⚛️Quantum Motion and NVIDIA Run Full Quantum Chemistry Pipeline on a Single GPU
Quantum chemistry just got a lot cheaper to prototype
TL;DR
Quantum Motion and NVIDIA built an end-to-end quantum chemistry pipeline by adding CUDA-Q as a backend to the open-source QUICHE interface. They modeled the spin energetics of Fe(SH)4- on a single GPU and checked the results against CASSCF. If you're doing quantum chemistry research, the workflow is now open and reproducible.
Quantum Motion and NVIDIA just showed that you can run a full quantum chemistry workflow, from classical integrals to quantum phase estimation, on one GPU. They plugged NVIDIA CUDA-Q into QUICHE, the open-source interface co-developed by Quantum Motion, FACCTs, and Riverlane, and it works end to end. If you work on electronic structure problems, this matters because the pipeline takes the same problem spec you'd use for fault-tolerant resource estimation and runs it directly as GPU-accelerated QPE. You export active-space integrals from ORCA, feed them to QUICHE with your Hamiltonian and precision targets, and CUDA-Q builds the qubit Hamiltonian and dispatches the kernels. No glue code you have to maintain yourself. The test case was the iron-sulfur complex Fe(SH)4-, modeled in a five-electron, five-orbital active space. The QPE energies were verified against exact CASSCF benchmarks, and the whole run fit on a single GPU. The SC26 demo notebook and the QUICHE codebase are both available now.

Key Points
The test system was the iron-sulfur complex Fe(SH)4-, modeled within a five-electron, five-orbital active space running on a single GPU for the full pipeline.
Keep reading — it's free
Enter your email to keep reading — plus the best of AI & tech, daily. Free, forever.
Already a member? Sign in
Comments
Be the first to comment