A180Draft-based articleExpanded article4 min read

Freeze parameters before predicting spectroscopy

A spectroscopy test becomes predictive when its operator family, coefficients and calibration data are fixed before the reserved transitions are examined. The freeze record makes the dependency between constants and atomic predictions auditable.

In this article

Why freezing parameters changes the strength of a claim
Define the calibration set before the test set
A simple algebraic example
Freeze the operator as well as the numbers
Connection to the constants program
A concrete freeze record

Key ideasSources & reading

01Why freezing parameters changes the strength of a claim

A model can often reproduce a chosen spectral line by adjusting a coefficient. Its explanatory value increases when the same coefficient is fixed independently and successfully predicts other lines. The post-freeze strategy in Paper 13 makes this distinction explicit. For SMM/DTT, upstream quantities might include a source-selected coupling, an action normalization and particle masses. If these are still empirical inputs, that is acceptable for a conditional atomic test, but they must be labeled. The crucial rule is that the target transition must not silently choose the parameters used to predict itself.

02Define the calibration set before the test set

List every datum used to set a parameter, including indirect inputs. A Rydberg constant or nuclear radius from an adjustment that included the target lines can carry those lines into the prediction even when they are not explicitly fitted again. The clean approach is to use independently determined inputs or an adjustment that excludes the reserved tests. Record the parameter covariance, units and theory conventions. A parameter freeze is a versioned numerical and conceptual specification, not simply a promise not to change one decimal after seeing the answer.

A single nonrelativistic n equals two level branches into fine-structure levels. A further radiative step splits the lower pair into 2S one-half and 2P one-half levels. — **Figure 1. Fine structure and Lamb shift: distinct benchmarks.** Schematic only, with deliberately exaggerated separations. For hydrogen n = 2, Dirac fine structure separates j = 1/2 from j = 3/2; radiative corrections further separate 2S₁/₂ and 2P₁/₂. No measured scale is plotted. View full size ↗

03A simple algebraic example

Suppose a toy spectrum has residuals ri=A si, where the shape si is fixed by a derived operator. If A is obtained from one designated calibration line, every other residual is predicted relative to that line. Two additional independent lines can then test whether the shared shape works. If instead each line has its own Ai, any collection can be matched and the model has lost cross-line predictive power. The distinction is not whether fitting is ever allowed; it is which data determine which parameters and how many independent consequences remain.

Calibration: A = r₁/s₁. Reserved prediction: r₂ = r₁(s₂/s₁).

04Freeze the operator as well as the numbers

A model can also overfit by changing its functional form after every discrepancy. For a contact correction, the nS pattern is 1/n³. For a spin-sensitive operator, matrix elements depend on angular labels. Choosing among many shapes after inspecting all data consumes information even if the final model has only one amplitude. State the operator family and model-selection rule beforehand, or use a separate exploratory set and a fresh test set. This keeps the physical claim tied to a source mechanism rather than to a retrospective arrangement of residuals.

A curve in a two-parameter plane represents models sharing the same calibrated observable. A second contour crosses it at a point and removes that illustrative ambiguity. — **Figure 2. Calibration can leave hidden parameter directions.** A fixed measurement may select a curve of models rather than one point. A genuine prediction must be constant on that calibration fiber, or require additional independently measured diagnostics. View full size ↗

05Connection to the constants program

The broader SMM/DTT program seeks noncircular derivations of α, action scale and masses. Spectroscopy can test that program only if the dependency direction is respected. Deriving α from a structural receiver and then checking hydrogen differs from choosing a receiver normalization using hydrogen and reporting the resulting agreement. Both can be useful stages, but only the first is an independent prediction of the coupling. The same care applies to outer-time calibration: a frequency standard imported from the target transition cannot simultaneously be evidence that the model predicted its absolute frequency.

06A concrete freeze record

A publishable analysis should save a compact manifest with the source revision, equations, fitted or derived parameters, calibration observations, reserved transitions and covariance convention. It should specify the approximation order and how theoretical remainders enter the error budget. If a failure leads to a revised model, the old prediction remains a documented test result and the revision gets a new validation set. This practice makes successive SMM/DTT papers accumulate evidence rather than erase unsuccessful attempts. The dependency is explicit: independent source derivation and calibration determine one frozen Hamiltonian, whose shared levels produce the reserved spectral predictions.

KEEP THE CONNECTION

Key ideas

A target line cannot independently validate parameters chosen from that line.
Freeze the operator family and provenance as well as parameter values.
Shared predictions across reserved transitions test the source program.

Scope of the argument. A proposed analysis protocol; no calibrated data set or successful DTT spectroscopy comparison is asserted.

Sources & further reading

Temporal Reconstruction of Hydrogen III, skeleton research outlinepaper13_hydrogen_fine_structure_lamb_shift.pdf, §§2–7, pp.3–9
Minimal hydrogen note: transition consistency and contact correctionsmm_dtt_minimal_hydrogen.pdf, §8.1, p.7
NIST CODATA theoretical treatment of hydrogen and deuterium ↗

Source editions and catalogue entries

Skeleton draft

Temporal Reconstruction of Hydrogen III: Fine Structure, Lamb Shift, and Radiative Inner-Time Loops

Working manuscript. Article-level references identify the consulted sections and the scope of the interpretation. No public URL was supplied for this exact manuscript version.

Source register →