What exactly is the Yang–Mills mass gap problem?

The problem asks us to turn an extraordinarily successful physical framework into a completely defined quantum theory, and then prove that its vacuum is separated from every nonzero physical excitation by a positive energy threshold.

This lesson assumes the Lagrangian and Hilbert-space language of a first QFT course. It introduces the target; later lessons explain the machinery.

By the end:

• Separate existence from a positive physical gap.
• Identify the dimension and gauge-group quantifier.

Start with a familiar action

Let $G$ be a compact simple gauge group. For $SU(N)$, choose Hermitian generators with $\operatorname{tr}(T^aT^b)=\delta^{ab}/2$. Put the coupling outside the curvature:

$$F_{\mu\nu}=\partial_\mu A_\nu-\partial_\nu A_\mu-i[A_\mu,A_\nu], \qquad S_E[A]=\frac{1}{2g^2}\int d^4x\,\operatorname{tr}F_{\mu\nu}F_{\mu\nu}.$$

We work at zero temperature and $\theta=0$. “Pure” means there are no dynamical quarks, Higgs fields, or other matter fields. External test charges can still be used to probe the theory.

The classical action contains no mass parameter. Expanding around $A_\mu=0$ produces the familiar perturbative description in terms of massless gauge fields. Yet the physical states expected in the strongly interacting theory are gauge-invariant excitations, including glueballs. Understanding that distinction is the subject of the next lesson.

Two questions hidden in one name

Existence asks for a quantum theory whose observables, correlations, state space, and time evolution satisfy suitable mathematical requirements. A formal expression such as

$$\langle\mathcal O\rangle =\frac{\int \mathcal DA\,\mathcal O[A]e^{-S_E[A]}} {\int \mathcal DA\,e^{-S_E[A]}}$$

suggests what to compute, but does not define an infinite-dimensional integral by itself. A successful construction must supply actual mathematical objects and show that they describe Yang–Mills theory.

The mass gap is a property of the resulting physical Hamiltonian. Set the vacuum energy to zero, write the vacuum as $\Omega$, and suppose $H\Omega=0$. The desired statement is

$$\operatorname{spec}(H)\cap(0,\Delta)=\varnothing \quad\text{for some }\Delta>0,$$

with a finite, nonzero threshold for excitations. The notation $\operatorname{spec}(H)$ means the spectrum: possible energies, including continuous spectral values. It need not be a list of isolated levels.

The official Jaffe–Witten description requires a nontrivial four-dimensional theory for every compact simple gauge group, appropriate axiomatic properties, and a positive gap. Its discussion also specifies the intended short-distance Yang–Mills behavior. Calculating the entire glueball spectrum is a further goal, not an entry condition for solving the problem.

Why a cutoff is not the endpoint

On a finite lattice, each link carries an element of a compact group. There are finitely many integrals, and the partition function is well defined. This is a valuable starting point.

Two independent scales then matter. The lattice spacing $a$ limits spatial resolution; the physical box size $L$ limits the largest distances. The desired theory requires control as $a\to0$ and as $L\to\infty$, along a suitable choice of bare coupling. Their order, or a coordinated joint limit, must be justified.

A positive energy at one spacing and one box size does not tell us what happens in those limits. A sequence of positive numbers can converge to zero. Likewise, convergence of a few observables does not automatically provide the full physical quantum theory.

This is why the Proof Map records separate outcomes: continuum construction, infinite volume, physical reconstruction, identification of the theory, and the physical spectral gap.

What would count as progress?

Different achievements answer different questions:

Result	What it contributes
A controlled lattice area law	Information about large loops in a specified regulated theory
A numerical glueball spectrum	Quantitative evidence with statistical and systematic errors
An ultraviolet stability estimate	Control of fluctuations at short scales
A lower-dimensional construction	A rigorous model and methods that may suggest extensions
A gap in a compactified deformation	A controlled physical mechanism in that setting

Each can be important. Its hypotheses determine what it contributes to the four-dimensional target.

The Clay problem page lists the problem as unsolved at the official-source check on 24 October 2026. Our task as readers is to understand the strongest partial results without dropping their conditions.

Exercises

1. A sequence of gaps

A regulated model has gap $\Delta_n=1/n$ for each positive integer $n$. Has a positive limiting gap been established?

Hint. Compare “positive for each $n$” with “bounded below by a common positive number.”

Solution. No. Every regulated model is gapped, but $\inf_n\Delta_n=0$. A limit argument needs the relevant uniform estimate and a justification that the spectral statement passes to the limiting theory.

2. Diagnose a proposed shortcut

Someone constructs a free massive scalar theory in four dimensions and observes that it has a gap. Which part of the target is missing?

Hint. Ask what identifies the theory.

Solution. A gapped quantum theory is not automatically Yang–Mills. The field content, gauge-invariant observables, interactions, and short-distance behavior must correspond to the target. Existence and a gap cannot be detached from that identification.

Continue at your own depth

The exact construction target develops this lesson without adding a prerequisite to the next lesson.

Read further

• Jaffe and Witten, Quantum Yang–Mills Theory, §§3–5, printed pages 5–7: axiomatic setting, target, finite mass threshold, and extensions.
• David Tong, Gauge Theory, for a graduate physics introduction.

Course overview · Next: Physical states