Yang–Mills Existence and the Mass Gap

The four-dimensional Yang–Mills problem contains two logically separate tasks. First construct a nontrivial quantum theory for a compact simple gauge group on $\mathbb R^4$ with axiomatic control at least as strong as the standard Wightman or Osterwalder–Schrader frameworks. Then prove that the translation or Hamiltonian spectrum above the vacuum begins at a strictly positive energy. A finite lattice measure, perturbative renormalizability, numerical glueball masses, confinement diagnostics, and lower-dimensional constructions each address part of the motivation; none supplies both the continuum object and the spectral theorem.

Required background. Perturbative gauge-QFT constructions fix formal-series scope. The non-Abelian mass-gap statement fixes its physical meaning, and existence, uniqueness, and equivalence claims fix the constructive target. Helpful background. Lines of constant physics explain continuum extrapolation, while QCD confinement and the Yang–Mills mass gap supplies the physical evidence.

The four-dimensional Yang–Mills object

Fix a compact simple group $G$. In Euclidean language one possible target is a compatible family of gauge-invariant Schwinger functions for local composite fields on $\mathbb R^4$ satisfying Euclidean covariance, symmetry, reflection positivity, regularity, clustering, and nontriviality. Osterwalder–Schrader reconstruction would then produce a positive Hilbert space, a vacuum, local Lorentzian observables, and a unitary translation representation. An algebraic construction could instead begin with a local net and a vacuum state, but it must deliver comparably strong locality, covariance, positivity, and spectral structure.

The official problem asks for a nontrivial quantum Yang–Mills theory for every compact simple $G$ and a gap $\Delta>0$ Jaffe and Witten 2000, §§3–4, pp. 5–7. In a reconstructed vacuum representation, write the joint energy–momentum spectrum as

$$\operatorname{sp}(H,\mathbf P) \subset \overline V_+, \qquad H\Omega=0.$$

A vacuum gap means, after any required vacuum-uniqueness statement,

$$\operatorname{sp}(H)\cap(0,\Delta)=\varnothing.$$

This is an infinite-volume continuum assertion. On a periodic finite lattice every Hamiltonian can have a positive first level simply because the state space and box are regulated. A useful theorem needs uniform control through the thermodynamic and continuum limits and must identify the limiting Hamiltonian or translation representation.

As of 2026-08-10, the Clay Mathematics Institute continues to list the problem as unsolved and states that no proof of the mass gap is known Clay Mathematics Institute 2026, current problem page. This dated institutional status is consistent with, but logically separate from, the mathematical formulation in the original problem description.

What the lattice supplies

For a finite hypercubic lattice, the Wilson law

$$d\mu_{a,L}(U)=Z_{a,L}^{-1} \exp\!\left[-\beta(a)\sum_p \left(1-\frac1{\dim F}\operatorname{Re}\operatorname{tr}_F U_p\right)\right] \prod_e dU_e$$

is a genuine positive, gauge-invariant probability measure because the product of compact Haar spaces is finite-dimensional. The standard Wilson action also has the required finite-cutoff reflection-positivity property Osterwalder and Seiler 1978, §§2–4, pp. 440–458. These are indispensable inputs, not the continuum conclusion.

Regulator removal still requires a tuning prescription for $\beta(a)$, an ordered $L\to\infty$ and $a\to0$ limit, tightness or convergence for a separating collection of renormalized gauge-invariant observables, Euclidean regularity and covariance in the limit, reconstruction, and nontriviality. Wilson loops alone are nonlocal and do not automatically provide all local fields needed by an axiomatic construction.

Recent rigorous scaling work illustrates the importance of the model label. Chatterjee constructs, under a joint weak-coupling and large-Higgs-length scaling, a projected massive Gaussian limit for $SU(2)$ lattice Yang–Mills–Higgs in any $d\ge2$ 2026, Theorem 3.2 and §3.3, pp. 12–17. Higgs matter, Gaussianity, and the fast coupling scaling are part of that theorem. It neither removes the regulator for four-dimensional pure Yang–Mills nor proves its spectral gap.

First application: interpreting a glueball spectrum

Return to the non-Abelian spectral statement. A continuum-extrapolated lattice glueball calculation provides evidence for masses extracted from long Euclidean-time decay of chosen lattice operators. To turn that into the mathematical theorem one must separately establish:

1. convergence of the relevant renormalized correlations in the continuum and infinite-volume limits;
2. reflection positivity and reconstruction of a limiting Hilbert theory;
3. vacuum uniqueness or a precise sector statement;
4. a spectral lower bound uniform over all states orthogonal to the vacuum, not only the measured channels; and
5. nontrivial local observables and the remaining axioms.

The numerical mass is valuable quantitative evidence. Its error analysis does not establish tightness of the full measure or exclude unseen spectral weight in every gauge-invariant channel.

Mass gap and confinement are not equivalent

A mass gap implies exponential clustering for suitable local vacuum correlations under the usual axiomatic assumptions. It does not by itself imply an area law for Wilson loops, absence of colored states, a linearly rising static potential, or asymptotic completeness. Conversely, an area law established at strong lattice coupling can disappear along the path toward the continuum, and it is not by itself a construction of a continuum Hamiltonian.

The separation is also visible in other models: the Schwinger model is gapped and screens charge, while gauge–Higgs systems may connect regimes without a universal local order parameter. These are counterexamples to slogan-level implications, not models of four-dimensional pure Yang–Mills.

Failure test: a finite-volume gap

Assume transfer-matrix eigenvalues at each $a,L$ obey $E_1(a,L)>0$. Without a lower bound uniform along the tuned continuum sequence, the gap may tend to zero. Even a uniform bound for one symmetry channel does not control the full orthogonal complement of the vacuum. The strongest surviving result is the regulated spectral estimate in the declared box and channel.

Exercises

Why does reflection positivity at every lattice spacing not by itself give a continuum Hilbert space?

Solution

Reflection positivity passes to an appropriate limiting hierarchy only after that limit is shown to exist. One still needs tightness or correlation convergence, regularity, Euclidean covariance, and a complete set of limiting Schwinger functions. Without a limit there is no continuum positive form on which OS reconstruction can act.

References

• Chatterjee, Sourav. “A Scaling Limit of $SU(2)$ Lattice Yang–Mills–Higgs Theory.” Probability and Mathematical Physics 7 (2026): 339–381. DOI; Open PDF.
• Clay Mathematics Institute. “Yang–Mills and the Mass Gap.” Current through 2026-08-10. Problem page.
• Jaffe, Arthur, and Edward Witten. “Quantum Yang–Mills Theory.” In The Millennium Prize Problems, 129–152. Clay Mathematics Institute and American Mathematical Society, 2006; problem description released 2000. Official PDF.
• Osterwalder, Konrad, and Erhard Seiler. “Gauge Field Theories on a Lattice.” Annals of Physics 110 (1978): 440–471. DOI.