Essay · Plain · 7 min

The q⁺ / q⁻ bifurcation: why two branches?

At μ* = 15, the sieve splits into two natural branches: q⁺ for couplings (vertex, leptons, α_EM), q⁻ for geometry (propagator, quarks, metric). Why this bifurcation is unavoidable and what it separates.

Go deeper: L0 , T5 , T6

One line, two readings

At first glance, PT contains a single cascade: $s = 1/2$ , $\mu^* = 15$ , $\alpha_{\rm EM}$ . Yet, midway through, something surprising happens. The system splits into two parallel branches that share the same ingredients but use them in two different ways.

One measures couplings: how particles interact (vertex, $\alpha_{\rm EM}$ , lepton masses, PMNS oscillations).

The other measures geometry: the space those particles live in (propagator, Bianchi I metric, quark masses, CKM mixing, gravity).

These two branches are called $q^+$ and $q^-$ . They are tied to $\mu^* = 15$ in two profoundly different ways.

Theorem L0: maximum entropy

Everything starts with a simple theorem. Asking that a distribution on $\{2, 4, 6, \ldots\}$ be (i) memoryless, (ii) of fixed mean $\mu$ , and (iii) of maximum entropy gives a unique geometric law:

P(X = 2k) = (1 - q) \cdot q^{k - 1}.

This is theorem L0. With required mean $\mu$ , we find $q = 1 - 2/\mu$ .

But there is another derivation

Beside this exact discrete derivation, there is a natural continuous reading: the Boltzmann limit, where the gaps are seen not as even integers but as durations on a continuum. There, the distribution becomes $\propto e^{-x/\mu}$ , and the parameter is:

q^- = e^{-1/\mu}.

At the fixed point $\mu^* = 15$ :

q^+ = \frac{13}{15} = 0.8667 \qquad \text{and} \qquad q^- = e^{-1/15} = 0.9355.

Different. Yet both legitimate: nothing internal to the sieve favours one over the other.

How the bifurcation takes root

Here PT becomes subtle. At each prime $p$ , the holonomy formula T6 gives:

\sin^2\theta_p = \delta_p (2 - \delta_p), \qquad \delta_p = \frac{1 - q^p}{p}.

Evaluated with $q^+$ , we get the values $\sin^2(\theta_p, q^+)$ . Evaluated with $q^-$ , we get $\sin^2(\theta_p, q^-)$ . Two distinct families:

\begin{array}{c|cc} p & \sin^2(\theta_p, q^+) & \sin^2(\theta_p, q^-) \\ \hline 3 & 0.2192 & 0.1172 \\ 5 & 0.1940 & 0.1102 \\ 7 & 0.1726 & 0.1037 \end{array}

PT then reads a bifurcation: $q^+$ naturally lives on vertices (where particles interact pointwise), $q^-$ naturally lives on edges (the propagators between two vertices).

What the two branches separate

Pitfall. It is tempting to say “q⁺ is the discrete branch, q⁻ the continuous one”. That would be wrong. The discrete/continuous distinction only characterises the initial derivation route for q (L0 max-entropy for q⁺, Boltzmann limit for q⁻). Once fixed, both branches operate simultaneously on discrete and continuous objects: each computes continuous angles θ_p on discrete ℤ/pℤ circles, and each projects into a continuous metric on three discrete active directions. What truly separates the branches is their physical role.

Branch	$q^+$	$q^-$
Physical role	Vertex (point interaction)	Edge (propagation between points)
Value at μ = 15*	$13/15$	$e^{-1/15}$
Derivation route	L0 max-entropy on even integers	Boltzmann limit
Target	Couplings, leptons, PMNS, Higgs	Geometry, quarks, CKM, $G$ , cosmology
Discrete objects used	ℤ/3ℤ, ℤ/5ℤ, ℤ/7ℤ circles (CRT)	Three active directions 7 of Bianchi I
Continuous objects used	Holonomy angles $\theta_p$	Fisher metric, scale factors $a_p(\mu)$
Emblematic example	$\alpha_{\rm EM} = \prod \sin^2(\theta_p, q^+)$	$V_{us} = \sin^2(\theta_3, q^-) + \sin^2(\theta_5, q^-)$

This vertex/edge duality splits the 43 observables into two groups: those descending from the coupling branch, those from the geometry branch. No physical observable mixes the two at tree level — this is the cross-branch exclusion rule.

A signature: the factor 2

The difference between $\delta(q^+)$ and $\delta(q^-)$ is almost exactly a factor 2. This near-binary duality is what separates what is measured as “interaction” from what is measured as “space”.

Ablation test: permuting the branch assignment across the 43 observables degrades the discrepancies by a factor 106 on average. The bifurcation is therefore not cosmetic: it is forced by data.

Natural question: can we cross the routes?

If q⁺ comes from a discrete route and q⁻ from a continuous one, couldn’t we derive q⁺ along another continuous path, or q⁻ along another discrete one?

No, and that is what makes the bifurcation rigid. Here is why.

The continuum limit of q⁺ gives q⁻

Start from the discrete geometric distribution on even integers with mean $\mu$ . Take the continuum limit (event spacing $\Delta x \to 0$ with fixed mean): the geometric distribution becomes a continuous exponential:

P_{\rm disc}(X = 2k) = (1 - q^+)\,(q^+)^{k-1} \quad \xrightarrow{\Delta x \to 0} \quad P_{\rm cont}(x) = \frac{1}{\mu}\,e^{-x/\mu}.

The parameter of the continuous regime is:

q^- = \lim_{\Delta x \to 0} \left(1 - \frac{\Delta x}{\mu}\right)^{1/\Delta x} = e^{-1/\mu}.

So any attempt to re-derive q⁺ along a continuous route lands mechanically on q⁻.

Discretising q⁻ gives q⁺

Conversely, start from the continuous exponential. Discretise it on the even-integer lattice under the same mean constraint, and you recover exactly the geometric of parameter $q^+ = 1 - 2/\mu$ . No other result is compatible.

Conclusion: not an arbitrary choice

q⁺ and q⁻ are not two conventions among infinitely many possible. They are the two regimes (discrete-even / continuous-real) of the same variational problem (max-entropy under mean constraint). The route determines the regime, and the regime determines the parameter — no choice involved.

But are q⁺ and q⁻ then “the same thing seen differently”?

In the limit $\mu \to \infty$ , yes: $q^+ \to 1$ and $q^- \to 1$ , and the latent heat $L = q^- - q^+ \to 0$ . Both branches coincide.

At finite μ — and in particular at $\mu^* = 15$ — the gap $L \approx 0.069$ is non-zero, measurable, and physical. This finite gap between discrete and continuous separates the roles: point-interaction on one side, between-point propagation on the other. Without finite μ, no observable bifurcation.

The bifurcation is the arithmetic shadow of a thermodynamic fact: at finite μ, a discrete system and its continuum limit do not coincide exactly, and that gap carries half of Standard Model physics.

A first-order phase transition

The difference $L = q^- - q^+ = 0.069$ is a latent heat. PT reads the bifurcation as a first-order phase transition at $\mu^* = 15$ . Before bifurcation, the system is unitary; after, it is dual.

This latent heat is not lost: it reappears in the cosmological dark sector. The fraction $\Omega_{\rm info} = 26.48\%$ obtained by Clausius $L \sim D_{KL} = s^2 = 1/4$ matches the Planck measurement to 0.09 %. Not a fit.

Why it is unavoidable

The uniqueness $\mu^* = 15$ does not only say the fixed point is arithmetically determined. It also says that two independent derivations of $q$ from $\mu$ exist — an exact discrete one ( $q^+$ ), a continuous Boltzmann one ( $q^-$ ) — and neither can be removed.

Keeping only one would lose half the Standard Model. Artificially merging them would erase the natural separation between interactions and geometry. PT takes the cleanest path: both coexist and signal two faces of the same fixed point.

In one sentence

The $q^+ / q^-$ bifurcation is the unique way for the cascade to close its consistency at $\mu^* = 15$ while staying simultaneously (i) exactly discrete (L0 maximum entropy) and (ii) Boltzmann-continuous (natural Gibbs limit). Its necessity is the deep reason physics contains both couplings and geometry.