Reliable results from noisy quantum computers can require thousands of repeated circuit runs.
In a study posted Sept. 16, IBM researchers reported reducing that sampling overhead by as much as 63-fold by combining error detection with probabilistic error cancellation, or PEC, on the company’s ibm_aachen superconducting processor.
The result could make some calculations less resource-intensive on existing quantum hardware. IBM also frames the experiment as evidence that techniques associated with error mitigation and correction can work together before fully fault-tolerant quantum computers arrive.
What the Study Found
IBM researchers Laurin Fischer, Ali Javadi-Abhari, Simon Martiel, and Alireza Seif combined two existing techniques. Error detection flags and discards results with identifiable faults, and probabilistic error cancellation, or PEC, corrects for the errors that remain. They call the combined method spacetime PEC, as it models faults by both their location and timing within a circuit.
The team tested the method on IBM's ibm_aachen superconducting processor, using 22 data qubits and 27 check qubits across circuits with up to 648 controlled-Z gates. In that test, four error-detecting checks identified about 65% of the modeled elementary faults, according to IBM’s research blog, reducing the remaining noise that PEC had to mitigate. At six Trotter steps, the researchers estimated that the combined method would require 1,359 samples to reach the target precision, compared with 85,545 for PEC alone — approximately a 63-fold reduction for that experimental configuration.
Metric | Result |
|---|---|
Sampling overhead reduction, 6 Trotter steps | 63x |
Sampling overhead reduction, 4 Trotter steps | 15.9x |
Sampling overhead reduction, 2 Trotter steps | 3.7x |
Modeled elementary faults detected | ~65% |
Qubits used (data / check) | 22 / 27 |
Circuit depth (max controlled-Z gates) | 648 |
Built on IBM's Broader Fault-Tolerance Push
The result builds on IBM's other error-correction work this year. In a cryogenic engineering milestone toward Quantum Starling, the company connected two ultracold modules needed for a future fault-tolerant system, though it still has to demonstrate error correction itself at scale. The Quantum Computing Report notes the new study sits alongside separate hierarchical code designs. As per IBM, it could eventually support trillions of logical operations at a low physical error rate.
Part of a Broader Push Toward Usable Quantum Systems
IBM’s work is part of a broader industry effort to make quantum systems more usable, from US investments in competing quantum-computing architectures to software projects such as Fujitsu’s newly open-sourced OpenQARP package.
The paper's authors, Laurin Fischer, Ali Javadi-Abhari, Simon Martiel, and Alireza Seif, wrote in an IBM blog post, cited by The Quantum Insider, that "quantum error mitigation and quantum error correction are often framed as tools for different eras of quantum computing: mitigation for near-term devices and correction for scalable fault tolerance. " The new result doesn't erase that divide, but it shows the two techniques can be stacked productively before fault tolerance arrives.
What eWeek Found: The Catch Behind the 63x Claim
The 63x figure is the inferred sampling overhead at six Trotter steps. It comes from a preprint that has not yet been peer-reviewed. The result has not yet been independently replicated on another quantum system.
That distinction matters. Coverage of results like this one often collapses different metrics into one headline multiplier, the same way qubit counts get compared against fidelity or coherence numbers that measure something else entirely. This result counts reruns.
Peer review and replication on other hardware will show whether the reduction holds across different processors, noise profiles, and circuit designs. Until then, the 63-fold figure is best understood as a promising result from one experimental configuration — not a general reduction in the cost of quantum error correction.