UCLA researchers have developed a new method for online quantum process tomography (QPT) that efficiently characterizes orthogonal random unitary channels, extending beyond Pauli channels to capture both incoherent and coherent noise effects in quantum devices. This approach combines efficient Pauli channel tomography with an online-learning framework, producing accurate noise models that are crucial for error mitigation and calibration in quantum computing systems.
Characterizing noise in quantum devices is essential for building scalable, fault-tolerant quantum computers. Current techniques often focus on Pauli channels, a restricted class of random unitary channels widely used in noise modeling and error correction. While useful, Pauli channels cannot capture coherent noise effects such as cross-talk or systematic calibration errors, which degrade quantum gate fidelity. Full quantum process tomography can, in principle, capture coherent errors, but it is typically resource-intensive and impractical for large systems. Thus, there is a need for efficient, scalable QPT methods that extend to random unitary channels beyond Pauli noise.
This invention integrates:
Modified Pauli channel tomography to efficiently estimate noise channel probabilities.
Online-learning algorithms (including stochastic gradient descent and observable sampling) to iteratively isolate a global unitary component.
A combined iterative framework that constructs an approximate random unitary noise channel.
The method efficiently distinguishes and quantifies coherent and incoherent noise contributions in a device. It provides online adaptability (i.e., it can refine estimates continuously during device operation), making it practical for near-term quantum computers. This innovation builds on prior UCLA work (Case No. 2023-183, Peetz et al.), extending efficient tomography methods into the domain of general orthogonal random unitary channels.
Captures coherent errors (e.g., cross-talk, calibration drift) not modeled by Pauli channels.
Efficient and scalable, avoiding the exponential overhead of full QPT.
Online implementation allows real-time adaptation to device noise conditions.
Compatible with gradient-based learning and observable sampling strategies.
Provides a more accurate and actionable noise model for quantum error correction and mitigation.
Builds upon and extends proven efficient tomography frameworks (UCLA Case No. 2023-183).
Quantum error mitigation and correction: providing realistic noise models for decoding and fault tolerance.
Device benchmarking and calibration: identifying coherent vs. incoherent noise sources.
Quantum algorithm optimization: tailoring circuits to specific device noise models.
Quantum control: improving feedback and adaptive control strategies by learning real-time noise dynamics.
Research in open quantum systems: enabling more accurate experimental studies of system–environment interactions.
Theoretical framework established and described in detail in the thesis (arXiv:2501.17243).
Online-learning procedures and modified Pauli tomography techniques validated through simulations.
Integration with prior efficient tomography approaches (UCLA Case No. 2023-183).
Further experimental validation on quantum hardware is the next development step.
UCLA Case No. 2024-078 — Online Quantum Process Tomography for Orthogonal Random Unitary Channels
Related prior IP: UCLA Case No. 2023-183 (efficient Pauli channel tomography framework).
Peetz, et al. — Efficient Pauli channel tomography methods (UCLA Case No. 2023-183).
Lu, Peetz, et al. — Online Quantum Process Tomography for Orthogonal Random Unitary Channels. arXiv:2501.17243 (2025). DOI / Preprint