Quantum Motion is accelerating the path towards utility-scale quantum computing by working with NVIDIA to overcome a critical bottleneck for end-to-end quantum advantage: efficiently embedding real-world data in a quantum computer, known as the state preparation problem. By combining NVIDIA accelerated computing capabilities and the CUDA-Q platform for hybrid quantum-classical computing with quantum circuit compilation tools developed as part of this project, we are enabling quantum computing workflows focusing on chemistry calculations. Our collaboration will allow researchers to translate chemical systems to executable quantum circuits via advanced tensor network techniques, enabling high-quality inputs for quantum algorithms like Quantum Phase Estimation (QPE). This will help address a significant barrier to analysing complex molecular electronic structures and bring us closer to quantum-enabled breakthroughs in drug discovery and materials science.
We have released a newly-developed, GPU-accelerated Julia package called MPSCircuits.jl. Our pipeline encodes chemical systems by using matrix product states (MPS), a type of tensor network produced classically by the Density Matrix Renormalisation Group (DMRG) algorithm, and compiling these to quantum circuits for subsequent data-loading to a quantum computer. To show this in an end-to-end example using the NVIDIA CUDA-Q platform, we have published a tutorial. Here, we guide you through the end-to-end process: defining the Hamiltonian, solving an approximate ground state with DMRG, circuit compilation via ‘unitary disentangler’ methods, and transpilation to an elementary quantum circuit in CUDA-Q.
End-to-end QPE circuit with our custom MPS-to-circuit compiler to address state preparation.
The resulting state preparation circuit is used within QPE, which is the gold-standard algorithm for the coveted goal of studying electronic ground states. QPE relies upon a ‘guide state’ that partially describes the molecule’s ground state; if the starting state is poor, the algorithm cannot extract useful information. Although quantum computers can refine coarse initial estimates to a level of accuracy beyond any classical computer, they cannot directly solve the state initialisation problem. This is where our pipeline steps in, providing a high-quality input state that minimises the quantum resources needed for a successful QPE evaluation.
Target chemical systems for our pipeline range from simple diatomics for testing up to large benchmark systems that are challenging to solve classically, such as a Ruthenium-based catalyst or FeMoco (pictured above).
With these foundations set up, we can start exploring more complex chemical systems for sectors ranging from pharmaceuticals to materials and batteries design. Indeed, we are just at the start of our collaboration, with ongoing investigations into noise-aware compilation, improved circuit architectures, and other types of data-loading. We also plan to scale state preparation to large benchmark systems, such as the FeMoco complex that is responsible for organic nitrogen fixation.
Check out part 1 and part 2 of our tutorial, and be sure to catch us at ISC 2026 in Hamburg this week!
About the authors:
Matthew Sims-Goh is an Applications Scientist at Quantum Motion. Matt obtained a DPhil in quantum computing at the University of Oxford, where he was a Rhodes Scholar. At Quantum Motion, Matt works on algorithm development and end-use applications of quantum computing, with a particular focus on state preparation and augmenting quantum algorithms with classical simulation tools.
Annina Lieberherr is an Applications Scientist at Quantum Motion. Annina obtained a DPhil in Theoretical Chemistry from the University of Oxford for research on (classical) simulation methods for spectroscopy. At Quantum Motion, Annina is working on identifying use cases for quantum computing and algorithm development, both with a focus on chemistry.
Zohim Chandani is a Quantum Algorithms and Applications Engineer at NVIDIA. His work spans scaling up quantum algorithms, use of AI across the quantum stack and the utility of quantum-GPU supercomputing.
Thomas R. Bromley is the Applications Lead at Quantum Motion, working to identify quantum computing use cases and develop the underlying quantum algorithms. Tom has over a decade of experience in quantum computing and leads a team of researchers and software developers to build applications that complement Quantum Motion’s silicon spin quantum dot hardware.