AI Summary of Scholarly Research

This page presents an AI-generated summary of a published research paper. The original authors did not write or review this article. [See full disclosure ↓]

Reinforcement learning improved quantum error correction stability

Computer Science research
Onri Jay Benally, Wikimedia Commons, CC BY-SA 2.0 · CC BY-SA 2.0
Research area:physics-astronomyquantum-physics-computing

What the study found

The study found that reinforcement learning, a machine-learning method that learns from feedback, can be combined with quantum error correction so a quantum computer can continuously self-calibrate during computation. The authors report improved stability and record logical error performance in their experiments and simulations.

Why the authors say this matters

The authors conclude that this approach addresses the problem of having to stop a quantum computation for recalibration, which they say is incompatible with the long runtimes expected for future quantum algorithms. They suggest this enables a quantum computer that learns from its errors and keeps computing.

What the researchers tested

The researchers unified calibration with computation by using error-detection events from quantum error correction as a learning signal for a reinforcement learning agent. They tested the framework on a Willow superconducting processor and also ran numerical simulations of large codes with tens of thousands of control parameters.

What worked and what didn't

In experiments, the framework improved the logical stability of the surface code 3.5-fold against injected drift. The authors report record performance for the surface and colour codes, with average logical error per cycle of 7.72(9) × 10−4 and 8.19(14) × 10−3, respectively. The simulations indicate that the optimization speed is independent of system size.

What to keep in mind

The abstract does not describe detailed limitations beyond the stated scope of the experiments and simulations. The reported results are specific to the tested processor, the surface and colour codes, and the simulated large-code setting.

Key points

  • Reinforcement learning was used to continuously steer control parameters during quantum error correction.
  • The method repurposed error-detection events as a learning signal.
  • On a Willow superconducting processor, the surface code’s logical stability improved 3.5-fold against injected drift.
  • The authors report record logical error rates for the surface and colour codes.
  • Simulations with tens of thousands of control parameters showed optimization speed independent of system size.

Disclosure

Research title:
Reinforcement learning improved quantum error correction stability
Authors:
Volodymyr Sivak, Alexis Morvan, Michael Broughton, Rodrigo G. Cortiñas, Johannes Bausch, Andrew W. Senior, M. Neeley, Alec Eickbusch, Noah Shutty, Laleh Aghababaie Beni, James S. Spencer, Francisco J. Heras, Thomas Edlich, Dmitry A. Abanin, Amira Abbas, Rajeev Acharya, Georg Aigeldinger, Ross Alcaraz, Sayra Alcaraz, Trond I. Andersen, M. Ansmann, Frank Arute, Kunal Arya, Walt Askew, Nikita Astrakhantsev, Juan Atalaya, Brian Ballard, Joseph C. Bardin, Hector Bates, Andreas Bengtsson, Majid Karimi, Alexander Bilmes, Simon Bilodeau, Felix Borjans, Alexandre Bourassa, Jenna Bovaird, Dylan Bowers, L. Brill, Peter Brooks, David A. Browne, Brett Buchea, Bob B. Buckley, Tim Burger, Brian Burkett, Nicholas Bushnell, Jamal Busnaina, Anthony Cabrera, Juan Campero, Hung-Shen Chang, Silas Chen, B. Chiaro, Liang-Ying Chih, Agnetta Y. Cleland, Bryan Cochrane, Matt Cockrell, Josh Cogan, Roberto Collins, Paul Conner, Harold J. Cook, William Courtney, Alexander L. Crook, Ben Curtin, Martin Damyanov, Sayan Das, Dripto M. Debroy, Sean Demura, Paul Donohoe, Ilya Drozdov, Andrew Dunsworth, Valerie Ehimhen, Aviv Moshe Elbag, Lior Ella, Mahmoud Elzouka, David Enriquez, Catherine Erickson, Vinicius S. Ferreira, Marcos Flores-Álamo, Leslie Flores Burgos, Ebrahim Forati, Jeremiah Ford, Austin G. Fowler, Brooks Foxen, Masaya Fukami, Alan Wing Lun Fung, Lenny Fuste, Suhas Ganjam, Gonzalo Cerruela García, Christopher Garrick, Robert Gasca, Helge Gehring, Robert Geiger, Élie Genois, William Giang, Dar Gilboa, James E. Goeders, E. Gonzales, Raja Gosula, Stijn J. de Graaf, Alejandro Grajales Dau, Dietrich Graumann, Joel Grebel, Alex Greene, Jonathan A. Gross, Jose Guerrero, Loïck Le Guevel, Tan Ha, Steve Habegger, Tanner Hadick, Ali Hadjikhani, Michael C. Hamilton, Matthew P. Harrigan, Sean D. Harrington, Jeanne Hartshorn, Stephen Heslin, Paula Heu, Oscar Higgott, Reno Hiltermann, Hsin-Yuan Huang, Mike Hucka, Christopher Hudspeth, Ashley Huff, William J. Huggins, E. Jeffrey, Shaun Jevons, Jiang Zhang, Xiaoxuan Jin, Chaitali Joshi, Pavol Juhás, A. Kabel, Dvir Kafri, Hui Kang, Kiseo Kang, Amir H. Karamlou, Ryan Kaufman, Kostyantyn Kechedzhi, Tanuj Khattar, Mostafa Khezri, Seon Kim, Can M. Knaut, Bryce Kobrin, Fedor Kostritsa, John Mark Kreikebaum, Ryuho Kudo, Ben Kueffler, Arun Kumar, Vladislav D. Kurilovich, Vitali Kutsko, Nathan Lacroix, David Landhuis, Tiano Lange-Dei, Brandon W. Langley, Pavel Laptev, Kim-Ming Lau, Justin Ledford, Joy Lee, Kenny Lee, Brian Lester, Wendy Leung, Yali Li, Wing Yan Li, Ming Li, Alexander T. Lill, William P. Livingston, Matthew T. Lloyd, Aditya Locharla, Laura de Lorenzo, Daniel Lundahl, Aaron Lunt, Sid Madhuk, Aniket Maiti, Ashley Maloney, Salvatore Mandrà, Leigh S. Martin, Orion Martin, Eric Mascot, Paul Masih Das, Dmitri Maslov, M. Geo Mathews, Cameron Maxfield, Jarrod R. McClean, Matt McEwen, Seneca Meeks, Kevin C. Miao, Zlatko Minev, Reza Molavi, Sebastian Molina, Shirin Montazeri, C. Neill, Michael Newman, Anthony Nguyen, Murray Nguyen, Chia-Hung Ni, Murphy Yuezhen Niu, Logan Oas, Raymond Orosco, Kristoffer Ottosson, Alice Pagano, Agustín Di Paolo, Sherman Peek, David R. Peterson, A. Pizzuto, Elías Portolés, R. Potter, Orion Pritchard, Michael C. Qian, Chris Quintana, Arpit Ranadive, Matthew J. Reagor, Rachel Resnick, David M. Rhodes, Daniel Riley, Gabrielle Roberts, Roberto Rodriguez, Emma Ropes, L. Rose, Eliott Rosenberg, Emma Rosenfeld, Dario Rosenstock, Elizabeth Rossi, P. Roushan, David A. Rower, Robert D. Salazar, Kannan Sankaragomathi, Murat Can Sarihan, Kevin J. Satzinger, Max Schaefer, Sebastian Schroeder, Henry F. Schurkus, Aria Shahingohar, Michael J. Shearn, Aaron Shorter, Vladimir Shvarts, Spencer Small, W. Clarke Smith, David A. Sobel, Barrett Spells, Sofia Springer, George Sterling, Jordan Suchard, Aaron Szasz, Alex Sztein, Madeline Taylor, Jothi Priyanka Thiruraman, Douglas Thor, Doǧan A. Timuçin, Eifu Tomita, A. Torres, Mustafa Mert Torunbalcı, Hao Tran, Abeer Vaishnav, Justin Vargas, Sergey Vdovichev, Guifré Vidal, Catherine Vollgraff Heidweiller, Meghan Voorhees, Steven Waltman, Jonathan Waltz, Shannon X. Wang, Brayden Ware, James D. Watson, Yonghua Wei, Travis Weidel, T. White, Kristi Wong, Bryan W. K. Woo, Christopher J. Wood, Maddy Woodson, Cheng Xing, Z. Jamie Yao, P. Yeh, Bicheng Ying, Juhwan Yoo, Noureldin Yosri, Elliot Young, Grayson Young, Adam Zalcman, Ran Zhang, Yaxing Zhang, Ningfeng Zhu, Nicholas Zobrist, Zhenjie Zou, Ryan Babbush, Dave Bacon, Sergio Boixo, Yu Chen, Zijun Chen, Michel Devoret, Monica Hansen, Jeremy Hilton, Cody Jones, J. Kelly, Alexander N. Korotkov, Erik Lucero, A. Megrant, Hartmut Neven, William D. Oliver, Ganesh Ramachandran, Vadim Smelyanskiy, Paul V. Klimov
Institutions:
Google (United Kingdom), Google (United Kingdom), Google (United Kingdom), Google (United Kingdom), Google (United Kingdom), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google (United States), Google DeepMind (United Kingdom), Google DeepMind (United Kingdom), Google DeepMind (United Kingdom), Google DeepMind (United Kingdom), Google DeepMind (United Kingdom), University of California, Santa Barbara, University of California, Santa Barbara, University of Massachusetts Amherst
Publication date:
2026-07-08
OpenAlex record:
View
Image credit:
Onri Jay Benally, Wikimedia Commons, CC BY-SA 2.0
AI provenance: This post was generated by gpt-5.4-mini (OpenAI). The original authors did not write or review this post.