Two charts can look almost identical and come from very different kinds of experiments. The execution label is part of the result.
Key takeaways
- Theoretical probabilities are not measured counts.
- Finite sampling varies even in ideal simulation.
- Hardware access and compatibility need their own checks.
Chapter 01
An ideal model answers a clean question
An ideal state-based calculation describes the program without physical-device noise. It is useful for understanding intended gate behavior and checking a small circuit. When a chart shows theoretical probabilities, label it as such rather than calling the values measured counts.
Chapter 02
Sampling adds ordinary variation
A sampled simulation repeats a modeled measurement a finite number of times. Its counts fluctuate even when the underlying probabilities stay fixed. For an ideal Bell-state preparation, a run need not split exactly evenly between 00 and 11. Record the number of shots before comparing histograms.
Read the execution label
Similar charts, different questions
| Result | What it describes | Keep beside it |
|---|---|---|
| Ideal probability | The model’s expected distribution | Circuit and measurement basis |
| Sampled simulation | Finite repetitions of a model | Shots and any noise model |
| Hardware measurement | A specific physical-device run | Target, compilation and run settings |
Chapter 03
A noise model is still a model
A simulated noise process can help explore possible errors, but it is not automatically a measurement of the current device. Record the model and its assumptions. Differences between that model and a physical system are part of the comparison, rather than a detail to leave out.
Chapter 04
Hardware adds access and device conditions
A QPU run depends on a reachable target, valid credentials, supported operations and the provider’s current conditions. Compilation and measurement mapping can affect what you see. Check those details before attributing every unexpected outcome to noise.
In QFlow, visible devices depend on personal access, organization policy and administrator configuration. A hardware option in an article does not create entitlement to use it.
Chapter 05
Choose the mode that answers your question
If you are learning what H and CNOT do, start with an available simulator. If you are studying device behavior, preserve the ideal reference and make the hardware comparison explicit. Both can be useful; they answer different questions.
Save the circuit, execution mode, target and shot count with the result. Your future self should not have to guess which kind of experiment produced the chart.
