IBM and Rutgers run nearly 5,000 measure and reset cycles across 100 qubits mid calculation

[CMPunk96]

Researchers from Rutgers University, IBM and several other institutions have shown a quantum processor repeatedly measuring and resetting its qubits while calculations carry on, Interesting Engineering reports. The work is published in Nature Physics. It ran on an IBM Quantum Heron processor with 156 qubits, using a connected chain of up to 100 of them. The headline's 100 qubit chip refers to that chain, not a separate new chip

The experiment involved two competing processes taking turns along the chain. One spread disorder between neighbouring qubits. The other measured individual qubits and pushed them back towards a set state. In total, the run used nearly 5,000 two qubit operations and almost 5,000 measurement and reset cycles

By changing how often each process happened, the team found a phase transition between chaotic and controlled behaviour. When scrambling dominated, the system stayed chaotic. When measurement and reset dominated, it moved towards an orderly, controllable state. The sharpest change came near an even balance, where a small adjustment in relative frequency triggered a dramatic shift across the whole system

The transition held up in a quantum system, where it had not been certain it would survive. It also became clearer in larger systems, matching theoretical predictions. The team tested systems too large to simulate on classical computers, which strengthened confidence in the explanation. IBM researcher Maika Takita described coordinating operations, measurements and resets thousands of times across a large system as an important step towards practical error correction

The article is careful to say this does not establish a fully fault tolerant quantum computer. What it does show is that current hardware can step in repeatedly while a calculation is running. That is exactly what error correction will need, since errors have to be caught and fixed on the fly. Real time feedback like this could help future processors manage errors through long calculations

How close does this bring us to practical error correction? And is IBM still the one to beat on superconducting hardware?

IBM's 100-qubit chip advances self-correcting quantum computersinterestingengineering.com