L7-07 · Classical shadows and observable structure

Learn¶

A classical shadow stores randomized measurement data in a form that estimates many observables. Each measurement setting and outcome defines an inverted snapshot. Averaging snapshots gives an estimator for selected properties. The snapshot need not itself be a physical density operator.

Efficiency depends on the measurement ensemble and the target observables. For local random Pauli measurements, a high-weight observable can have a large shadow norm. A logarithmic dependence on the number of requested observables does not mean every observable can be estimated with a system-size-independent constant cost.

Use a measurement ensemble suited to the property family. State the error and confidence target and account for reconstruction time. Classical shadows do not provide a free complete state description at arbitrary accuracy. They also do not remove causal identification assumptions when the target is an intervention property.

Practice¶

Check your understanding¶

What controls the measurement cost beyond the observable count?

• A. The measurement ensemble and observable-dependent shadow norm

• B. Only the number of stored files

• C. Only the Hilbert-space label

Sources¶

This lesson draws on Huang et al., 2020Watrous, 2018.

Next: L7-08 · Reproducible quantum causal benchmarks