Easy Money: Yang-Mills on the Poincaré Homology Sphere

The Millennium Prize asks whether pure Yang-Mills theory on $\mathbb{R}^4$ has a positive mass gap. On flat space, confinement must emerge dynamically. On the Poincaré Homology Sphere $M = S^3/2I$, the answer is forced by geometry.

Positive Ricci curvature provides a universal floor. The finite fundamental group yields exactly three isolated vacua: trivial, standard, and Galois conjugate. The McKay correspondence for the extended $E_8$ diagram filters the spectrum at each vacuum, producing a ninefold enhancement at the Galois sector.

The same curvature that enters the Λ conversion guarantees confinement.

I. The Geometry

The Weitzenböck identity on $S^3$ decomposes the gauge-covariant Hodge Laplacian on bundle-valued 1-forms. In the Hamiltonian formulation of Yang-Mills on $M \times \mathbb{R}$, the mass gap of the 4D theory is determined by the lowest nonzero eigenvalue of this spatial operator around the vacuum. For a flat connection ($F_A = 0$), the identity reduces to:

$\Delta_{\text{Hodge}} = \nabla_A^\ast \nabla_A + \text{Ric}$

On the round $S^3$ of radius $R$, the Ricci tensor is $R_{ij} = (2/R^2)\,g_{ij}$. Since $\nabla_A^\ast \nabla_A \geq 0$ and $\text{Ric} = 2/R^2 > 0$, every gauge fluctuation around any flat connection satisfies:

$\lambda_{\min} \geq \frac{2}{R^2}$

This bound is universal: it holds for any quotient $S^3/\Gamma$ inheriting the round metric, independent of $\Gamma$ and the choice of flat connection. The existence of a positive floor is topological: $S^3/2I$ has finite fundamental group and universal cover $S^3$, so it admits no flat metric, and any Riemannian metric has positive scalar curvature. The numerical value $2/R^2$ depends on the round metric. The actual lowest eigenvalue exceeds this floor. The following sections compute it exactly.

II. Three Vacua

Flat $\text{SU}(2)$ connections on $M$ are classified by $\text{Hom}(2I, \text{SU}(2))/\text{conj}$. A homomorphism $\rho: 2I \to \text{SU}(2)$ is equivalently a 2-dimensional unitary representation with determinant 1. The normal subgroups of $2I$ are $\{1\}$, $\{\pm 1\}$, and $2I$ itself, yielding three possibilities: the trivial map, and two inequivalent faithful embeddings distinguished by their character values on elements of order 10.

The standard and Galois embeddings are distinguished by the golden ratio: $\rho_{\text{std}}$ has trace $\varphi = (1+\sqrt{5})/2$ on order-10 elements, while $\rho_{\text{gal}}$ has trace $\bar\varphi = (1-\sqrt{5})/2$. Since $\text{SU}(2)$ conjugation preserves traces and $\varphi \neq \bar\varphi$, the two define distinct conjugacy classes. The kernel $\{\pm 1\}$ case (factoring through $A_5$) produces no additional class because $A_5$ has no faithful 2-dimensional representation.

Vacuum isolation

Vanishing $H^1(M;\,\text{ad}\,\rho) = 0$ at each flat connection guarantees that no continuous moduli connect the families and no massless Goldstone modes bridge them. For the irreducible flats (standard and Galois), this follows from standard deformation theory. For the trivial connection, $H^1(M;\,\text{ad}\,\rho_\text{triv}) = H^1(M;\,\mathbb{R})^3$; since $\pi_1(M) = 2I$ is a perfect group ($2I$ equals its own commutator subgroup), $H_1(M;\,\mathbb{Z}) = 0$, so $H^1(M;\,\mathbb{R}) = 0$ by universal coefficients. Each vacuum is isolated with positive-definite Hessian. The number three is the count of flat $\text{SU}(2)$ connections on $S^3/2I$. Three vacua, three families.

III. The Spectral Filter

The McKay graph of $2I$ is the extended (affine) $E_8$ Dynkin diagram. Each node corresponds to an irreducible representation of $2I$; the edges encode tensor product with the fundamental 2-dimensional representation. The restriction of the spin-j representation of $\text{SU}(2)$ to $2I$ follows the McKay recursion along this graph.

Coexact spectrum on $S^3$

On the round $S^3$ of radius $R$, coexact (divergence-free) 1-forms organize into levels $k = 1, 2, 3, \ldots$ with eigenvalue:

$\lambda_k = \frac{(k+1)^2}{R^2}$

On the quotient $M = S^3/2I$, physical modes at level $k$ must satisfy a twisted equivariance condition: the adjoint representation $\sigma$ of the vacuum must appear in the McKay decomposition of the spin content at that level. If $\sigma$ is absent, the level is filtered out entirely.

Standard vacuum (3a)

The adjoint representation 3a is precisely $V_1 \lvert_{2I}$: it appears at the first spin level. The first allowed coexact mode sits at $k = 1$:

$\Delta^2 = \frac{(1+1)^2}{R^2} = \frac{4}{R^2}$

Galois vacuum (3b)

On the round $S^3$, coexact 1-forms at level $k$ carry $\text{SU}(2)$ representation content determined by the Peter-Weyl decomposition under the full isometry group. Restricting to $2I$ via the McKay recursion determines which $2I$ irreps appear at each level. The adjoint 3b ($R_4$) sits at graph distance 6 from $R_0$ on the $E_8$ diagram (through the branch at $R_8$). The twisted equivariance condition (adjoint must appear in the $2I$-decomposition of the coexact representation at level $k$) filters levels $k = 1$ through $k = 4$. The first allowed coexact mode for the 3b sector sits at $k = 5$:

$\Delta^2 = \frac{(5+1)^2}{R^2} = \frac{36}{R^2}$

The $k = 5$ claim is verifiable by explicit computation from the character table of $2I$: decompose the coexact 1-form representation at each level $k = 1, \ldots, 5$ into $2I$ irreps and check where $R_4$ first appears. This computation is listed in the falsification table.

The 9x enhancement is topological: it counts filtered levels, not metric parameters. The mass hierarchy between generations follows from this filter. Each vacuum accesses different levels of the McKay decomposition. The Galois vacuum sees the largest gap. Vacuum-to-generation assignment is open; no universal mass hierarchy holds across irreps.

IV. Three Pillars

The mass gap result rests on three independent arguments, each verifiable with standard mathematical tools:

What is topological vs metric-dependent

The existence of the gap, the discreteness of vacua, the vanishing $H^1$, the number of filtered levels, and the 9x enhancement are all topological. The numerical value of $\lambda_{\min}$ scales as $1/R^2$ and is metric-dependent.

Connection to MIT

Within Mode Identity Theory, $R$ is fixed by cosmological parameters, making $\Delta^2 = 4/R^2$ a determinate physical constant. The same positive curvature $\text{Ric}(S^3) = 2/R^2$ that guarantees confinement here enters the Gauss-Codazzi conversion that produces Λ. One geometry, two consequences: the cosmological constant and the mass gap share a common origin in the curvature of $S^3$.

The $2I$ structure performs three roles: it partitions phase space into the 120-domain (the scaling law), it filters gauge fluctuations into three isolated vacua with computed mass gaps (this result), and it provides the spectral geometry whose McKay multiplicities are the $E_8$ root system.

V. Falsification

Every prediction is checkable by finite computation.

Physical predictions (conditional on compact topology)

---

SU(2) Yang-Mills on the Poincaré Homology Sphere has a positive mass gap. Positive curvature forces it. Three isolated vacua produce three families. The McKay correspondence filters the Galois sector through four empty levels, yielding a ninefold enhancement at the branch tip of the $E_8$ graph. The existence of the gap is topological, its value is geometric. The computation reduces to finite group representation theory, standard gauge theory, and Riemannian geometry.

On curved ground, confinement is easy money.