VISUAL REFERENCE DESK

Keep the context in the picture.

Two original teaching diagrams, with readable explanations, primary sources and reference files. These are conceptual aids, not measurements or source figures.

Where did the circuit run?

Three execution environments; three descriptions to keep separate.

Three separate panels explain local simulation, remote simulation and a physical quantum processor. Access alone does not establish advantage; local does not guarantee zero cost. — **Access does not establish advantage. Local does not guarantee zero cost.** This is a classification of execution environments, not a progression from worse to better. AWS documents simulators and quantum devices as different ways of running tasks. A local simulator uses the local environment, which can itself be a hosted notebook.

Use the idea

When explaining an experiment, identify the task, environment and output first. A result from one environment cannot silently stand in for another, and service access alone supplies no performance comparison.

Local simulator Classical resources in the local environment. Outputs model the specified circuit.
Remote simulator Classical resources provided remotely. Confirm the service, method and costs.
Quantum processor A physical device executes the task. Record the device, settings, shots and result.

Sources and permission

AWS: What is Amazon Braket?Living documentation; publication date not stated
AWS: Testing a quantum task with the local simulatorLiving documentation; publication date not stated

Original QubitWire schematic. Reference download; no general republication license is granted. Share the context link. QubitWire terms apply. Read the terms.

Version 1 · Sources checked 24 September 2026

landscape SVG · 1200 × 880 landscape PNG · 1200 × 880 portrait SVG · 600 × 1350 portrait PNG · 600 × 1350 Context and citation TXT

Selectable text with sources

Where did the circuit run? QubitWire, version 1. Sources checked 24 September 2026. https://qubitwire.com/resources/visuals#environments

Access does not establish advantage. Local does not guarantee zero cost.

Original QubitWire schematic. Reference download; no general republication license is granted. Share the context link. QubitWire terms apply.
Terms: https://qubitwire.com/terms

Sources:
AWS: What is Amazon Braket?: https://docs.aws.amazon.com/braket/latest/developerguide/what-is-braket.html
AWS: Testing a quantum task with the local simulator: https://docs.aws.amazon.com/braket/latest/developerguide/braket-send-to-local-simulator.html

Read the explanation in context →

Make a comparison you can inspect.

A worksheet for reading a quantum performance claim.

Three questions ask whether the task, counted resources and evidence match. A quantum memory result is not an application benchmark; experimental conditions must stay with the claim. — **A memory result is not an application benchmark. Keep the conditions with the claim.** The surface-code paper reports below-threshold quantum memories and discusses the further work needed for logical computation. Its April 2026 correction changes labels and keys in Figure 3a. This original worksheet does not reproduce the paper’s figures.

Use the idea

The three questions are QubitWire’s reading aid, not a scoring system or a result reported by the paper. They help expose mismatched tasks and missing context before making a numerical comparison.

Same task? Specify the input, output and success condition. Memory protection and a useful application are different tasks.
Resources counted? Name the device or simulator, encoding, repetitions and classical work. Keep the stated conditions visible.
Evidence attached? Link the methods, comparison and corrections. Separate what was measured from what you infer.

Put two claims on the same worksheet

Write each claim with its source and version, then compare the task, accepted output, counted resources and conditions. A faster time does not establish a like-for-like improvement when the task or accepted quality differs. A reported fidelity may concern a different operation or a selected subset of events.

Fields to keep beside each claim
Field group	What to record
Source and version	Exact claim, claimant, source title, URL, version and source date.
Task and output	Input, operation, architecture, accepted output and quality target.
Resources	Physical count, logical scope, repetitions and classical costs; state the counting boundary.
Conditions	Noise model, decoder, acceptance rule and denominator.
Timing	Start and end of the measured interval. Keep throughput and feedback latency distinct.
Metric and comparison	Metric, uncertainty and the actual baseline. State what is comparable.
Evidence and limits	Supporting result, correction relationship and what is not established.

Keep missing information visible. “Not reported” means the checked source does not supply the field. “Not checked” means this comparison has not verified it. “Not applicable” means the field does not belong to the task. “Not comparable” means the values measure different things. None means zero.

End with a sentence explaining what can and cannot be compared, rather than a combined score. For logical resources, continue to the error-correction guide. For a recent claim’s evidence, open the Breakthrough Ledger.