Here “glue results” means concrete consistency relations within stated models. Established theories and the examples below support this method. Interpreting them as one universal EEML principle remains a research proposal.
Take ẋ=ax+by and ẏ=cx−dy, with constant coefficients and d>0. Solving for y and substituting gives exactly:
ẋ(t) = ax(t) + bc ∫₀ᵗ exp[−d(t−s)]x(s) ds + b exp(−dt)y(0)
The integral is memory. An unknown y(0) can be described statistically; the final term is then correlated across times. With variance σ² its covariance is b²σ² exp[−d(t+s)]. This example is not stationary and claims no general equilibrium noise law. It shows why deleting y is not the same as deleting its effects.
General context: Gouasmi, Parish & Duraisamy, Mori-Zwanzig memory effects.
On a finite space with p_h∝exp[−(S−hB)/ε], ε>0 and h-independent A and B, differentiation gives:
∂ₕ ⟨A⟩_h = Cov_h(A,B)/ε
This relates static response to fluctuations in that ensemble. A dynamical fluctuation-dissipation relation needs further equilibrium and dynamical assumptions. The formula does not automatically apply to arbitrary nonequilibrium systems.
K3’s existence as a consistent geometry is distinct from its relative weight or unique selection by EEML. K3 need not be the most efficient or only realisation. χ=24 is a property of K3, not a universal EEML score. N_gen=χ/8 requires independent physical justification; 24/8=3 alone does not count generations.
Equip the real four-torus with a complex two-dimensional torus structure. The involution x↦−x has 16 fixed points. The quotient has A₁ singularities there. Their minimal resolution produces a smooth Kummer K3 surface. The second arrow denotes resolution, not ordinary equality or time evolution.
χ(T⁴) = 0 χ(T⁴/ℤ₂) = (0 + 16)/2 = 8 χ(K3) = 8 + 16·(2−1) = 24
Each resolved singularity replaces a point with an exceptional ℂP¹. Not every complex K3 surface is itself a Kummer surface. Huybrechts, Lectures on K3 Surfaces, §1.1 and §1.3.
A well-known string duality relates type IIA on K3 to heterotic string theory on T⁴, as six-dimensional theories with corresponding moduli and charges. It does not equate K3 and T⁴ as geometric spaces, nor is it the Kummer resolution. It illustrates different descriptions of corresponding physics within their theoretical framework, not experimental confirmation of EEML.
χ=24, signature −16 and scalar heat coefficient 4/15 are exact under their geometric assumptions. They provide concrete starting points for EEML research. Selecting K3 from the meta-law and connecting it to electromagnetism remain separate questions.
Take a smooth degree-four surface X in complex projective space ℂP³, for example the Fermat quartic:
z₀⁴ + z₁⁴ + z₂⁴ + z₃⁴ = 0
It is smooth: all partial derivatives vanish simultaneously only when every zᵢ is zero, which is not a projective point. Write H for the hyperplane class on X. The tangent-normal exact sequence gives:
c(TX) = (1+H)⁴ / (1+4H)
= 1 + 6H² (through complex degree 2)
∫X H² = 4
χ(X) = ∫X c₂(TX) = 6 · 4 = 24Adjunction gives a trivial canonical bundle; the hyperplane theorem gives simple connectedness. Thus X is a K3 surface. This calculation explains 24 within this geometry. It does not yet select the geometry from all possibilities.
With the complex orientation, the signature formula σ=(c₁²−2c₂)/3 gives:
σ = −16 b₂ = χ−2 = 22 b₂⁺ = 3, b₂⁻ = 19 h¹,¹ = b₂−h²,⁰−h⁰,² = 22−1−1 = 20
Here b₁=b₃=0 and b₀=b₄=1. These numbers count topological and complex structures. Without a physical compactification model they do not count particles or observable spectral lines.
The local dimension of Ricci-flat K3 metric moduli at fixed volume is 3·19=57: positive three-planes in H²(X,ℝ) with signature (3,19). Including the volume scale gives 58. Additional fields or constraints change the physical parameter accounting.
Geometric background: Daniel Huybrechts, Lectures on K3 Surfaces.
Now take a smooth closed K3 with a Ricci-flat Riemannian metric. For the nonnegative scalar Laplacian Δ, using the full real tensor norm:
∫X |Riem|² dV = 32π²χ = 768π²
The integrated scalar heat coefficient is therefore:
a₄ = [1 / (360 · 16π²)] ∫X 2|Riem|² dV = χ/90 = 4/15
At volume 1, as t↓0:
Tr exp(−tΔ) = 1/(16π²t²) + 4/15 + O(t)
The general coefficient and tensor conventions appear in Vassilevich, Heat kernel expansion, equation 4.28. Other operators, such as a coupled Dirac², have different terms. Ricci terms cannot simply be discarded for an approximate metric that is not exactly Ricci-flat.
Define ζΔ(s)=Σλ>0λ−s for Re(s)>2, then analytically continue. A connected closed K3 has one scalar constant zero mode, so:
ζΔ(0) = a₄ − dim ker Δ = 4/15 − 1 = −11/15
This is not ζ′Δ(0). The regularised determinant is det′Δ=exp(−ζ′Δ(0)) and requires further spectral information. A topologically fixed ζ(0) does not fix its derivative.
The scale must also be explicit. For a positive factor c:
ζ(cΔ)(s) = c^(−s) ζΔ(s) ζ′(cΔ)(0) = ζ′Δ(0) − (ln c)ζΔ(0)
Under g→r²g, λ→r⁻²λ and V→r⁴V. Thus λ√V is scale invariant in four real dimensions. This justifies the old scripts' volume convention, but does not validate their matrix assembly.
The old k=3,…,6 files have nonzero first eigenvalues. An independent check locates a problem in the energy matrix of spectrum_matrices. In the basis φₐᵦ=sₐs̄ᵦ/D, the sum ΣGₐᵦφₐᵦ is exactly the constant function 1. Its Dirichlet energy must be zero, even for finite samples.
| Check on the same 1,600 points | Result |
|---|---|
| Largest error in the constant value | less than 10⁻¹² |
| Largest derivative of that constant | 7.53 × 10⁻¹⁵ |
| Rayleigh quotient, original energy matrix | 1.656743 × 10⁻³ |
| Rayleigh quotient, independent real assembly | −5.24 × 10⁻¹⁸ |
This is a matrix consistency error, not physical energy of a constant function. Independent derivatives were also checked with finite differences. The matrices differ by about 19.5% in relative Frobenius norm. This percentage is not an error estimate for individual eigenvalues.
The old precise eigenvalues, estimated ζ′(0) and predictions depending on them therefore do not return as verified results. The check uses a small diagnostic sample; it does not provide a converged replacement K3 spectrum.
For a minimal elliptic K3 Weierstrass model over ℂP¹, y²=x³+f x+g, f and g have degrees 8 and 12. With f=a h², g=b h³ and h of degree four:
j = 1728 · 4f³/(4f³+27g²) = 6912a³/(4a³+27b²)
j is constant over the base but still depends on a³/b². Taking h with four distinct zeros and a nonzero discriminant factor gives the generic four-I₀* family. This does not select a unique constant value of j.
The elementary count of fibre multiplicity patterns is reproducible. For g=0, count partitions of 8 with parts 1,2,3: ten patterns. For f=0, count partitions of 12 with parts 1,…,5: 47 patterns. Remove the four-I₀* pattern from each and count that type once:
1 + (10−1) + (47−1) = 56
This counts types of multiplicity patterns in this description, not 56 individual K3 surfaces or metrics. Zero locations may still vary. Some old orbifold labels were also incorrect: the standard ℤ₄ example has 2 III* + I₀*, and the ℤ₆ example II* + IV* + I₀*.
For these families and orbifold examples: Justin Sawon, Isotrivial elliptic K3 surfaces and Lagrangian fibrations. An elliptic modulus or axiodilaton is not automatically the electromagnetic α; a physical dictionary is needed.
Stationary action means δA=0, not necessarily a minimum. Critical points of the Einstein-Hilbert functional on closed manifolds at fixed volume are Einstein metrics Ric=λg, not necessarily Ricci-flat metrics. A round sphere illustrates the distinction. See Introduction to Einstein manifolds.
Nor is the real weight e−S/ε the oscillatory quantum amplitude eiA/ℏ without an additional construction. Geometry, fields, action, couplings and observations must form one explicit physical model.
Download the verification bundle · Reproduction and scope
The checked status of the route to α →Sources for anomaly constraints: Bilal, Lectures on Anomalies; Golkar & Sethi, Global Anomalies and Effective Field Theory.
v0.25 · 27 September 2026