[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-82342-en":3,"doc-seo-82342-105":29,"detail-sidebar-cat-0-en-105":90},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},82342,1374391974585,"Genevieve","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",8,"Research & Report","Action-Factored Multi-Agent Reinforcement Learning for Scalable Quantum Device Tuning","Cooperative multi-agent reinforcement learning suits large parameter spaces with exploitable local structure, exemplified by electrostatically defined quantum-dot array tuning. Strong parameter cross-talk makes each agent’s environment non-stationary and can destabilize learning, mirroring difficulties in manual calibration. The framework QADAPT learns an online factored action representation to decouple agents and reduce interference, enabling efficient shared-policy learning from local measurements and rewards. The modular design delivers zero-shot generalization to unseen device sizes while keeping convergence steps approximately constant, supporting rapid calibration of large quantum processors.","Action-Factored Multi-Agent Reinforcement Learning for Scalable Quantum Device Tuning  \narXiv :2607 .09422v 1 [ cs .LG] 10 Jul 2026  \nEdwin De Nicolo∗ Department of Physics University of Oxford  \nUnited Kingdom [edwin.denicolo@merton.ox.ac.uk](edwin.denicolo@merton.ox.ac.uk)  \nCornelius Carlsson  \nDepartment of Engineering Science University of Oxford  \nUnited Kingdom [cornelius.carlsson@spc.ox.ac.uk](cornelius.carlsson@spc.ox.ac.uk)  \nRahul Marchand†  \nDepartment of Engineering Science University of Oxford United Kingdom  \n[rahul.marchand@new.ox.ac.uk](rahul.marchand@new.ox.ac.uk)  \nPranav Vaidhyanathan  \nDepartment of Engineering Science University of Oxford United Kingdom [pranav@robots.ox.ac.uk](pranav@robots.ox.ac.uk)  \nNatalia Ares  \nDepartment of Engineering Science  \nUniversity of Oxford  \nUnited Kingdom  \n[natalia.ares@eng.ox.ac.uk](natalia.ares@eng.ox.ac.uk)  \nAbstract  \nCooperative multi-agent reinforcement learning is well suited to problems with large parameter spaces and exploitable local structure, such as the tuning of electrostatically-defined quantum-dot arrays. However, if parameter cross-talk is strong, a non-stationary environment from the perspective of any individual agent can destabilize learning – the same effect that plagues manual tuning of such systems. We propose using a factored representation of the action space, learned online, to decouple agents and minimize their interference. Our framework, QADAPT, uses this factorization to efficiently learn shared policies based on local measurements and rewards. With this modular strategy, we achieve zero-shot generalization to unseen quantum device sizes and maintain an approximately constant number of convergence steps to reach target regimes. This work provides a scalable route toward the rapid calibration of large-scale quantum processors.  \n1 Introduction  \nReinforcement learning (RL) provides a general framework for sequential decision-making under uncertainty, where agents learn policies through interaction with an environment [1] . Combining RL with deep neural networks has achieved human-level performance on complex, high-dimensional tasks [2], and its extension to continuous control domains has allowed agents to learn efficiently from sensory inputs [3] . This opens practical applications for robotics and locomotion, and more recently, has bridged RL with the field of quantum computing for tasks such as error-robust quantum  \ngate design [4], analog pulse shaping [5], real-time qubit feedback [6, 7], and voltage tuning of ∗ Equal Contribution. Order determined alphabetically.  \n†Equal Contribution. Order determined alphabetically.  \nPreprint.  \nquantum dots [8] . While RL, and machine learning in general, is becoming central to quantum control [9, 10], a key challenge lies in scaling these computational methods efficiently to larger devices [11, 12, 13, 14, 15, 16] .  \nSemiconductor spin qubits in particular, while a compelling quantum computing platform [17, 18, 19, 20], face enormous control complexity due to measurement noise, device non-uniformities, and cross-talk. The tuning procedure involves calibrating dc gate voltages to create confined islands of charge, i.e. quantum dots, as shown in Fig. 1a. There exist two main gate types; plunger gates predominantly control the electrochemical potential of quantum dots, hence their charge occupations, and barrier gates modulate the tunnel coupling between dots. Because of dense gate electrode arrangements, adjusting one gate voltage perturbs neighboring dots via capacitive cross-talk [21] . Measuring the system’s response as a function of two gate voltages produces images known as charge stability diagrams (CSDs) . These diagrams contain relevant tuning and cross-talk information, which may be inferred from the position and curvature of the line features they contain. As shown in Fig. 1b, the objective is to reach a voltage configuration that brings all charge occupations and tunnel-couplings to a ","cbCaieLpRwXBuVMF","https://ap.wps.com/l/cbCaieLpRwXBuVMF","pdf",3872921,1,31,"English","en",105,"# Introduction\n## Quantum control and reinforcement learning context\n## Quantum dot tuning challenge and objectives\n# QADAPT framework overview\n## Action-space factorization and decoupled control\n## CTDE training and role-shared policies\n# Contributions","[{\"question\":\"Why does parameter cross-talk cause instability in cooperative multi-agent reinforcement learning for quantum tuning?\",\"answer\":\"Strong cross-talk makes the environment non-stationary from the perspective of any single agent, destabilizing learning. The same interference issue also complicates manual tuning of these systems.\"},{\"question\":\"What is QADAPT and what problem does it address?\",\"answer\":\"QADAPT is a multi-agent reinforcement learning framework for scalable tuning of quantum devices. It addresses action-space growth and interference by learning a factored action representation online for decoupled control.\"},{\"question\":\"How does QADAPT achieve generalization to unseen quantum device sizes?\",\"answer\":\"QADAPT uses the learned action factorization and a modular architecture so newly introduced agents can inherit existing policies. This enables zero-shot generalization to quantum device sizes not seen during training.\"}]",1784179766,78,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":85,"head_meta":87,"extra_data":89,"updated_unix":27},"action-factored-multi-agent-reinforcement-learning-for-scalable-quantum-device-tuning","",{"@graph":35,"@context":84},[36,53,67],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/action-factored-multi-agent-reinforcement-learning-for-scalable-quantum-device-tuning/82342/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":61,"encodingFormat":60,"isAccessibleForFree":62,"interactionStatistic":63},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-16",true,{"@type":64,"interactionType":65,"userInteractionCount":4},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"Why does parameter cross-talk cause instability in cooperative multi-agent reinforcement learning for quantum tuning?","Question",{"text":74,"@type":75},"Strong cross-talk makes the environment non-stationary from the perspective of any single agent, destabilizing learning. The same interference issue also complicates manual tuning of these systems.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"What is QADAPT and what problem does it address?",{"text":79,"@type":75},"QADAPT is a multi-agent reinforcement learning framework for scalable tuning of quantum devices. It addresses action-space growth and interference by learning a factored action representation online for decoupled control.",{"name":81,"@type":72,"acceptedAnswer":82},"How does QADAPT achieve generalization to unseen quantum device sizes?",{"text":83,"@type":75},"QADAPT uses the learned action factorization and a modular architecture so newly introduced agents can inherit existing policies. This enables zero-shot generalization to quantum device sizes not seen during training.","https://schema.org",{"og:url":51,"og:type":86,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":88,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":91},[92,96,100,104,109,114,119,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":93,"show_sort_weight":94,"slug":95},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":97,"show_sort_weight":98,"slug":99},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":101,"show_sort_weight":102,"slug":103},"Exam",70,"exam",{"id":105,"doc_module":4,"doc_module_name":45,"category_name":106,"show_sort_weight":107,"slug":108},5,"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":45,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":45,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":45,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":45,"category_name":136,"show_sort_weight":105,"slug":137},19,"General","general"]