Sandia and Quantinuum map the noise of an error-corrected logical qubit

Started by BareMinimum, Today at 11:12 AM

Previous topic - Next topic

0 Members and 2 Guests are viewing this topic.

Topic: Sandia and Quantinuum map the noise of an error-corrected logical qubit   Views(Read 31 times)
Active members in this topic:
BareMinimum(1)

BareMinimum

Researchers have found a new way to see exactly how error corrected qubits go wrong. Quantum Computing Report says Sandia National Laboratories and Quantinuum have introduced logical gate set tomography, a technique for fully characterising the operations performed on an error corrected logical qubit

Gate set tomography is normally used on individual physical qubits. Here it was applied to a logical qubit encoded in Quantinuum's 20 qubit H1-1 trapped ion processor using the Steane code, with errors corrected in real time. Quantum Computing Report says it was the final experiment run on that machine

The team studied several logical operations, including H and S gates, an idle cycle of error correction, measurement and initialisation, and a T gate made through a process called magic state injection, which succeeded 95.9 percent of the time. Hardware infidelities ranged from around 0.1 percent for measurement to 1.2 percent for the T gate, although several error bars were larger than the values themselves

With only 50 shots on real hardware, the results fitted a simple random error model. A much larger run of 14,100 shots on Quantinuum's emulator did find significant memory effects in the noise, along with small coherent over rotations, which the authors trace mainly to dephasing while ions are moved and cooled. They also found that the error correction process itself can create these memory effects at the logical level, even when the physical noise has none

Understanding logical errors in detail matters, because error correction only works if engineers know what kind of noise they are fighting. The most precise findings come from emulation rather than hardware, and the work is a preprint. Still, it gives the field a valuable new diagnostic tool

How important is detailed noise characterisation for fault tolerant quantum computing? And can memory effects in noise be engineered away?