Quemix, Toyota, Toyota Central R&D Labs, and the University of Tokyo Demonstrate Efficient 'Task Distribution' in Quantum Chemistry Calculations Using Classical-Quantum Hybrid Computers
A joint research team including Quemix, Toyota, Toyota Central R&D Labs, and the University of Tokyo has demonstrated an efficient task distribution method for hybrid classical-quantum computing in quantum chemistry. By using the Density Matrix Renormalization Group (DMRG) method on classical computers to maximize precision before passing the state to the proprietary 'PITE®' quantum algorithm, the team significantly reduced computational costs.
📋 Article Processing Timeline
- 📰 Published: June 1, 2026 at 20:15
- 🔍 Collected: June 1, 2026 at 11:35
- 🤖 AI Analyzed: June 1, 2026 at 18:33 (6h 58m after Collected)
Quemix, a subsidiary of TerraSky specializing in quantum computer algorithm and software R&D, along with Toyota Motor Corporation, Toyota Central R&D Labs, and the University of Tokyo's Graduate School of Science, have announced the results of a joint research project on efficient task distribution between classical and quantum computers for quantum chemistry calculations. This study presents a new guideline for the 'efficient allocation of computational resources' by combining the Density Matrix Renormalization Group (DMRG) method with the Probabilistic Imaginary-Time Evolution (PITE®) algorithm, maximizing the strengths of both device types.
[Background: The Challenge of 'Initial State Preparation' in Quantum Chemistry]
In recent years, high-precision quantum chemistry calculations using quantum computers have gained significant attention. Because molecular electronic states follow quantum mechanics, they can be naturally represented on qubits, allowing for precise simulation of complex electronic behaviors that are difficult for classical computers. This is expected to accelerate the development of high-performance new materials.
The core theme of quantum chemistry is identifying the 'ground state' with minimum energy. Knowing the ground state allows for the extraction of various physical properties, such as molecular reactivity and experimental spectra. While Quantum Phase Estimation is a well-known algorithm for ground state calculations, Quemix has proposed its own algorithm, 'Probabilistic Imaginary-Time Evolution (PITE®),' to further accelerate calculations. However, 'State Preparation' remains a common challenge. Often, the solution from mean-field approximation calculations performed on classical computers is used as the initial state. The more difficult a problem is for mean-field approximation, the greater the benefit of high-precision quantum computing. However, in such cases, the mean-field solution is far from the true ground state. Using this as an initial state leads to enormous computational costs (time and error risk) on the quantum computer. Therefore, preparing an initial state close to the true solution is key to practical application.
There were two major problems in preparing good initial states on quantum computers. First, seeking a good initial state on classical computers leads to exponentially increasing computational loads for large systems. Second, even if a good quantum state is prepared classically, encoding it onto a quantum computer incurs massive costs. Thus, the optimal task distribution between classical and quantum devices remained unclear. As hybrid computing becomes a next-generation foundation, clarifying this distribution is a critical question.
[Research Content: Finding the 'Best Balance' Between Classical and Quantum]
This study empirically clarified the boundary of 'how much to prepare on a classical computer and where to delegate to a quantum computer.'
■ Results
Thorough 'Refinement' by Classical Computers: The study showed that processing as much as possible on the classical side is key to overall efficiency. Specifically, using the DMRG method, the team approached the true ground state to the 'limit' allowed by classical memory and cost.
Seamless Connection and Quantum Benefits: The high-precision state obtained classically is encoded as a Matrix Product State (MPS) into a quantum circuit and handed over to Quemix's proprietary 'PITE®' algorithm.
Breaking Classical Limits with Quantum: For large-scale problems, relying solely on classical computers leads to resource exhaustion. By adopting a role-sharing approach—approaching the true solution as closely as possible classically and then delegating the rest to the quantum computer—the team demonstrated the possibility of reaching the 'true solution' that was previously unattainable.
[Significance: Practical Guidelines for Hybrid Operations]
This research reaffirms that while quantum computers are powerful, their benefits cannot be fully realized without a sufficient initial state. Providing a concrete balance for allocating classical and quantum resources is a major achievement toward practical application. Quemix will continue to promote algorithm development to solve societal issues through materials development.
[Background: The Challenge of 'Initial State Preparation' in Quantum Chemistry]
In recent years, high-precision quantum chemistry calculations using quantum computers have gained significant attention. Because molecular electronic states follow quantum mechanics, they can be naturally represented on qubits, allowing for precise simulation of complex electronic behaviors that are difficult for classical computers. This is expected to accelerate the development of high-performance new materials.
The core theme of quantum chemistry is identifying the 'ground state' with minimum energy. Knowing the ground state allows for the extraction of various physical properties, such as molecular reactivity and experimental spectra. While Quantum Phase Estimation is a well-known algorithm for ground state calculations, Quemix has proposed its own algorithm, 'Probabilistic Imaginary-Time Evolution (PITE®),' to further accelerate calculations. However, 'State Preparation' remains a common challenge. Often, the solution from mean-field approximation calculations performed on classical computers is used as the initial state. The more difficult a problem is for mean-field approximation, the greater the benefit of high-precision quantum computing. However, in such cases, the mean-field solution is far from the true ground state. Using this as an initial state leads to enormous computational costs (time and error risk) on the quantum computer. Therefore, preparing an initial state close to the true solution is key to practical application.
There were two major problems in preparing good initial states on quantum computers. First, seeking a good initial state on classical computers leads to exponentially increasing computational loads for large systems. Second, even if a good quantum state is prepared classically, encoding it onto a quantum computer incurs massive costs. Thus, the optimal task distribution between classical and quantum devices remained unclear. As hybrid computing becomes a next-generation foundation, clarifying this distribution is a critical question.
[Research Content: Finding the 'Best Balance' Between Classical and Quantum]
This study empirically clarified the boundary of 'how much to prepare on a classical computer and where to delegate to a quantum computer.'
■ Results
Thorough 'Refinement' by Classical Computers: The study showed that processing as much as possible on the classical side is key to overall efficiency. Specifically, using the DMRG method, the team approached the true ground state to the 'limit' allowed by classical memory and cost.
Seamless Connection and Quantum Benefits: The high-precision state obtained classically is encoded as a Matrix Product State (MPS) into a quantum circuit and handed over to Quemix's proprietary 'PITE®' algorithm.
Breaking Classical Limits with Quantum: For large-scale problems, relying solely on classical computers leads to resource exhaustion. By adopting a role-sharing approach—approaching the true solution as closely as possible classically and then delegating the rest to the quantum computer—the team demonstrated the possibility of reaching the 'true solution' that was previously unattainable.
[Significance: Practical Guidelines for Hybrid Operations]
This research reaffirms that while quantum computers are powerful, their benefits cannot be fully realized without a sufficient initial state. Providing a concrete balance for allocating classical and quantum resources is a major achievement toward practical application. Quemix will continue to promote algorithm development to solve societal issues through materials development.
FAQ
What is Quemix's role in this project?
Quemix provides the proprietary PITE® quantum algorithm and leads the research on quantum chemistry algorithms.