Modeling a nitrogen-vacancy center with NVIDIA CUDA-Q Dynamics: University of Washington Capstone Project

Defects in crystals are leading qubit candidates for quantum information networks. The ability to efficiently model defects, like nitrogen-vacancy (NV) centers, and their environment would deepen our understanding of these complex quantum systems and accelerate the use of these defects in scalable quantum networks. To study complex quantum systems like these and advance basic research to practical implementation requires both advanced tools and skilled practitioners. To train the next generation of quantum technologists, the Accelerating Quantum-Enabled Technologies (AQET) program at the University of Washington (UW) offers students a multidisciplinary curriculum that builds on the existing quantum information science and engineering offerings at UW. The program helps students consolidate their understanding of quantum technologies across physics, chemistry, materials science, engineering, and computer science. As part of the program, students complete a 10-week industry mentorship project. Last year, students working with AWS mentors used Amazon Braket to analyze the performance of the quantum approximate optimization algorithm (QAOA) in the presence of noise.

In this post, we’re excited to share the results of the 2025 AQET capstone project sponsored by Amazon Web Services (AWS) and NVIDIA. Students used Amazon Braket notebook instances with custom Jupyter kernels and NVIDIA CUDA-Q to model a NV center in diamond. CUDA-Q Dynamics is a new library within the NVIDIA CUDA-Q platform that enables performant and scalable simulations of time-domain quantum processes that describe how quantum systems evolve in time according to their interactions with other quantum objects, control pulses, noise, and the environment. Researchers can leverage CUDA-Q to design higher fidelity quantum hardware, elucidate underlying noise processes, and engineer better control sequences for qubits.

The Technology Challenge

A defect qubit must possess both a ground state electron spin to store quantum information and optical states to transmit this information. On top of the requirements for individual qubits, schemes to scale up quantum networks require multiple qubits at every node. Fortunately, a single defect can act as a multi-qubit node by harnessing the nuclear spins in its environment. The position and interaction of these nuclear spins is random and depends on the concentration of the targeted spin isotope. The NV center in diamond is the most advanced defect studied today for quantum networks. However, every NV diamond center has a unique nuclear spin environment with fully quantum mechanical interactions.

Project Implementation

The students studied the dynamics of an NV center subject to the presence of near-by carbon-13 nuclear spins. The system modeled included a single coherently coupled carbon-13 nuclear spin, a single electron spin, a single nitrogen-14 nuclear spin, and an incoherently interacting bath of carbon-13 nuclear spins incorporated as pure dephasing. A Ramsey protocol, a series of predefined pulse sequences, was simulated by constructing a Hamiltonian that described the quantum mechanical interactions of the system and applying the appropriate pulses.

Students used Amazon Braket Jupyter notebooks and CUDA-Q Docker containers to develop a CUDA-Q application. With the Amazon Braket service the students were able to use cost-effective ml.p3.2xlarge instances with a single NVIDIA GPU for testing and development with the option to scale up to 8 NVIDIA GPUs on ml.p3.16xlarge instances. As the complexity of the system increased, in the form of additional carbon-13 nuclear spins, so did the computational costs of these experiments. The students were able to leverage P4, P5, and P6 instances accelerated by NVIDIA A100, H100, and Blackwell GPUs, respectively. Using all 8 GPUs per instance connected with NVIDIA NVLink & NVLink Switch, the following reductions in computational runtime were observed for systems with 10 carbon-13s P4 to P5: 42%, P4 to P6: 64%, and P5 to P6: 38%. This included making use of latest NVIDIA Blackwell GPUs – which are the most performant tools available for performing comprehensive simulations of qubit dynamics.

Summary

In this blog, we shared the results of the 2025 AQET capstone project sponsored by AWS and NVIDIA. UW students used Dynamics functionality within NVIDIA CUDA-Q Platform, running on AWS, to model the dynamics of a NV center in diamond a promising qubit candidate for quantum information networks. The same services and tooling can be used to develop novel dynamics experiments that can be run on AWS at scale. If you are a member of an accredited research intuition and would like to get started running your own dynamics simulations check out our Cloud Credit for Research Program.

Acknowledgements

The AQET program is sponsored by NSF award 2021540, NRT-QL: Accelerating Quantum-Enabled Technologies.