What the study found
The study finds that uncorrected local noise can effectively shorten quantum circuits to logarithmic depth when estimating observable expectation values. It also finds that quantum circuits under any non-unital noise do not exhibit barren plateaus for cost functions made from local observables.
Why the authors say this matters
The authors conclude that, unless quantum circuits are carefully designed to take advantage of noise, noisy quantum computers are unlikely to beat shallow ones for algorithms that estimate observable expectation values. They also note that this is relevant to many variational quantum machine learning proposals.
What the researchers tested
The researchers studied the effect of uncorrected local noise on logical quantum circuits, motivated by near-term hardware without successful fault-tolerant quantum error correction. They analyzed observable expectation values, non-unital noise, and cost functions composed of local observables, and they also designed a classical algorithm for estimating expectation values.
What worked and what didn't
For observable expectation values, the authors show that noise effectively truncates most quantum circuits to logarithmic depth. They prove that non-unital noise removes barren plateaus for local-observable cost functions, and they also obtain an efficient classical algorithm that estimates expectation values within any constant additive accuracy with high probability over the choice of circuit, for any circuit architecture.
What to keep in mind
The abstract does not describe experimental data or a specific hardware implementation. The results apply to the setting studied here: noisy quantum circuits, observable expectation values, local observables, and the stated noise conditions.
- Uncorrected local noise can effectively reduce many quantum circuits to logarithmic depth for expectation-value estimation.
- Quantum circuits under any non-unital noise do not exhibit barren plateaus for cost functions built from local observables.
- The authors designed a classical algorithm that estimates expectation values within any constant additive accuracy with high probability.
- The study suggests noisy quantum circuits are unlikely to outperform shallow ones for many expectation-value-based algorithms.