Variational quantum algorithms offer a promising route towards practical quantum computation, but currently demand the most powerful, and therefore limited, quantum hardware to deliver reliable results. Yuqian Huo, David Quiroga, Anastasios Kyrillidis, and Tirthak Patel, all from Rice University, present a new technique, Nest, that fundamentally alters how these algorithms operate. Their work demonstrates that by strategically shifting the algorithm’s reliance between low and high-fidelity qubits during computation, they not only maintain accuracy and accelerate the process of finding optimal solutions, but also unlock the potential for running multiple algorithms simultaneously on a single computer. This achievement represents a significant step towards maximising the performance and utility of near-term quantum devices, effectively balancing accuracy, speed, and overall system efficiency.

Adaptive Optimisation for Variational Quantum Algorithms

Variational quantum algorithms (VQAs) hold significant promise for demonstrating quantum advantage on near-term quantum computers. Recent research highlights that optimising solely for performance overlooks crucial aspects of VQA execution, namely convergence speed and overall system efficiency. This work introduces NEST, a novel optimisation strategy that simultaneously improves performance, convergence, and system throughput. NEST achieves this by adaptively selecting the optimal quantum circuit ansatz and optimiser parameters based on the specific problem instance and hardware characteristics, effectively balancing competing objectives. The method employs a multi-objective optimisation framework considering both the energy expectation value and the optimisation steps required to reach a satisfactory solution, leading to significant improvements in resource utilisation and scalability. Through numerical simulations on quantum chemistry problems, the team demonstrates that NEST consistently outperforms state-of-the-art VQA implementations, achieving up to a two-fold improvement in performance and throughput while maintaining comparable convergence rates.

Variational Algorithms and Quantum Error Mitigation

Current research in quantum computing focuses heavily on optimising variational quantum algorithms (VQAs) and mitigating the effects of noise. A range of techniques are being explored, including methods for error detection and correction, strategies like Qismet which navigates dynamic noise landscapes, and VAQEM, a variational approach to quantum error mitigation. Other approaches, such as Clapton, CutQC, and Relaxed Peephole Optimization, aim to improve algorithm performance through problem transformation and circuit optimisation. Researchers are also investigating techniques to enhance measurement fidelity, like Minimum Clique Cover, and mitigate crosstalk through commutativity-based instruction reordering.

Addressing barren plateaus, a common issue in VQAs where gradients vanish, is another key area of focus. Circuit concurrency, which accelerates VQAs by exploiting parallelism, and advanced circuit compilation techniques like Paulihedral and QuCloud are also under investigation. Furthermore, researchers are developing tools like PARALLAX and PachinQo for modelling specific hardware architectures, such as neutral atom quantum computers. Noise-adaptive search, or QuantumNAS, aims to find robust quantum circuits less susceptible to noise. Research extends to quantum hardware and architecture, exploring heterogeneous microarchitectures and superconducting and Rydberg atom quantum computers.

Qubit mapping, assigning logical qubits to physical qubits, and crosstalk mitigation are crucial considerations. Variability-aware policies account for differences in qubit performance. Resource management and scheduling, with techniques like multi-device job scheduling (Qoncord), improve reliability. Understanding job characteristics and managing cloud quantum computing resources are also key areas of focus. Quantum machine learning, particularly quantum autoencoders for anomaly detection (Quorum), is also being explored.

Underlying these advancements are fundamental optimisation techniques, including gradient descent and low-precision training inspired by deep learning. Scheduling algorithms, also inspired by deep learning training, are used to optimise quantum computations. Tools like SciPy, CAFQA, and Qutracer support these efforts. Key themes emerging from this research include addressing noise, hardware-software co-design, efficient resource management, and leveraging classical techniques to enhance quantum computing. This research highlights the breadth and depth of current efforts to unlock the potential of quantum computing.

Dynamic Qubit Mapping Improves VQA Performance

Scientists have developed a technique, NEST, that dynamically adjusts quantum circuit mapping during variational quantum algorithm (VQA) execution, leveraging the varying fidelity of qubits within a single quantum computer. This innovative approach moves beyond static mapping or switching between predefined configurations, instead progressively adapting qubit assignments to improve overall performance. Experiments reveal that NEST not only enhances performance, bringing results closer to the optimal value, but also accelerates convergence, reducing the time required to reach a solution. This is achieved through a structured qubit walk, a methodical and incremental remapping of individual qubits that avoids disruptive changes to the optimisation process.

The technique capitalises on the heterogeneous noise profiles found in superconducting qubit architectures, where individual qubits exhibit differing levels of noise and fidelity. Furthermore, NEST enables concurrent execution of multiple VQAs by assigning non-overlapping sets of qubits to each job, significantly increasing system throughput. By strategically allocating qubits with appropriate fidelity to different algorithms, the team successfully co-located multiple jobs on the same quantum processor, maximising resource utilisation. This multi-programming capability addresses three key challenges in near-term quantum computing: improving VQA performance, accelerating convergence, and increasing system throughput through more effective resource allocation. The work demonstrates that dynamic fidelity variation offers a powerful strategy for optimising VQA execution and unlocking the full potential of near-term quantum hardware.

Nest Optimizes Variational Quantum Algorithm Performance

Researchers have developed a new technique, termed Nest, that significantly improves the performance of variational quantum algorithms (VQAs) on near-term quantum computers. This work addresses a key challenge in quantum computing: achieving good results despite the limitations of current hardware. Nest dynamically adjusts the qubit fidelity used during algorithm execution, leveraging the varying quality of qubits within a single device to optimise both the accuracy of solutions and the speed at which they are found. The team demonstrated that by carefully managing qubit fidelity, Nest not only enhances the quality of VQA outcomes but also accelerates convergence, achieving improvements of 12.

7% and 47. 1% compared to existing methods. Furthermore, Nest enables the concurrent execution of multiple VQAs on the same computer, substantially increasing system throughput and offering a pathway to scalable quantum computation. This approach represents a simple yet powerful strategy for unlocking the potential of VQAs on existing and near-future quantum hardware.

👉 More information
🗞 Three Birds with One Stone: Improving Performance, Convergence, and System Throughput with Nest
🧠 ArXiv: https://arxiv.org/abs/2510.09578