New method verifies accuracy of quantum error correction

Researchers found a way to precisely measure how well quantum codes and circuits suppress errors.

A new method certifies code and circuit distances at scale. These distances show how strongly quantum error correction suppresses logical errors.

The researchers used tropical partition functions to express these distances. They evaluated them with rank-based dynamic programming and weight-bounded search. Circuit reductions and relaxations shrink the problems.

Proved lower bounds with explicit logical errors of matching weight certify exact values. This method certifies more instances than all other tested exact solvers combined. It works for codes with up to 31,200 qubits.

The method also works for memory circuits with up to 8×1058\times10^5 fault mechanisms. An audit of 26 recent papers found overestimated distances in 10 papers. One discrepancy was at least eight.

Certified distances guide circuit design and show which defective chips remain usable. For IBM's two-gross memory circuits, the standard schedule has a certified distance of 16, not the assumed 18. Redesigned schedules reach 18 at one noisy cycle.

At 99.98% junction yield, keeping memory blocks with dead checks certified at distance 10 or more helps. This raises the estimated usable share from 53% to 94% with an improved schedule.

Key facts

  • The method certifies distances for codes with up to 31,200 qubits.
  • An audit of 26 papers identified overestimated distances in 10 of them.
  • Redesigned schedules for IBM's two-gross memory circuits reach a distance of 18 at one noisy cycle.
  • Using improved schedules with dead checks certified at distance 10 or more raises usable share to 94% at 99.98% junction yield.

All sections

Edition