{"id":"https://openalex.org/W7123272391","doi":"https://doi.org/10.48550/arxiv.2601.05475","title":"MaxCode: A Max-Reward Reinforcement Learning Framework for Automated Code Optimization","display_name":"MaxCode: A Max-Reward Reinforcement Learning Framework for Automated Code Optimization","publication_year":2026,"publication_date":"2026-01-09","ids":{"openalex":"https://openalex.org/W7123272391","doi":"https://doi.org/10.48550/arxiv.2601.05475"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2601.05475","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2601.05475","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2601.05475","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5044588154","display_name":"Jiefu Ou","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Ou, Jiefu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122803433","display_name":"Sapana Chaudhary","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chaudhary, Sapana","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5050663933","display_name":"Kaj Bostrom","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bostrom, Kaj","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5062395513","display_name":"Nathaniel Weir","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Weir, Nathaniel","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122775324","display_name":"Shuai Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Shuai","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122808730","display_name":"Huzefa Rangwala","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Rangwala, Huzefa","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5110359290","display_name":"George Karypis","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Karypis, George","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":7,"corresponding_author_ids":["https://openalex.org/A5044588154"],"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":0.15850000083446503,"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":0.15850000083446503,"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/T10260","display_name":"Software Engineering Research","score":0.093299999833107,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T10036","display_name":"Advanced Neural Network Applications","score":0.06300000101327896,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/speedup","display_name":"Speedup","score":0.8580999970436096},{"id":"https://openalex.org/keywords/modular-design","display_name":"Modular design","score":0.5781000256538391},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.5778999924659729},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.4778999984264374},{"id":"https://openalex.org/keywords/coding","display_name":"Coding (social sciences)","score":0.4657000005245209},{"id":"https://openalex.org/keywords/program-optimization","display_name":"Program optimization","score":0.39809998869895935},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.3937999904155731},{"id":"https://openalex.org/keywords/cuda","display_name":"CUDA","score":0.3889000117778778},{"id":"https://openalex.org/keywords/generative-model","display_name":"Generative model","score":0.3878999948501587}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8634999990463257},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.8580999970436096},{"id":"https://openalex.org/C101468663","wikidata":"https://www.wikidata.org/wiki/Q1620158","display_name":"Modular design","level":2,"score":0.5781000256538391},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.5778999924659729},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.4778999984264374},{"id":"https://openalex.org/C179518139","wikidata":"https://www.wikidata.org/wiki/Q5140297","display_name":"Coding (social sciences)","level":2,"score":0.4657000005245209},{"id":"https://openalex.org/C139571649","wikidata":"https://www.wikidata.org/wiki/Q1156793","display_name":"Program optimization","level":3,"score":0.39809998869895935},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.3937999904155731},{"id":"https://openalex.org/C2778119891","wikidata":"https://www.wikidata.org/wiki/Q477690","display_name":"CUDA","level":2,"score":0.3889000117778778},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.3878999948501587},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.38749998807907104},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3840000033378601},{"id":"https://openalex.org/C39890363","wikidata":"https://www.wikidata.org/wiki/Q36108","display_name":"Generative grammar","level":2,"score":0.37279999256134033},{"id":"https://openalex.org/C133162039","wikidata":"https://www.wikidata.org/wiki/Q1061077","display_name":"Code generation","level":3,"score":0.3691999912261963},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.36469998955726624},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.35830000042915344},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.35350000858306885},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.34209999442100525},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.33329999446868896},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.33309999108314514},{"id":"https://openalex.org/C48372109","wikidata":"https://www.wikidata.org/wiki/Q3913","display_name":"Binary number","level":2,"score":0.32679998874664307},{"id":"https://openalex.org/C106189395","wikidata":"https://www.wikidata.org/wiki/Q176789","display_name":"Markov decision process","level":3,"score":0.31940001249313354},{"id":"https://openalex.org/C2777138346","wikidata":"https://www.wikidata.org/wiki/Q1714153","display_name":"Performance tuning","level":2,"score":0.3158000111579895},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.3095000088214874},{"id":"https://openalex.org/C137287247","wikidata":"https://www.wikidata.org/wiki/Q1329550","display_name":"Static program analysis","level":4,"score":0.2948000133037567},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.28790000081062317},{"id":"https://openalex.org/C2778915421","wikidata":"https://www.wikidata.org/wiki/Q3643177","display_name":"Performance improvement","level":2,"score":0.2831999957561493},{"id":"https://openalex.org/C2777904410","wikidata":"https://www.wikidata.org/wiki/Q7397","display_name":"Software","level":2,"score":0.27649998664855957},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2687999904155731},{"id":"https://openalex.org/C63435697","wikidata":"https://www.wikidata.org/wiki/Q864135","display_name":"Binary code","level":3,"score":0.2685000002384186},{"id":"https://openalex.org/C199519371","wikidata":"https://www.wikidata.org/wiki/Q942695","display_name":"Source lines of code","level":3,"score":0.25459998846054077},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2538999915122986},{"id":"https://openalex.org/C527412718","wikidata":"https://www.wikidata.org/wiki/Q855395","display_name":"Interpretation (philosophy)","level":2,"score":0.25060001015663147}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2601.05475","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2601.05475","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2601.05475","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2601.05475","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education","score":0.6233535408973694}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"demonstrate":[4],"strong":[5],"capabilities":[6],"in":[7,37,192],"general":[8],"coding":[9],"tasks":[10],"but":[11],"encounter":[12],"two":[13],"key":[14],"challenges":[15],"when":[16],"optimizing":[17],"code:":[18],"(i)":[19],"the":[20,68,96,106,130,142,168],"complexity":[21],"of":[22,47],"writing":[23],"optimized":[24,180],"code":[25,143,181],"(such":[26],"as":[27],"performant":[28],"CUDA":[29],"kernels":[30],"and":[31,40,43,52,98,126,129,171,188,196],"competition-level":[32],"CPU":[33],"code)":[34],"requires":[35,45],"expertise":[36],"systems,":[38],"algorithms":[39,65],"specific":[41],"languages":[42],"(ii)":[44],"interpretation":[46],"performance":[48,127,182],"metrics":[49],"like":[50],"timing":[51],"device":[53],"utilization":[54],"beyond":[55],"binary":[56],"correctness.":[57],"In":[58],"this":[59],"work,":[60],"we":[61,109,151],"explore":[62],"inference-time":[63],"search":[64,87],"that":[66,116,177],"guide":[67],"LLM":[69],"to":[70,141,162,184],"discover":[71],"better":[72],"solutions":[73],"through":[74],"iterative":[75],"refinement":[76],"based":[77],"on":[78,167],"execution":[79,119],"feedback.":[80],"Our":[81],"approach,":[82],"called":[83],"MaxCode":[84,178],"unifies":[85],"existing":[86],"methods":[88],"under":[89],"a":[90,111,153],"max-reward":[91],"reinforcement":[92],"learning":[93],"framework,":[94],"making":[95],"observation":[97,107],"action-value":[99],"functions":[100],"modular":[101],"for":[102],"modification.":[103],"To":[104,146],"enhance":[105],"space,":[108],"integrate":[110],"natural":[112],"language":[113],"critique":[114],"model":[115,156],"converts":[117],"raw":[118],"feedback":[120],"into":[121],"diagnostic":[122],"insights":[123],"about":[124],"errors":[125],"bottlenecks,":[128],"best-discounted":[131],"reward":[132],"seen":[133],"so":[134],"far.":[135],"Together,":[136],"these":[137],"provide":[138],"richer":[139],"input":[140],"proposal":[144],"function.":[145],"improve":[147],"exploration":[148],"during":[149],"search,":[150],"train":[152],"generative":[154],"reward-to-go":[155],"using":[157],"action":[158],"values":[159],"from":[160],"rollouts":[161],"rerank":[163],"potential":[164],"solutions.":[165],"Testing":[166],"KernelBench":[169],"(CUDA)":[170],"PIE":[172],"(C++)":[173],"optimization":[174],"benchmarks":[175],"shows":[176],"improves":[179],"compared":[183],"baselines,":[185],"achieving":[186],"20.3%":[187],"10.1%":[189],"relative":[190,197],"improvements":[191],"absolute":[193],"speedup":[194,198],"value":[195],"ranking,":[199],"respectively.":[200]},"counts_by_year":[],"updated_date":"2026-01-13T12:58:16.504669","created_date":"2026-01-13T00:00:00"}