, ,

IonQ and Synopsys Cut LS-DYNA Simulation Time by Up to 14.6%

Jessica Nicole
IonQ and Synopsys Cut LS-DYNA Simulation Time by Up to 14.6%

IonQ and Synopsys published today (September 17, 2026) a hybrid-quantum result that lands squarely in industrial CAE: embedding an advanced quantum algorithm inside Synopsys’s Ansys LS-DYNA stack cut total simulation time by up to 14.6% on large dynamic models. The work earned a 1st Place Best Paper Award at IEEE Quantum Week 2026 in Toronto and targets the same bottleneck every crash, aero, and structural desk knows—reorganizing enormous systems of equations before the solver ever starts marching.

Large product simulations (virtual crash tests, aerodynamic loads, jet-engine assemblies) routinely solve systems with hundreds of millions of variables. Depending on setup, classical clusters spend heavy time on avoidable fill-in and reorder work. IonQ and Synopsys plugged a quantum-assisted sorting step into that preprocessing path so the hybrid system acts like a traffic cop: it finds a better data organization once at the start, then the classical LS-DYNA run pays that savings repeatedly across the rest of the job.

Vertical wind tunnel facility used for aerodynamic validation—physical counterpart to large CFD and structural CAE campaigns. Photo: Wikimedia Commons.
Physical tunnels and crash rigs still set the ground truth; the win is shaving days off the digital runs that decide which hardware test to fund. Photo: Wikimedia Commons (rights-safe).

Dr. Martin Roetteler, IonQ VP of Quantum Solutions and co-author, put the industrial math plainly: jet-engine and automotive crash simulations can occupy large classical clusters for days, and a 14.6% wall-clock cut on a seven-day stress test is roughly a full day of continuous compute returned. Across the suite—an automobile, an industrial drill component, a fluid impeller, and a jet-engine assembly—improvements were at least 5.9%, peaking at 14.6% for complex dynamic cases, with meshes up to 35 million points.

The numerical campaign used up to 150 qubits in simulation, with physical validation on IonQ’s 36-qubit Forte trapped-ion machine. Synopsys SVP of Innovation Prith Banerjee framed the partnership as near-term NISQ-era acceleration inside tools customers already run, not a distant fault-tolerant rewrite of every solver. That is the CAE buyer’s question: does the quantum step drop into the incumbent stack without breaking audit, mesh, and material cards?

Fine finite-element mesh on a rectangular domain in Ansys-class FEA software. Photo: Wikimedia Commons.
Mesh quality and equation ordering still decide whether a CAE night shift finishes before the design review. Photo: Wikimedia Commons (rights-safe).

Same week, the design-desk side of CAE moved too: SimScale launched an Engineering AI Agent inside PTC Onshape at IMTS 2026, reasoning through CFD, thermal, Emag, and FEA setup from natural-language prompts without leaving CAD. Together the headlines say simulation is being pulled earlier into the decision—and squeezed harder at the HPC end.

For N23D, the bar stays one native Rust binary and a digital thread that owns the model, the mesh assumptions, and the results you ship—whether the accelerator is a quantum reorder inside LS-DYNA or an agent proposing boundary conditions in CAD. Faster CAE only helps if every run stays auditable back to the geometry and materials you actually own, not a rented black box that drifts from the PLM vault.

Sources/References