Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errorsQuantum state tomography estimates the state itself; classical shadows create a compact record from randomized measurements that can be used to estimate selected properties. Shadows can reuse measurement data for many predictions, but they do not generally reconstruct the full state or make every property cheap to estimate. Choose based on whether you need a complete state description or answers to particular questions about it.
What each method gives you
| Question | Quantum state tomography | Classical shadows |
|---|---|---|
| Primary output | An estimate of the quantum state, commonly represented by a density matrix. | A compact classical record, or “shadow,” used to estimate selected properties. |
| Measurement strategy | Measurements chosen to be tomographically complete, so their outcomes can determine the state elements. | Randomized measurement settings and outcomes are combined into classical snapshots and processed with an estimator. |
| Best fit | Questions that require the state description itself. | Questions about a useful set of properties, especially when the target properties can be selected after collecting the measurements. |
| What it does not guarantee | A state estimate does not by itself make every later calculation inexpensive. | It does not provide a free compressed encoding from which every property can be accurately recovered. |
The comparison is about the goal and the measurement protocol, not simply “old” versus “new.” Both methods infer information from measurements on quantum systems, but they organize that information for different outputs.
How quantum state tomography works
In conventional state tomography, an experimenter measures copies of a system using a set of measurements that is tomographically complete for the chosen state representation. The measurement outcomes are used to estimate the density matrix or another chosen parameterization. Tomographic completeness matters: without it, different states may be consistent with the collected measurement data, so the state elements cannot be determined unambiguously.
This is the natural approach when the research objective is to characterize the state broadly, rather than answer only a preselected list of questions. The estimate can then be used to calculate properties of interest, subject to the quality and limits of that estimate.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →#1 Best Overall
How classical shadows work
Classical shadows use randomized operations or measurement settings on state copies. A setting and its outcome are recorded together as a classical snapshot. A reconstruction map or estimator processes those snapshots to predict properties of interest; the method need not first produce a full density matrix.
- Choose a suitable measurement ensemble. The random measurement settings must support the properties you want to estimate.
- Measure copies and retain the settings and outcomes. Each resulting snapshot contributes to the classical record.
- Apply estimators to the recorded snapshots. The same record can be used to estimate multiple supported properties.
Examples discussed in Hsin-Yuan Huang, Richard Kueng, and John Preskill’s 2020 paper and Huang’s 2022 review include local observables, fidelities, entanglement entropy, and an expected Hamiltonian value. These are examples of possible targets, not a promise that one measurement design estimates every such quantity equally well.
Rank #2
Why “shadow tomography” can mean different things
“Shadow tomography” is also used for a broader task: estimating many measurement outcome probabilities. Protocols for that task may use collective measurements. The classical-shadows method introduced by Huang, Kueng, and Preskill is a particular property-prediction approach based on randomized measurements.
The distinction is experimentally relevant. The 2021 study Experimental Estimation of Quantum State Properties from Classical Shadows contrasts the difficulty of directly realizing the original collective-measurement approach with a classical-shadows procedure using separable measurements on individual copies. When comparing papers, check which measurement model they mean rather than assuming that the names describe the same protocol.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
What the sample-efficiency result does—and does not—say
Huang, Kueng, and Preskill’s 2020 paper states that, for its protocol and stated guarantee, order log(M) measurements suffice to predict M functions with high success probability; the result is described as independent of system size under those stated conditions. This is a specific theoretical result, not a universal measurement count for arbitrary observables, hardware, noise, or accuracy.
For a particular task, the required sample count depends on factors including the shadow norm or other protocol-specific quantities, the target observables, the chosen measurement ensemble, the desired accuracy and confidence, and noise. A later study, Lower Bounds for Learning Quantum States with Single-Copy Measurements (2025), further emphasizes that sample complexity depends on the available measurement choices. The number of samples also does not by itself capture all experimental and classical-computing costs.
What has been demonstrated experimentally
The 2021 PRX Quantum study reports classical-shadow estimates of operator mean values and fidelity using quantum-optical, high-dimensional spatial states of photons. It reports access to Hilbert spaces of dimension up to 32 in that experiment and compares fidelity estimation with conventional reconstruction under limited measurements. That dimension describes the reported experiment; it is not a general capacity limit or guarantee for classical shadows.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Which approach should you use?
- Use state tomography when the deliverable is a broad state estimate, such as a density matrix, and your measurement design can be tomographically complete.
- Consider classical shadows when the scientific question is about a set of properties and an available randomized measurement protocol supports estimating them. A benefit is that targets may be chosen after measurements are complete, as described in the CaltechAUTHORS record for the 2020 work.
- Compare protocols before deciding if measurements are constrained, the targets are unusual or numerous, or noise and precision requirements are important. The measurement ensemble and target property determine whether the apparent measurement savings apply.
Classical shadows can avoid full reconstruction for a suitable prediction task, but they are not a universal replacement for tomography. The 2022 review notes fundamental limits on accurately predicting some classes of properties by classical post-processing. Extensions such as classical shadows for quantum process tomography concern quantum channels; they should not be conflated with state tomography.
Quick Recap
Best Value
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
