Quantum processors are expected to speed up various applications, including, but not limited to, chemistry, finance, and secure communication, on the premise that the device's error rate is well below the quantum error correction threshold.Thus, the infrastructure provider must perform routine characterization and calibration (C&C) to maintain the device's quality.The user cannot access the device during the C&C routine, which is not ideal but necessary for both the user and the provider.To maximize device availability, the provider must minimize the time required for the C&C routine.This thesis introduces a modular C&C framework designed to reduce system downtime, implemented as an opensource Python library named Inspeqtor.To enable broad data efficiency, the framework establishes compatible abstractions for Bayesian Optimal Experimental Design (BOED) tailored for general-purpose models.Alongside this capability, Inspeqtor provides firstclass support for a Graybox characterization method adaptable to multiple qubit implementations.Through statistical analysis of this Graybox method, we derive a performance bound showing that, while the expected minimum of the empirical error depends on the number of measurement shots, the predictive performance bound on average gate fidelity remains independent of the shot count due to the cancellation of data variance.This reveals that once the model's test error approaches this shot-dependent variance, it can accurately predict average gate fidelity, suggesting that strategic balancing of shot count and sample size can be exploited to train highly accurate predictive models.Furthermore, we develop a Probabilistic Graybox Model (PGM) to enable the integration of Graybox characterization with BOED.We validate this approach via a numerical simulation, demonstrating that the PGM maintains high accuracy even under stochastic noise and with limited training sample sizes.By building each module directly around the system's mathematical representation, Inspeqtor bridges the gap between software implementation and mathematical abstraction.This functional, composable approach narrows the domain-knowledge gap for infrastructure engineers and quantum researchers, widening the path to highly available quantum computing platforms.
Quantum processors are expected to speed up various applications, including, but not limited to, chemistry, finance, and secure communication, on the premise that the device's error rate is well below the quantum error correction threshold.Thus, the infrastructure provider must perform routine characterization and calibration (C&C) to maintain the device's quality.The user cannot access the device during the C&C routine, which is not ideal but necessary for both the user and the provider.To maximize device availability, the provider must minimize the time required for the C&C routine.This thesis introduces a modular C&C framework designed to reduce system downtime, implemented as an opensource Python library named Inspeqtor.To enable broad data efficiency, the framework establishes compatible abstractions for Bayesian Optimal Experimental Design (BOED) tailored for general-purpose models.Alongside this capability, Inspeqtor provides firstclass support for a Graybox characterization method adaptable to multiple qubit implementations.Through statistical analysis of this Graybox method, we derive a performance bound showing that, while the expected minimum of the empirical error depends on the number of measurement shots, the predictive performance bound on average gate fidelity remains independent of the shot count due to the cancellation of data variance.This reveals that once the model's test error approaches this shot-dependent variance, it can accurately predict average gate fidelity, suggesting that strategic balancing of shot count and sample size can be exploited to train highly accurate predictive models.Furthermore, we develop a Probabilistic Graybox Model (PGM) to enable the integration of Graybox characterization with BOED.We validate this approach via a numerical simulation, demonstrating that the PGM maintains high accuracy even under stochastic noise and with limited training sample sizes.By building each module directly around the system's mathematical representation, Inspeqtor bridges the gap between software implementation and mathematical abstraction.This functional, composable approach narrows the domain-knowledge gap for infrastructure engineers and quantum researchers, widening the path to highly available quantum computing platforms.
Mr. Poramet Pathumsoot's doctoral dissertation, titled "Data-Efficient Characterization & Calibration Framework for Quantum Processors," addresses a critical problem in evaluating the health of a quantum computer in order to minimize errors.In chapter 1, Poramet introduces the promises that quantum devices have to offer, and the challenges in the development.He points out that although we can use quantum error correction to suppress the error, the underlying operations have to operate well below some error tolerant threshold.To achieve that, we need to perform routine maintenance on the quantum device, which increase downtime and reduce the device availability.Furthermore, performing maintenance operations also require collecting data from the device which induce yet another cost.Thus, it is important to reduce the data needed for maintenance and consequently reduce the device downtime.In addition, there are a many possible physical implementations of the device.It is also essential to also address that requirements of each implementation.In his dissertation, he purposes a data-efficient characterization and calibration framework for quantum processors.Specifically, he develops Inspeqtor, an open-source Python library that implement the framework.The library offers a Graybox characterization method that can be applied to a variety of quantum devices.Furthermore, Inspeqtor can help reduce data needed for maintenance routines via Bayesian Optimal Experimental Design (BOED), Data variance, Bayesian optimization approaches.The framework integrates each module together for the ease of usage to both researching and infrastructure maintaining fronts.In the following chapter 2, he reviews the theoretical background required to read the dissertation.He also reviews the maintenance approaches in the literature, which divided into the calibration and characterization phases, and then he discusses the approaches that aims to reduce the data collected from the quantum device.He also reviews the existing library and tools for quantum device maintenance.Chapter 3 lays the theoretical foundation needed in development of Inspeqtor.He starts from statistical machine learning approach using the Graybox characterization method as a toy model.By statistical analysis, he derives expected minimum value of mean squared error between experimental and predicted expectation values, which become a bound for model predictive performance.He verified the analytical result using numerical simulation and perform experiment on the superconducting qubit device.He then step-up the characterization using probabilistic approach using Bayesian inference so that it is compatible with the expected information gain calculation using BOED.Finally, he reviews data-efficient characterization approaches including data variance, and BOED.In chapter 4, he discusses the necessarily of the quantum network to realize the distributed quantum computation.He presents the experimental realization of the quantum network protocol on the superconducting device.He then discusses how characterization and calibration improve the experiment using the Graybox characterization method.Chapter 5 presents and discusses the core design decisions of Inspeqtor.The framework is designed
Abstract Characterization and calibration of quantum devices are necessary steps to achieve fault-tolerant quantum computing. As quantum devices become more sophisticated, it is increasingly essential to rely not only on physics-based models but also on machine learning models with open-loop optimization. The Graybox approach is a recently proposed platform-agnostic method offering flexibility in modeling the implicit noise. Despite promising results in photonic qubits, its suitability for other platforms is still unknown. We investigate its performance and limitations under realistic conditions, both numerically and experimentally, for the characterization and calibration of gates on superconducting-qubit devices. We find that the Graybox approach performs well on the devices’ pulse-level controls and is able to search for pulse shapes of high-fidelity gates. However, an important limiting factor of the model’s ability to reduce the training errors is the finite number of measurement shots. We therefore derive analytical bounds to explain this limitation, and use them to indicate users whether the characterization performance is optimized. Furthermore, the model’s predictability also depends on other factors such as parameter drift that may occur between the characterization and calibration processes. Our results provide insights into quantum device characterization and gate optimization on superconducting qubits under realistic scenarios using the Graybox approach.
Quantum sensors offer significant advantages over classical devices in spatial resolution and sensitivity, enabling transformative applications across materials science, healthcare, and beyond. Their practical performance, however, is often constrained by unmodelled effects, including noise, imperfect state preparation, and non-ideal control fields. In this work, we report the first experimental implementation of a graybox modelling strategy for a solid-state open quantum system. The graybox framework integrates a physics-based system model with a data-driven description of experimental imperfections, achieving higher fidelity than purely analytical (whitebox) approaches while requiring fewer training resources than fully deep-learning blackbox models. We experimentally validate the method on the task of estimating a static magnetic field using a single-spin quantum sensor, performing Bayesian inference with a graybox model trained on prior experimental data. Using roughly 10,000 training datapoints, the graybox model yields several orders of magnitude improvement in mean squared error over the corresponding physics-only model, as well as significant improvement over a blackbox model with comparable size. These results are broadly applicable to a wide range of quantum sensing platforms, not limited to single-spin systems, and are particularly valuable for real-time adaptive protocols, where model inaccuracies can otherwise lead to suboptimal control and degraded performance.
Maximizing the computational utility of near-term quantum processors requires predictive noise models that inform robust, noise-aware compilation and error mitigation. Conventional models often fail to capture the complex error dynamics of real hardware or require prohibitive characterization overhead. We introduce a data-efficient framework that first constructs a physically motivated, parameterized noise model, and subsequently employs machine-learning-driven Bayesian optimization to identify its parameters. Our approach circumvents costly characterization protocols by estimating algorithm- and hardware-specific error parameters directly from readily available experimental data derived from existing application and benchmark circuit executions. The generality and robustness of the framework are demonstrated across diverse algorithms and superconducting devices, yielding high-fidelity predictions by estimating an independent parameter set tailored to each specific algorithm-hardware context. Crucially, we show that a model calibrated exclusively on small-scale circuits accurately predicts the behavior of larger validation circuits. Our data-efficient approach achieves up to a 65% improvement in model fidelity quantified by the Hellinger distance between predicted and experimental circuit output distributions, compared to standard noise models derived from device properties. This work establishes a practical paradigm for application-aware noise characterization, enabling compilation and error-mitigation strategies tailored to the specific interplay between quantum algorithms and device-specific noise dynamics.
While the Graybox characterization method allows for implicit noise models and is platform-agnostic, the method lacks uncertainty quantification. Characterization of quantum devices is a crucial process that enables researchers to gain insight from experimental settings. Graybox characterization combines known system dynamics with unknown transformations, where the latter is modeled using machine learning. Prediction uncertainty helps researchers make informed decisions. It allows valuable insights from the devices without overconfidence. We therefore develop a probabilistic Graybox characterization model using probabilistic machine learning, specifically Bayesian Neural Networks, and utilize binary measurement outcomes directly for inference. With stochastic noise in a quantum device, we analyze statistical properties of the measurement data. Our results show that the model's prediction performance solely depends on its ability to capture the expected value of the true expectation value. Our proposed probabilistic Graybox model outperforms the original model by up to 1.9 times in capturing the distribution of observed data. We expect that our results will serve as an additional tool for characterizing quantum devices with uncertainty estimation, as they provide a flexible choice that can be utilized even without extensive prior knowledge of the noise model of the devices.
We experimentally study error mitigation and finite-shot sampling on superconducting quantum hardware under a constrained execution budget. The study combines calibration-aware qubit selection, circuit-depth scaling, zero-noise extrapolation, dynamical decoupling, readout-error mitigation, and repeated-shot estimation on an IBM Quantum processor. Experiments are organized across ideal simulation, noise-model simulation, and physical-device execution to separate sampling uncertainty from device-induced error. We investigate how mitigation performance changes with circuit depth, effective noise scale, qubit connectivity, and measurement budget, and quantify accuracy using expectation-value error, mean-squared error, statistical uncertainty, and mitigation gain. A fixed hardware-execution budget is used to evaluate shot allocation strategies and repeated measurements without relying on unlimited sampling. The resulting analysis provides a hardware-aware characterization of when mitigation improves expectation-value estimation and when finite-shot fluctuations offset the benefit of additional mitigation overhead. The implementation uses contemporary Qiskit and IBM Quantum Runtime workflows and is designed to provide reproducible experimental evidence for error-mitigation studies on current quantum processors.