The Control Variational Quantum Eigensolver (ctrl-VQE) directly optimizes microwave pulses to enable faster and lower-error quantum-state preparation, but its continuous control landscape re- quires efficient search strategies. We demonstrate that a reinforcement-learning agent based on a deep Q learning network can autonomously discover high-performance pulse sequences using only system parameters and a reward function. The approach is fully general for superconducting qubit platforms, requires no ansatz, and operates at nanosecond resolution compatible with hardware con- straints. As a proof of concept, we apply the method to ground-state preparation of the Hydrogen molecule on a simulated superconducting device. The agent consistently identifies optimized control sequences that achieve high fidelity and outperform random-search baselines. These results highlight adaptive learning as a promising hardware-ready framework for pulse-level quantum control.
The Control Variational Quantum Eigensolver (ctrl-VQE) directly optimizes microwave pulses to enable faster and lower-error quantum-state preparation, but its continuous control landscape re- quires efficient search strategies. We demonstrate that a reinforcement-learning agent based on a deep Q learning network can autonomously discover high-performance pulse sequences using only system parameters and a reward function. The approach is fully general for superconducting qubit platforms, requires no ansatz, and operates at nanosecond resolution compatible with hardware con- straints. As a proof of concept, we apply the method to ground-state preparation of the Hydrogen molecule on a simulated superconducting device. The agent consistently identifies optimized control sequences that achieve high fidelity and outperform random-search baselines. These results highlight adaptive learning as a promising hardware-ready framework for pulse-level quantum control.
Variational quantum algorithms (VQAs) are key enablers of practical computations on near-term quantum computing hardware. However, gate-based ansatz suffer from decoherence and control errors, limiting scalability. Pulse-level approaches, such as control variational quantum eigen solver (ctrl-VQE), address these challenges by directly optimizing analog control fields and enabling time-optimal state preparation. Building on this paradigm, we present a reinforcement-learning (RL) framework that autonomously discovers high-fidelity, time-efficient control pulses. Using a hydrogen dimer testbed, our RL agent achieves state-fidelity above 0.995 and ground state energy errors near 0.005 Hartree, demonstrating the promise of AI-driven, hardware-efficient quantum algorithms.
Variational quantum sensing offers a promising route to high-precision parameter estimation, but its performance depends strongly on the circuit architectures used for probe preparation and measurement. Existing approaches typically optimize continuous parameters within predefined ansätze, restricting the accessible design space and limiting adaptation to sensing tasks and hardware constraints. Here, we introduce \textsc{AutoQSense}, a reinforcement-learning framework that searches circuit architectures using Fisher-information-based objectives. For few-qubit systems, a single agent sequentially constructs preparation and measurement circuits. For larger systems, a distributed formulation assigns local circuit design to subsystem agents and inter-block entanglement to a budgeted agent. Numerical results show that the learned architectures recover known benchmark strategies, adapt to dephasing noise, and outperform fixed hardware-efficient ansätze while using fewer entangling gates. These results establish \textsc{AutoQSense} as a resource-aware approach to adaptive and hardware-compatible quantum sensing.
Quantum optimal control theory (QOCT) typically addresses the control of physical qubits. Finding explicit pulse control sequences in such a framework is challenging, especially when an underlying physical model is unknown. We propose a deep reinforcement learning (DRL) method, which doesn't require any underlying gate model or qubit pre-calibration, capable of controlling a superconductive qubit via analog pulses acting in the IBM Qiskit Pulse environment. We applied the method to build a single-qubit gate with high fidelity and short duration at pulse level. In particular, the DRL agent approximated the X90 gate at the physical layer on the IBM Armonk transmon superconductive qubit simulated by the Qiskit Pulse simulator. The learned sequence has an average gate fidelity greater than 0.978 and a duration of 58 ns only, faster than the default X90 pulse IBM implementation, which has a runtime of 140 ns. Without prior knowledge and gate model knowledge, the agent learned a non-traditional shaped microwave pulse, providing an alternative strategy for controlling noisy quantum states.
This paper investigates the application of reinforcement learning (RL) for adaptive quantum control. Traditional quantum control methods often rely on pre-designed pulse shapes optimized for specific scenarios, lacking adaptability to variations in system parameters or environmental noise. This work proposes a novel framework where an RL agent learns to dynamically adjust quantum control pulses in real-time, maximizing the fidelity of quantum operations. The agent receives feedback from the quantum system, typically measured through state variables, and utilizes this information to optimize the control pulse sequence. This adaptive approach addresses the limitations of static control schemes and promises enhanced robustness and performance in complex quantum systems. The core claim centers on utilizing reinforcement learning to design optimal quantum control pulses. The central mechanism involves a trained controller dynamically adjusting pulses based on system feedback. The presented methodology offers a new approach to quantum control, particularly relevant in noisy environments and systems with parameter variations. The research demonstrates the potential of RL for achieving superior quantum control outcomes compared to conventional techniques. ---
Quantum hardware control -- the precise calibration of microwave pulses, flux biases, and laser parameters that drive qubit operations -- is a critical bottleneck in scaling superconducting and trapped-ion quantum processors. Manual calibration is time-consuming, requires expert domain knowledge, and fails to adapt to rapid parameter drift between calibration cycles. AI-assisted control systems offer automated, adaptive alternatives using reinforcement learning (RL), Bayesian optimisation (BO), and neural network surrogate models to continuously optimise hardware control parameters. This paper proposes the AI Quantum Control Framework (AQCF), evaluating six AI control strategies across five hardware optimisation tasks: single-qubit gate calibration, two-qubit gate optimisation, crosstalk suppression, readout fidelity maximisation, and real-time drift compensation. AQCF is evaluated on IBM Quantum Eagle and IonQ Aria hardware over 60-day continuous operation. Key results: RL-based pulse optimisation improves two-qubit gate fidelity by 18.4% over manual calibration; Bayesian optimisation reduces calibration time by 84% (from 4.2 hours to 40 minutes); neural surrogate models enable real-time drift compensation with 12.8-minute response latency. The framework introduces the AI Control Efficiency Score (ACES) and a deployment roadmap for autonomous quantum hardware management.
Real-time adaptive control of quantum systems requires rapid generation of robust, high-fidelity pulses across a continuous range of operating conditions. Standard optimization algorithms such as gradient-ascent pulse engineering (GRAPE) solve each instance independently, discarding information between runs and requiring costly reinitialization when parameters change. We present an approach to robust optimal quantum control based on model-based reinforcement learning, in which a single neural network -- embedding the Hamiltonian directly into the training pipeline -- generates robust gates across an entire family of gate configurations, without pre-computed training data. Demonstrated on a single-spin (two-level) system, the trained networks produce pulses for arbitrary rotation angles over a range of pulse durations, detunings, and field inhomogeneities in milliseconds, at fidelities comparable to multi-seed GRAPE. The framework is inherently adaptable: any parameter entering the Hamiltonian can serve as a network input, extending the approach to different systems and control settings. Beyond speed, the network reveals structure in the control landscape: it discovers the same structured phase profiles that appear in GRAPE solutions -- made identifiable through fidelity-invariant symmetry transformations -- but more consistently than independent optimization. This consistency enables smooth interpolation across the entire trained parameter space.