A CHECKABLE RESULT
Interactive demonstrations are useful when their numbers can be recovered from the equations and their assumptions can be varied without quietly changing the question.
Record the page and model version, all inputs, units, preparation, outcome rule and calibration. Identify which inputs were measured, fitted or chosen for illustration.
Choose a small exactly solvable setting. Check normalization, signs, conservation rules and limiting cases. A plot that looks plausible is weaker evidence than a baseline with a known answer.
Vary a clock law, memory partition, detector overlap or response operator while keeping the other definitions fixed. Ask whether the claimed implication survives.
Save the inputs and outputs together. Include unsuccessful trials when acceptance or postselection is involved. State numerical tolerances and avoid presenting rounded agreement as an exact identity.
Use a quantity that did not set the parameters. Report the reference model, uncertainty and parameter range. Reproducing a fitted input is a calibration check; predictive content comes from constraints on further observations.
For a deterministic map F and a record map R, choose two states with R(x) = R(y). Compare the sequences R(Fkx) and R(Fky). One unequal future record is enough to disprove predictive closure of that partition.
If equality holds for every future step in a finite deterministic model, the states can be grouped for that observation protocol. A changed input alphabet or instrument can require a finer partition.
Run the finite memory example →Choose B = 4 and r = 1/2. Then D = log 4 / log 2 = 2. The illustrative choices u = n log 4 and τ = n log 2 give u/τ = 2 for n > 0.
With τ′ = τ + u/2, the same archive gives u/τ′ = 1. Geometry alone therefore does not rule out this clock. A physical selection argument must explain which clocks are admissible.
Explore clock selection →For a balanced two-path benchmark, P± = [1 ± V cos φ]/2, with 0 ≤ V ≤ 1. Thus P+ + P− = 1. At V = 1 and φ = 0, the outputs are 1 and 0; at V = 0, both are 1/2.
These are checks of the supplied benchmark. An SMM/DTT interpretation becomes distinguishable only through an additional rule that changes or constrains an independently accessible observable.
Explore the interference benchmark →| Comparison | What it can support | What is still needed |
|---|---|---|
| Code versus an exact baseline | The implementation reproduces that calculation within stated precision. | Additional cases and checks that target the actual failure modes. |
| Two equivalent mathematical descriptions | The transformation preserves the named observables. | A new observable or additional constraint to distinguish interpretations. |
| Model versus calibration data | The parameters can fit the chosen data under the supplied assumptions. | Out-of-sample observables, uncertainty and comparison with alternatives. |
| A theorem under stated premises | The conclusion follows in its quantified domain. | An argument that the physical system satisfies those premises. |
| An explicit countermodel | The original premises do not force the claimed conclusion. | A justified additional principle, a narrower claim, or a revised target. |
Page / model / date: Question being checked: Inputs and units: Preparation and measurement rule: Fixed calibration: Exact or independently calculated baseline: Varied assumption and allowed range: Outputs, tolerance, and retained failures: Reference model: Which observation was not used for fitting? Conclusion and limits:
The new labs provide downloadable run data where their controls support it. Keep that data with the model definition. A saved number without its preparation, input settings and observation rule is not a complete record.
When sending a correction or extension, link the exact page, supply the smallest example that demonstrates the issue, and distinguish a mathematical discrepancy from a numerical or display error. Continue with prediction and comparison, memory and repeatability, or contact the author.
SMONAD
Share this page, or keep a clear reference to the work you are reading.
Use this reference for the explanation on this page. “n.d.” means its first publication date is not specified; the update date is recorded separately.
