A groundbreaking advancement in the classical simulation of quantum circuits has emerged from a collaborative effort between The University of Osaka’s Center for Quantum Information and Quantum Biology (QIQB) and Fixstars Corporation. This pioneering research has successfully demonstrated one of the largest-scale classical simulations of iterative quantum phase estimation (IQPE) circuits applicable to quantum chemistry. By utilizing an unprecedented scale of computational resources—specifically up to 1,024 NVIDIA H100 GPUs—the team has surpassed the previous upper boundary of 40-qubit quantum circuit simulations, unlocking new horizons in simulating molecular systems for quantum algorithm development.
Quantum computing stands poised as a transformative technology for solving complex quantum chemical problems that classical supercomputers struggle to address efficiently. The accurate simulation of molecular interactions, electronic structures, and reaction dynamics lies at the core of drug discovery and materials science. Yet, the current limitations in both hardware and algorithms restrict the scale and complexity of these simulations. Fault-tolerant quantum computers (FTQCs), which are still under active development, promise to breach these limits. However, to prepare for their eventual functional deployment, robust quantum algorithms must be developed and rigorously validated. The classical simulation of quantum algorithms such as IQPE before deploying on quantum hardware is critical to this preparatory phase.
Iterative Quantum Phase Estimation (IQPE) is a quantum algorithm instrumental in extracting eigenvalues of unitary operations, pivotal in leveraging quantum systems to solve eigenvalue problems prevalent in quantum chemistry. Its strength lies in requiring fewer qubits relative to traditional QPE approaches, making it a favorable candidate for near and medium-term quantum hardware architectures. Professor Wataru Mizukami and his research team concentrated their efforts on implementing IQPE within the quantum circuit simulator specially designed for quantum chemistry applications, known as “chemqulacs-gpu.” This software framework orchestrates quantum chemistry simulations over GPU clusters, enabling high-fidelity quantum circuit emulation.
Concomitant with software innovations, the research group developed an advanced parallel computing methodology tailored to harness the immense computational power of large-scale GPU clusters effectively. The dense inter-GPU communication and synchronization requirements usually become a formidable bottleneck when simulating complex quantum circuits at scale. By innovating in interconnect optimization and load balancing across 1,024 GPUs within AIST’s ABCI-Q infrastructure, the team overcame these challenges. The result is a scalable, highly efficient classical simulation framework capable of pushing quantum circuit simulations beyond the previously imposed 40-qubit limit.
The simulations conducted propelled the computational chemistry frontier, achieving simulation of a 42-spin-orbital system representing an H₂O molecule with qubit reduction techniques applied. This advancement allows intricate modeling of molecular electronic states with unprecedented accuracy. Additionally, the team simulated a 41-qubit quantum circuit benchmark involving an Fe₂S₂ molecule, marking a milestone that tests pure circuit-scale performance. Such achievements highlight the remarkable progress in simulating realistic, chemically significant molecular systems, paving the way for more intricate quantum algorithm testing and refinement.
From a computational architecture perspective, leveraging 1,024 NVIDIA H100 GPUs required intricate parallelism and data management strategies. The GPUs, renowned for their tensor cores and high throughput, provided the raw computing capability essential for managing the enormous state vectors characteristic of these high-qubit simulations. Employing sophisticated inter-GPU communication protocols, mesh topologies, and communication schedules, the team optimized the bandwidth and latency constraints inherent in traditional large-scale distributed systems. This meticulous engineering was indispensable for sustaining the performance and accuracy of the quantum circuit simulation at scale.
Such classical simulations provide a pivotal intermediate step in quantum computing evolution. By extending the boundaries of classical simulation capacity, researchers can rigorously benchmark quantum algorithms, discover potential error sources, and develop error mitigation strategies prior to implementation on quantum hardware. The insights gained contribute directly to the design of more robust, scalable quantum algorithms while informing hardware architects on the practical requirements for future fault-tolerant quantum processors.
Professor Mizukami remarked on the technical rigor and perseverance involved in this endeavor, noting the challenges of coordinating a vast distributed system within limited computational windows while troubleshooting emergent issues. The dedication of the promising researchers, Yusuke Teranishi and Shoma Hiraoka, alongside the critical support from ABCI-Q operations staff, was fundamental to achieving these large-scale simulation outcomes. Their success not only exemplifies the synergy between academia and industry but also invigorates the quest toward practical quantum algorithm deployment.
The significance of this research also underscores the synergy between QIQB’s academic expertise and Fixstars Corporation’s specialization in GPU performance optimization. While the QIQB team focused on the simulation algorithms and quantum chemistry implementation layers, Fixstars introduced GPU performance profiling and code optimization tailored for the ABCI-Q platform. Their collaborative effort resolved intricate inter-GPU bottlenecks, facilitating robust scalability and efficient resource utilization that underpins these unprecedented simulation scales.
This breakthrough was formally presented at NVIDIA GTC 2026 in San Jose, California, where the team detailed methodologies and performance metrics under the session titled “Efficient Iterative QPE Simulations for Quantum Chemistry using Distributed State-Vector Methods.” The presentation highlighted innovative techniques in computational simulation, GPU cluster coordination, and quantum algorithm emulation, offering a pathway for other research groups to replicate or extend these advancements.
Funding for this research was provided by prominent Japanese institutions, including the Japan Society for the Promotion of Science, Ministry of Education, Culture, Sports, Science and Technology, and Japan Science and Technology Agency. Computational resources were made available through the National Institute of Advanced Industrial Science and Technology’s (AIST) Quantum-AI Hybrid Computing Infrastructure (ABCI-Q) under the “ABCI-Q Grand Challenge” program, which facilitates cutting-edge scientific computation on powerful GPU clusters.
This milestone not only bridges current technological gaps in quantum chemistry simulations but also provides critical validation tools necessary for quantum algorithm researchers globally. By expanding the complexity and scale of classically simulatable quantum circuits, the research accelerates the development pipeline toward practical quantum advantage in industrial and scientific applications. Ultimately, such progress enhances our capacity to tackle grand challenges in drug discovery, sustainable materials, and quantum information sciences through hybrid classical-quantum computational paradigms.
Subject of Research: Not applicable
Article Title: Efficient Iterative Quantum Phase Estimation Simulation Breaks 40-Qubit Barrier for Quantum Chemistry
News Publication Date: Not specified (expected presentation at NVIDIA GTC 2026)
Image Credits: QIQB, The University of Osaka
Keywords: Computational science, Quantum computing, Quantum algorithms
