Choose a route and keep an experiment record¶
Once you can run a program and read its result, choose the subject closest to your work. You do not have to finish every branch before starting a useful experiment.
| Background or question | Suggested route | A result to aim for |
|---|---|---|
| New to quantum programming | Foundations, then Bell circuit | Explain counts, probability and bit order |
| Familiar with gate-based circuits | Digital, then Hybrid | Build a parameterized circuit and a continuous D-A-D experiment |
| Interested in neutral-atom controls | Experimentalist route | Compare waveforms or local controls in a small array |
| Studying optimization | Optimization route | Compare candidate solutions with an explicit classical baseline |
| Integrating or extending the SDK | Platform or compiler | Explain a job lifecycle or compiled parameter mapping |
Record enough to repeat the run¶
Save the script and the returned result together. Alongside them, record the following information in a short text file:
Question:
SDK version and source revision:
Python version and dependency environment:
Program, units and parameter values:
Backend, simulation method and precision settings:
Shots, random seeds, repeated runs and trajectories when applicable:
Expected observation and comparison baseline:
Output files:
What the result supports, and what it does not establish:
git rev-parse HEAD records the source revision. python -m pip freeze lists installed dependencies. A fixed seed is useful, but does not replace either record. An exact random sample is not guaranteed across dependency or SDK changes.
Choose one independent variable before sweeping several at once. Keep all other settings fixed, and decide how you will compare results before selecting the most favorable run. The experimental-design lesson works through a small example.
Practice: another student has your HTML report but cannot reproduce it. Name three missing items that a screenshot alone is unlikely to preserve.
Possible answers
The executable source and its revision; all bound parameter values and units; the simulator method and numerical settings; the seed and sampling budget; or the dependency environment. A chart title is not a complete experiment record.