What makes a derivation non-circular?

A derivation should reduce the freedoms left by its inputs

A non-circular physical derivation defines its inputs independently of the target and shows that the target follows across the whole admitted model class. It can use measured calibration data, but it must state which data and which outputs remain genuinely uncalibrated. Solving a model after choosing its parameters is useful; explaining why those parameters are selected is an additional task.

This standard applies to every research framework. For SMM/DTT it is particularly valuable because one source is meant to connect geometry, clocks, matter and constants. A shared name or controller does not establish those connections if each output is governed by an independently chosen rule.

The prediction-after-calibration guide expresses the criterion precisely: the intended output must be fixed, or restricted to a stated interval, among all admissible models agreeing with the independent inputs.

How a valid count can conceal a fitted choice

The recurrence r₁=1, r₂=q+3 and r₃=qr₂+7 gives r₃=q²+3q+7. At q=10 it yields 137. At q=9 it yields 115, and at q=11 it yields 161. The arithmetic is exact in every case.

If q is defined by solving q²+3q+7=137, its positive solution q=(√529−3)/2=10 simply recovers the inserted target. If an independently specified receiver task instead forces q=dim Sym²(ℝ⁴)=10, the count follows from that task. The next physical question is why the source selects that four-dimensional envelope, symmetric receiver and retention grammar.

Thus the mathematical count can be rigorous while its source selection remains an assumption. The structural-137 discussion separates these stages. Improving the derivation means proving one of the remaining selections, rather than replacing the word “assumption” with “necessity.”

An exact family exposes an unfixed prediction

Let a,b>0 and suppose an independent calibration establishes ab=1. The admissible family is a=t,b=1/t for every t>0. A proposed output a/b then equals t² and can be any positive number. The calibration has determined a product but leaves the ratio entirely free.

Adding a+b=2 forces a=b=1, because (a−b)²=(a+b)²−4ab=0. The ratio is now fixed. By contrast, calibrations ab=2 and a+b=3 leave the labeled pairs (1,2) and (2,1), giving ratios 1/2 and 2. An additional asymmetric measurement or source law is required to distinguish them.

This example shows why counting the number of equations is insufficient. Continuous freedoms, discrete ambiguities and symmetries must all be considered. A claimed prediction should be checked over the complete solution family, including branches that a numerical optimizer may not visit.

When is a clock equality only a definition?

Suppose a measured refinement count and scale give D, and a source activity is u. Defining a new variable T=u/D makes u/T=D automatically. This is a valid coordinate definition, but it is not a derivation of how a physical oscillator or material transition records duration.

To establish a clock law, define the clock reading through an independent device model, prepare it by a fixed procedure and compute its response while the source varies. If the same update then forces u/T=D across a nonempty family, a real freedom has been removed. The clock-universality problem asks whether that relation holds for a stated class of material clocks rather than one selected counter.

The six nested inner levels can organize the source trajectory, and the seventh completion can mark when a record is released. Their physical clock relation still needs a measurement dictionary supplied before the desired equality is evaluated.

A coordinate normalization cannot choose an invariant response

Consider an action containing Z(∂φ)²/2−gJφ, with Z>0. Under φ′=aφ, the coefficients become Z′=Z/a² and g′=g/a. The combination g²/Z is unchanged. One may set Z′=1 by choosing a=√Z, but this transformation does not determine the invariant interaction strength.

This is why a physical normalization problem cannot always be solved by declaring convenient units. A common action or clock scale may cancel from a dimensionless ratio; an independent dimensionless response coefficient can still remain. The coupling discussion gives explicit stiffness deformations that preserve counts while changing observable response.

A careful proof separates coordinate redundancy from physical variation. First remove changes that merely relabel the same model. Then ask whether any inequivalent models satisfying the premises still predict different measurements.

Check existence before proving necessity

A collection of assumptions can imply any target if it admits no model. For example, a=1, b=2 and a=b are inconsistent. A source theorem should therefore supply a nonempty example before claiming that all admissible sources have a desired property.

It should also state the permitted variation. A theorem about one isolated hand-selected matrix does not show that a relation survives a changed preparation, refinement setting or material species. A useful family may be narrow, provided its boundaries are explicit and motivated. Counterexamples outside that family clarify which premise protects the conclusion.

Finite calibration precision should be carried through the result. If an independent measurement gives a∈[2−ε,2+ε] with ε<2, then y=1/a lies in [1/(2+ε),1/(2−ε)]. At ε=0.1 this is [10/21,10/19]. A controlled interval is an honest prediction even when an exact value is unavailable.

Independence concerns the measurement, not its label

Two formulas may use differently named parameters while obtaining both from the same fitted observable. For example, defining a length L from a target energy E and then defining a time T=L/c does not provide two independent calibrations. A later relation involving L and T may simply reproduce their shared origin.

A useful audit follows each input back to an actual mathematical selection or measurement procedure. Ask what would happen if the target outcome changed while those procedures were held fixed. If an upstream parameter would be retuned, the calculation has used the target indirectly.

This does not prohibit reuse of well-established calibrated quantities. It makes their role explicit and identifies what remains to be predicted. A source programme can then aim for a new dimensionless relation or a family of outputs that no remaining retuning can reproduce arbitrarily.

What should a collaborator submit?

Choose one source-to-observable claim and provide a dependency table. List the state space, autonomous update, allowed preparations, retained memory, completion test and independently defined output. Mark which quantities are fixed by mathematics, supplied as physical hypotheses or determined by calibration. Then vary every remaining admissible parameter.

The strongest deliverable is a uniqueness or stability theorem together with an explicit working example and a counterexample after one necessary premise is removed. A negative result can be equally useful when it isolates the exact missing relation. Such a contribution gives the SMM/DTT programme a clear next step: a source law that removes a real degree of freedom and produces a result that another reader can reproduce without inserting the answer.

Continue reading

• Prediction after calibration →
• SMM/DTT research claim map: results, assumptions and open targets →
• Open problem: one dynamics for geometry, clocks and matter →
• Open problem: fixing a normalized physical coupling →

Read this discussion as part of a path

• Enter through the mathematics →
• Meaning, explanation and distinction →