IonQ, NVIDIA and quantum software startup qBraid have published joint research showing a 54% reduction in logical error rates on deep Trotterized quantum chemistry simulations, executed on IonQ's Barium trapped-ion hardware and NVIDIA's GH200 Grace Hopper Superchip.
Beating the deep Trotter dilemma
Announced on September 1, 2026, the paper attacks a long-standing problem in near-term quantum chemistry: as Trotterization slices continuous time evolution into deep gate sequences, physical noise accumulates exponentially and destroys the target signal. The joint team replaces long non-local Jordan-Wigner strings with local Majorana-loop stabilizers using a technique called Generalized Superfast Encoding, then combines it with a Clifford Noise Reduction protocol that teleports verified Clifford operations onto the data register before hook errors propagate.
Active mid-circuit measurement is the trick
The most striking finding: the 54% fidelity improvement drops entirely to zero when stabilizer readouts are deferred to the end of the circuit. That confirms active, mid-circuit fault detection — measuring ancilla qubits mid-flight, resetting them and correcting corrupted operations dynamically — as the mechanism driving the quantum advantage, rather than post-processing.
ML-guided stabilizer selection on GH200
To choose which stabilizer pairs to measure across deep circuits, the researchers trained a machine-learning model on an NVIDIA GH200 GPU using 57,536 samples with 992 pairs per graph. During inference the model rapidly scored 105 candidate pairs to pick optimal verification operators, comfortably outperforming random selection. The 6-qubit encoded simulation was executed on a Barium-based IonQ development system similar to the forthcoming IonQ Tempo architecture, using NVIDIA's CUDA-Q and cuStabilizer libraries.
The work slots into an increasingly fast IonQ execution cadence — the company recently paired real-time quantum error correction decoding with an Apple M4 Max and signed a co-design MOU with Sandia National Laboratories.
Reporting based on coverage from Quantum Computing Report, IonQ Blog, arXiv and Quantum Zeitgeist.