Skip to content
Quantum.world

The Signal

Quantinuum and JPMorganChase report error-corrected algorithm runs

A paper published in Physical Review X on November 19 reports fault-tolerant-gadget runs of QAOA and HHL on Quantinuum trapped-ion processors. Its largest QAOA trial used 12 logical qubits, 97 physical qubits and 2,132 physical two-qubit gates.

Quantinuum and JPMorganChase report error-corrected algorithm runs

A team from Quantinuum and JPMorganChase has run two quantum algorithms using error-corrected logical qubits on trapped-ion processors. The paper reports that its largest QAOA trial used 12 logical qubits, 97 physical qubits and 2,132 physical two-qubit gates. Published November 19 in Physical Review X, the study reports tests on Quantinuum H2 and Helios trapped-ion processors using the [[7,1,3]] Steane code.

Source journals.aps.org

The authors describe the algorithms as built from fault-tolerant circuit components, but say they did not establish fault tolerance in the stronger, whole-computation sense. The distinction matters to how the result is described: the paper demonstrates complete algorithm runs assembled from fault-tolerant gadgets, while noting that the full computation did not meet that stronger standard.

Two algorithms on encoded qubits

The tests covered the Quantum Approximate Optimization Algorithm, or QAOA, and the Harrow-Hassidim-Lloyd algorithm, or HHL. The QAOA workloads addressed small instances of low autocorrelation binary sequences and portfolio optimization; HHL was applied to a lattice Poisson equation that the paper describes as a structured toy problem.

The Steane code encodes one logical qubit in seven physical qubits and has code distance three. The team ran the experiments on Quantinuum H2-1, H2-2 and Helios systems. Its approach combined encoded operations with measurement-dependent feedback in dynamic circuits, along with state-preparation routines that could repeat until they succeeded.

For QAOA circuits with five and six logical qubits, the researchers report better performance as they increased the number of algorithm layers and the number of T gates used to approximate logical rotations, despite the added physical circuit complexity. They also report that circuits with up to eight logical qubits and nine logical T gates performed similarly to unencoded circuits.

Results and limits

The largest QAOA case was a 12-variable portfolio optimization problem on Helios. It used 97 physical qubits and, in its deeper tested circuit, 2,132 physical two-qubit gates. The paper reports that both tested circuits performed better than random guessing, although the deeper circuit performed worse than the shallower one.

The authors caution that this portfolio instance was highly simplified and that its circuits were produced in a non-scalable way. They report that increasing the number of allowed attempts in repeat-until-success state preparation reduced the discard rate, while active quantum error-correction cycles improved the HHL circuit’s logical fidelity. In examining HHL, they also identify dynamic program-compilation artifacts as a performance bottleneck.

The paper reports a fault-tolerant logical T-gate implementation with an infidelity of 2.7(4) × 10−3. On H2-1, its Ramsey-sequence benchmark measured fidelity of 0.960(6) for a sequence containing 16 T gates.

Sources

  1. Original source · Paper American Physical Society journals.aps.org

Other pages read

  1. arXiv arxiv.org

Related stories

Explore

More articles