Abstract A hybrid physics–machine-learning framework is introduced for the joint optimization of separate fidelity and Grover algorithm performance in superconducting quantum processors. The method combines a Hamiltonian-based open-quantum-system model of a two-qubit architecture coupled through a bus resonator with a QGHNN. Readout performance is evaluated via the Lindblad master-equation simulation of the dispersive measurement process, in which state-dependent resonator dynamics are mapped to realistic I/Q distributions and classified to estimate single-shot measurement fidelity. In parallel, algorithmic performance is assessed by simulating a two-qubit Grover search circuit in the presence of dissipation and dephasing, using effective cross-Kerr interactions derived from the underlying circuit parameters. A physically constrained dataset is constructed by sampling of some critical variables relevant to system. The QGHNN is trained to learn the nonlinear relationship between device parameters and the dual objectives of readout (separation) fidelity and Grover success probability. Gradient-based optimization is then carried out directly in parameter space using the trained network, yielding operating points that simultaneously improve measurement distinguishability and Grover algorithm accuracy. Simulation results demonstrate that the readout separation fidelity exceeds 99.4%, while the Grover success probability reaches approximately 86%, and this is comparable with the accuracy achieved by IBM Q5 Tenerife processor. Finally, the optimized parameters are used to implement the quantum circuit layout in Qiskit Metal, and their validity is further confirmed through EPR-based certification.
Abstract A hybrid physics–machine-learning framework is introduced for the joint optimization of separate fidelity and Grover algorithm performance in superconducting quantum processors. The method combines a Hamiltonian-based open-quantum-system model of a two-qubit architecture coupled through a bus resonator with a QGHNN. Readout performance is evaluated via the Lindblad master-equation simulation of the dispersive measurement process, in which state-dependent resonator dynamics are mapped to realistic I/Q distributions and classified to estimate single-shot measurement fidelity. In parallel, algorithmic performance is assessed by simulating a two-qubit Grover search circuit in the presence of dissipation and dephasing, using effective cross-Kerr interactions derived from the underlying circuit parameters. A physically constrained dataset is constructed by sampling of some critical variables relevant to system. The QGHNN is trained to learn the nonlinear relationship between device parameters and the dual objectives of readout (separation) fidelity and Grover success probability. Gradient-based optimization is then carried out directly in parameter space using the trained network, yielding operating points that simultaneously improve measurement distinguishability and Grover algorithm accuracy. Simulation results demonstrate that the readout separation fidelity exceeds 99.4%, while the Grover success probability reaches approximately 86%, and this is comparable with the accuracy achieved by IBM Q5 Tenerife processor. Finally, the optimized parameters are used to implement the quantum circuit layout in Qiskit Metal, and their validity is further confirmed through EPR-based certification.
This study introduces a superconducting quantum chip architecture designed to simultaneously preserve entanglement and readout fidelity, addressing one of the key trade-offs in the development of scalable quantum hardware. In conventional quantum circuits, strong qubit qubit coupling enhances entanglement but often leads to undesired crosstalk, dephasing, and reduced measurement fidelity. To mitigate these effects, we propose a hybrid multiqubit configuration consisting of nine transmon qubits organized into interior and exterior groups, interconnected via a flux tunable qubit and a network of distributed resonators. The interior qubits along with tunable qubit form an entanglement core, while the exterior qubits operate in the dispersive regime under large detuning to enable readout. The degree of entanglement can be dynamically tuned by adjusting the coupling between the central tunable qubit and the interior qubits. The total Hamiltonian includes all significant coupling contributions, encompassing effective exchange interactions among interior and exterior qubits, as well as their mediated couplings through interface resonators. By numerically solving the complete Hamiltonian alongside the Lindblad master equation, the system dynamics are characterized, allowing evaluation of both spectroscopic features and separation fidelity. Simulation results demonstrate that the proposed design maintains strong entanglement by creating the avoided-crossing region while sustaining measurement fidelity around 0.995 under realistic noise conditions. These findings confirm that entanglement strength and readout fidelity can be co-optimized within a single, reconfigurable architecture, establishing a viable route toward high-performance and scalable superconducting quantum processors.
Although quantum computers will not necessarily transform ML in general, they hold immense potential for specific uses, such as dealing with high-dimensional data and complex correlations that cannot be effectively modeled using classical systems. It is, however, difficult to design efficient quantum circuits, especially with the existing Noisy Intermediate-Scale Quantum (NISQ) hardware that has constraints including decoherence, a limited number of qubits, and gate fidelity limitations. The authors propose an Adaptive Quantum Circuit Architecture Search (AQCAS) framework whose structure and parameters are optimized dynamically to improve the performance of the task-specific ML. The framework integrates quantum circuit encoding, a hybrid search strategy that is based on reinforcement learning and evolutionary algorithms, and a task-specific adaptivity mechanism. Benchmark datasets were used for experimental evaluations of both simulation using Qiskit Aer and classical optimization using the Python frameworks TensorFlow and PennyLane. Fixed and standard variational quantum circuits were compared with performance measurements like the classification accuracy, circuit depth, convergence speed, and computational efficiency. The results show that the accuracy of AQCAS can be raised to 94.1%, and the circuit depth is lowered at the same time. It converged rapidly and is computationally efficient. Ablation studies show that both hybrid search and adaptivity mechanisms help to enhance generalization and stability under NISQ constraints. The results demonstrate the potential of AQCAS as a practical approach to using adaptive quantum circuits in specialized ML applications, and lay the groundwork for future development of scaling to larger devices, implementing noise-aware optimization, and integrating with hybrid quantum-classical ML pipelines.
Measurement is an essential component of robust and practical quantum computation. For superconducting qubits, the measurement process involves the effective manipulation of the joint qubit-resonator dynamics, and it should ideally provide the highest quality for qubit state discrimination with the shortest readout pulse and resonator reset time. Here, we harness model-free reinforcement learning (RL), together with a tailored training environment, to achieve this multifaceted optimization task. Using the IBM quantum device, we demonstrate that the pulse obtained by the RL agent not only successfully achieves state-of-the-art performance, with an assignment error of ( 4.6 ± 0.4 ) × 10 − 3 , but also executes the readout and the subsequent resonator reset almost three times faster than the system’s default process. Furthermore, the learned waveforms are robust against realistic parameter drifts and follow an intuitive form, making them readily implementable on existing hardware with little computational overhead. Our results provide an effective readout strategy to boost the performance of superconducting quantum processors and demonstrate the value of RL in providing optimal and practical solutions for complex quantum information processing tasks.
Open quantum systems can undergo dissipative phase transitions, and their critical behavior can be exploited in sensing applications. For example, it can be used to enhance the fidelity of superconducting qubit readout measurements, a central problem toward the creation of reliable quantum hardware. A recently introduced measurement protocol, named ``critical parametric quantum sensing'', uses the parametric (two-photon driven) Kerr resonator's driven-dissipative phase transition to reach single-qubit detection fidelity of 99.9\% [arXiv:2107.04503]. In this work, we improve upon the previous protocol by using machine learning-based classification algorithms to \textit{efficiently and rapidly} extract information from this critical dynamics, which has so far been neglected to focus only on stationary properties. These classification algorithms are applied to the time series data of weak quantum measurements (homodyne detection) of a circuit-QED implementation of the Kerr resonator coupled to a superconducting qubit. This demonstrates how machine learning methods enable a faster and more reliable measurement protocol in critical open quantum systems.
In this paper, a scalable Quantum Machine Learning (QML) algorithm is introduced to perform large-scale data classification and optimization with limited quantum resources. The method combines the amplitude-based encoding of quantum features with quantum circuits with parameters and hybrid optimization with RMSProp to lead to a high quality of accuracy and scalability in comparison with traditional machine learning models. The model has been run on 16-qubit NISQ processors with Qiskit and PennyLane frameworks and compared to the classical neural and kernel-based systems. Findings show that the suggested approach had a mean classification accuracy of 96.42 % better than classical models by an average of 5.2 % and with the same level of computation efficiency. The quantum kernel and entanglement optimization drastically enhanced high-dimensional similarity measurement and the noise mitigation strategies inherent in it also increased fidelity to 0.93. A test of the scalability of the system was a stable performance with sizes of up to 10,000 samples in datasets, with a sub-linear increase in time. These results confirm that quantum-enhanced architectures have the ability to learn complicated patterns and still be resource-efficient. The article forms a solid basis of the application of QML to other fields, including bioinformatics, finance, and IoT analytics where the dimensions of data and computation requirements are intensive.
Training parameterized quantum circuits requires balancing optimization progress with limited evaluation budgets and the statistical uncertainty of near-term quantum hardware. We present a modular benchmarking framework that separates quantum search-direction estimation from classical parameter-update rules, allowing their interactions and sensitivities to be examined independently. We evaluate a suite of gradient-based, stochastic, and derivative-free optimizers across combinatorial optimization using the Quantum Approximate Optimization Algorithm (QAOA), supervised quantum machine learning using Iris classification and a quantum convolutional neural network (QCNN) for binary MNIST classification, and quantum chemistry using the variational quantum eigensolver (VQE) for molecular hydrogen. By separating objective evaluations from sample-circuit costs under finite-shot simulation and physical execution on a 156-qubit processor, we compare optimizer behavior across workloads with different parameter counts and measurement requirements. Four selected simulator and hardware case studies report terminal and best observed objectives separately, alongside post-training classifier accuracies. The comparisons include selected MaxCut and hydrogen trajectories and illustrate workload-dependent objective changes and evaluation costs, while the simulator benchmark compares both peak and mean performance across three seeds. The hardware runs do not establish an optimizer ranking or isolate a causal effect of device noise.