IBM Quantum Processor Completes Million-Sample Task in 19 Seconds

A US-based research team, BlueQubit, has used IBM's Nighthawk r2 quantum processor to perform a random quantum circuit sampling experiment, generating one million samples in a mere 19 seconds. Researchers estimate that replicating this task on a supercomputer like Frontier would take approximately 110 years. The experiment was led by Tigran Sedrakyan, a condensed matter physicist at BlueQubit.

The Nighthawk r2 is a 120-qubit superconducting quantum processor accessible via IBM's cloud platform. The team utilized 61 qubits and ran increasingly complex random quantum circuits for up to 40 cycles. The optimal point was found at 36 cycles with 918 two-qubit gates.

The significance of the experiment lies not just in the speed of sample generation, but in the requirement that these results adhere to a specific probability distribution produced by the quantum circuit, a task known to be challenging for classical computers. Random Circuit Sampling (RCS) is a benchmark task used to test the capabilities of quantum processors.

To estimate the computational effort for a classical computer, the researchers employed tensor network contraction methods. The calculations indicated that producing one million samples would require approximately 1.2 × 10²⁷ computational operations. Using the Frontier supercomputer, with a peak performance exceeding 10¹⁸ operations per second, the task is estimated to take 110 years. However, this estimate is based on a specific classical simulation algorithm.

Researchers validated the results using methods such as "trimmed" quantum circuits and mirror circuits. While RCS itself is not a direct production application, the experiment highlights the capabilities of commercial quantum processors accessible via cloud platforms. The findings have been published as a preprint on arXiv.