Hypotheses: local response and experimental context
Compare nine recalibrated GHU benchmarks and inspect a separate CMS spectrum with its covariance. Each selection includes a reading of the calculated values and their scope.
Control names are kept as they appear in the application. Start with a reference, change one input at a time, and retain the assumptions with your result.
What this tool is for
Distinguish changes in model parameters, local couplings, full-field potential differences and experimental error structure. Theory and CMS are separate contexts until a compatible prediction exists.
First steps
- Open Same potential? and use the jump button for Hypotheses. Press Demo for the real-control walkthrough.
- Choose cases 1 and 6. Read the recalibrated coupling changes, then expand the potential interval.
- Choose CMS intervals i and j. Inspect the measured value, marginal error and signed correlation.
- Export the selected SVG, full JSON or observed CSV. Import the JSON to restore a validated selection.
- Open Certificates. Read each claim and its scope, then download its JSON or the existing catalogue ZIP.
Controls and inputs
| Control | Meaning | Try this |
|---|---|---|
| GHU case | One of nine fixed recalibrated cases; changing the case may change multiple inputs and field multiplicities. | Compare cases 1 and 6. |
| Potential interval | Field θ in 0.06–0.14 or 0–π. The vacuum θH remains fixed at 0.1. | Switch regions and read the vertical exponent. |
| CMS intervals i and j | Select one measured bin and a second bin for their experimental correlation. | Set i=j, then select different intervals. |
Reading the results
| Result | Meaning |
|---|---|
| Four linked figures | Parameter scatter and potential difference share the theory selection. Measured error bars and covariance-derived correlations share the CMS selection. |
| Downloads and imports | JSON preserves provenance, full covariance and theory samples. SVG is standalone; CSV contains observed bins. Import restores inputs, not an external dataset or imported verdict. |
| Scoped certificates | Ten rational Lean identities, exact covariance witnesses and three pointwise derivative enclosures. Full-integral error and observed-fit certification remain pending. |
Read axes, units and source before comparing panels. Small coupling changes after recalibration do not establish experimental indistinguishability. The potential scale is fixed across cases within a region, and changes when the region changes.
Worked example
Follow the interactive demo
- Open Same potential? and use the jump button for Hypotheses. Press Demo for the real-control walkthrough.
- Choose cases 1 and 6. Read the recalibrated coupling changes, then expand the potential interval.
- Choose CMS intervals i and j. Inspect the measured value, marginal error and signed correlation.
- Export the selected SVG, full JSON or observed CSV. Import the JSON to restore a validated selection.
- Open Certificates. Read each claim and its scope, then download its JSON or the existing catalogue ZIP.
What to expect: The demo ends with separate theoretical and experimental readings. It never assigns an exclusion to the nine benchmarks.
Scope and limitations
- Dominant-tower reconstruction, bottom omitted, physical uncertainty unquantified. The browser reads archived calculations; it does not rerun the Bessel integrals.
- This study is independent of the exact SU(7) equivalence theorem in its host section. Its video chapter and interactive Demo share the same practical walkthrough.
- No observed GHU fit or full physical certification is provided. A covariance alone does not supply a compatible theory prediction.
- The about 7.4% quartic fit/table discrepancy is unresolved. Quadrature convergence is not a rigorous total error bound.
If something is unclear
The selected curve is zero.
Case 1 is compared with itself. Select another case.
Import is rejected.
Use this card’s JSON export with valid integer selections. Unrelated datasets require a separate audited adapter.
Sources and attribution
Current instrument guide. Archived editions retain their own scope.
Back to top