Diffusion models for quantum error suppression

• Grant number: 25/27502-3
• Support Opportunities: Scholarships abroad - Research Internship - Post-doctor
• Start date: March 23, 2026
• End date: March 22, 2027
• Field of knowledge: Physical Sciences and Mathematics - Physics
• Principal Investigator: Celso Jorge Villas-Bôas
• Grantee: Tiago de Souza Farias
• Supervisor: Thomas Monz
• Host Institution: Centro de Ciências Exatas e de Tecnologia (CCET). Universidade Federal de São Carlos (UFSCAR). São Carlos , SP, Brazil
• Institution abroad: University of Innsbruck, Austria
• Associated to the scholarship: 23/15739-3 - Quantum Algorithms for solving Complex Problems and Commercial Applications, BP.PD

Abstract

Quantum computing presents the transformative capability of executing algorithms at speeds far exceeding those of current classical computers, enabling the solution of problems that are computationally prohibitive or even unattainable through classical approaches. Nonetheless, a major obstacle to its practical implementation lies in the pervasive presence of noise, inherently introduced by physical phenomena such as unintended quantum interactions and thermal fluctuations. These sources of noise pose considerable challenges for the effective scaling of quantum computers and currently constrain their practical utility.Machine learning encompasses algorithms that can learn from data and be trained to accomplish specific tasks or objectives. Within this field, diffusion models,constituting a subclass of machine learning algorithms, have demonstrated notable effectiveness in removing perturbations in the form of noise, enabling the generation of data that is statistically consistent with the training distribution. This denoising process is typically achieved through Monte Carlo methods, which provide numerical approximations of probability distributions.This research project proposes an approach that integrates concepts from machine learning and quantum computing. Specifically, it aims to investigate the potential of diffusion models to suppress perturbations and mitigate noise in quantum systems. Such an approach has the capacity to substantially enhance the reliability and quality of quantum computational outputs, reducing error rates and advancing quantum computing toward the realization of its theoretical promise. (AU)