent Systems and Negotiation","score":0.4154999852180481,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.361299991607666,"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/T10906","display_name":"AI-based Problem Solving and Planning","score":0.03669999912381172,"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/cooperative-multi-agent-reinforcement-learning","display_name":"cooperative multi-agent reinforcement learning","score":0.8560000061988831},{"id":"https://openalex.org/keywords/decentralized-communication","display_name":"decentralized communication","score":0.48500001430511475},{"id":"https://openalex.org/keywords/multi-agent-reasoning","display_name":"multi-agent reasoning","score":0.7799999713897705},{"id":"https://openalex.org/keywords/centralized-training-with-decentralized-execution","display_name":"centralized training decentralized execution","score":0.5920000076293945}],"concepts":[{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.7811999917030334},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7472000122070312},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.6050999760627747},{"id":"https://openalex.org/C166109690","wikidata":"https://www.wikidata.org/wiki/Q4677422","display_name":"Action selection","level":3,"score":0.6047999858856201},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.5860000252723694},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.531499981880188},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.5187000036239624},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.49630001187324524},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.45080000162124634},{"id":"https://openalex.org/C41550386","wikidata":"https://www.wikidata.org/wiki/Q529909","display_name":"Multi-agent system","level":2,"score":0.4309999942779541},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.42289999127388},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.4065999984741211},{"id":"https://openalex.org/C199845137","wikidata":"https://www.wikidata.org/wiki/Q145490","display_name":"Network topology","level":2,"score":0.36410000920295715},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.34709998965263367},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.33489999175071716},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.3343999981880188},{"id":"https://openalex.org/C2164484","wikidata":"https://www.wikidata.org/wiki/Q5170150","display_name":"Core (optical fiber)","level":2,"score":0.33000001311302185},{"id":"https://openalex.org/C205875254","wikidata":"https://www.wikidata.org/wiki/Q17156857","display_name":"Decentralised system","level":3,"score":0.29280000925064087},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.28929999470710754},{"id":"https://openalex.org/C67203356","wikidata":"https://www.wikidata.org/wiki/Q1321905","display_name":"Reinforcement","level":2,"score":0.27549999952316284},{"id":"https://openalex.org/C74072328","wikidata":"https://www.wikidata.org/wiki/Q1142726","display_name":"Intelligent agent","level":2,"score":0.2680000066757202},{"id":"https://openalex.org/C14646407","wikidata":"https://www.wikidata.org/wiki/Q1430750","display_name":"Bellman equation","level":2,"score":0.25189998745918274}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2604.00344","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2604.00344","pdf_url":"https://arxiv.org/pdf/2604.00344","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2604.00344","doi":"https://doi.org/10.48550/arxiv.2604.00344","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.00344","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:2604.00344","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2604.00344","pdf_url":"https://arxiv.org/pdf/2604.00344","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":"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/W7148554124.pdf","grobid_xml":"https://content.openalex.org/works/W7148554124.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7223042917","https://openalex.org/W7131384206","https://openalex.org/W7140125034","https://openalex.org/W4414690299","https://openalex.org/W4403575202","https://openalex.org/W7207616193","https://openalex.org/W7155050529","https://openalex.org/W7133267436","https://openalex.org/W4416054967","https://openalex.org/W4415428106"],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"have":[4],"shown":[5],"remarkable":[6],"performance":[7],"in":[8,126,195],"completing":[9],"various":[10],"tasks.":[11],"However,":[12],"solving":[13],"complex":[14],"problems":[15],"often":[16],"requires":[17],"the":[18,134,154,188,197],"coordination":[19],"of":[20,75,190,199],"multiple":[21],"agents,":[22],"raising":[23],"a":[24,42,51,73,81,91,101,112,163],"fundamental":[25],"question:":[26],"how":[27],"to":[28,139],"effectively":[29],"select":[30],"and":[31,97,129,147,175,181],"interconnect":[32],"these":[33],"agents.":[34],"In":[35],"this":[36],"paper,":[37],"we":[38],"propose":[39],"\\textbf{Agent":[40],"Q-Mix},":[41],"reinforcement":[43],"learning":[44],"framework":[45,110],"that":[46,78,115],"reformulates":[47],"topology":[48,193],"selection":[49],"as":[50,162],"cooperative":[52],"Multi-Agent":[53],"Reinforcement":[54],"Learning":[55],"(MARL)":[56],"problem.":[57],"Our":[58],"method":[59],"learns":[60],"decentralized":[61,192],"communication":[62,76,83],"decisions":[63],"using":[64,160],"QMIX":[65],"value":[66],"factorization,":[67],"where":[68],"each":[69],"agent":[70,150],"selects":[71],"from":[72],"set":[74],"actions":[77],"jointly":[79],"induce":[80],"round-wise":[82],"graph.":[84],"At":[85],"its":[86],"core,":[87],"Agent":[88,131,165,172],"Q-Mix":[89,132,166],"combines":[90],"topology-aware":[92],"GNN":[93],"encoder,":[94],"GRU":[95],"memory,":[96],"per-agent":[98],"Q-heads":[99],"under":[100],"Centralized":[102],"Training":[103],"with":[104,119],"Decentralized":[105],"Execution":[106],"(CTDE)":[107],"paradigm.":[108],"The":[109],"optimizes":[111],"reward":[113],"function":[114],"balances":[116],"task":[117],"accuracy":[118,137],"token":[120,145],"cost.":[121],"Across":[122],"seven":[123],"core":[124],"benchmarks":[125],"coding,":[127],"reasoning,":[128],"mathematics,":[130],"achieves":[133,167],"highest":[135],"average":[136],"compared":[138],"existing":[140],"methods":[141],"while":[142],"demonstrating":[143],"superior":[144],"efficiency":[146],"robustness":[148],"against":[149],"failure.":[151],"Notably,":[152],"on":[153],"challenging":[155],"Humanity's":[156],"Last":[157],"Exam":[158],"(HLE)":[159],"Gemini-3.1-Flash-Lite":[161],"backbone,":[164],"20.8\\%":[168],"accuracy,":[169],"outperforming":[170],"Microsoft":[171],"Framework":[173],"(19.2\\%)":[174],"LangGraph":[176],"(19.2\\%),":[177],"followed":[178],"by":[179,183],"AutoGen":[180],"Lobster":[182],"OpenClaw.":[184],"These":[185],"results":[186],"underscore":[187],"effectiveness":[189],"learned,":[191],"optimization":[194],"pushing":[196],"boundaries":[198],"multi-agent":[200],"reasoning.":[201]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-04-03T00:00:00"}