Principal-Bundle Sectors and Large Gauge Transformations

Gauge-field configuration space is first separated by the isomorphism class of the principal bundle and only then quotiented by the allowed gauge group within each class. Large gauge transformations are disconnected components of that group; they need not change the bundle class. Boundary conditions, global form, and matter representations decide which transformations are quotiented and therefore which apparent labels are physical.

Required background. Gauge configuration groupoids supplies the quotient object. Principal and associated bundles supplies transition functions and characteristic classes. Global form and faithful gauge group determines which bundle automorphisms act trivially on all fields. Helpful background. Large gauge transformations and topological sectors distinguishes bundle labels from disconnected transformations. Theta terms and periodicity explains how sector sums enter quantum amplitudes.

Bundle classes precede gauge orbits

For a fixed principal $G$-bundle $P\to M$, connections form the affine space $\mathcal A(P)$ and gauge transformations form $\mathcal G(P)=\operatorname{Aut}_M(P)$. The complete classical configuration groupoid over the declared class of bundles is schematically

$$\coprod_{[P]}[\mathcal A(P)/\mathcal G(P)].$$

The coproduct is not optional bookkeeping: connections on nonisomorphic bundles cannot be compared by an automorphism of one fixed $P$. Within a sector, $\pi_0\mathcal G(P)$ records large gauge transformations. Whether one divides by all components, only the identity component, or a subgroup fixed at a boundary changes the quotient. None of these choices by itself changes the characteristic class of $P$.

The faithful gauge group also depends on matter. If a subgroup of the center acts trivially on every field, then the effective global form may be $G/\Gamma$ rather than $G$. The allowed bundles and characteristic classes can consequently change. A Lie algebra alone does not determine this information.

U(1) bundles on the two-sphere

The first application is the sector analysis in Large Gauge Transformations and Topological Sectors. Principal $U(1)$ bundles over $S^2$ are classified by

$$c_1(P)=\frac{1}{2\pi}\int_{S^2}F=n\in\mathbb Z.$$

Cover $S^2$ by northern and southern charts. In the real-potential convention $\operatorname{Hol}_\gamma(A)=\exp(i\int_\gamma A)$, choose

$$A_N=\frac n2(1-\cos\theta)\,\mathrm d\phi, \qquad A_S=-\frac n2(1+\cos\theta)\,\mathrm d\phi.$$

Each potential is regular at the pole contained in its chart. On the overlap,

$$A_N-A_S=n\,\mathrm d\phi, \qquad g_{NS}=e^{in\phi},$$

and single-valuedness of $g_{NS}$ forces $n\in\mathbb Z$. Both local expressions give

$$F=\mathrm dA_N=\mathrm dA_S =\frac n2\sin\theta\,\mathrm d\theta\wedge\mathrm d\phi,$$

whose integral is $2\pi n$. Wu and Yang’s global monopole formulation makes the necessity of overlapping regular potentials and a transition function explicit in Wu and Yang 1975, pp. 3848–3851.

If $n\neq0$, there is no globally defined one-form $A$ with $F=\mathrm dA$: Stokes’ theorem on closed $S^2$ would give $\int_{S^2}F=0$. This is an independent, convention-insensitive check that the nontrivial sector cannot be represented by one global potential. Changing a local trivialization changes $A_N,A_S$, not $n$.

Large transformations are a different question. On a fixed bundle, a gauge transformation is a global section of the adjoint group bundle. Its connected components can be nontrivial on other base manifolds or under boundary conditions, yet it remains an automorphism of that same $P$. A singular transformation that changes a characteristic number is not an allowed smooth automorphism in the original configuration space.

Boundary conditions change the quotient

Let $M$ have boundary. The subgroup

$$\mathcal G_0(P)=\{u\in\mathcal G(P):u|_{\partial M}=1\}$$

acts as redundancy when boundary data are fixed. A transformation with nontrivial boundary value need not be quotiented; it may change an edge degree of freedom or carry a boundary charge. If instead all boundary values are declared gauge, the quotient is smaller. Thus a “sector count” without the boundary gauge group and boundary conditions is incomplete.

This distinction also limits theta-angle reasoning. A topological weight may distinguish bundle classes or winding sectors in a path integral, but periodicity follows only after the normalization of the characteristic number and the allowed set of bundles are fixed. Conversely, a disconnected gauge group does not automatically imply distinct quantum superselection sectors; the quantum representation must still be specified.

Failure test

The adversarial configuration is the $n=1$ bundle above. Assume one global potential exists. Then $F=\mathrm dA$ and Stokes gives zero flux, contradicting $\int F=2\pi$. Alternatively, keep a manifold with boundary and quotient by a transformation whose boundary value was explicitly forbidden; the resulting quotient identifies states with different boundary data. Either move changes the declared problem rather than simplifying it.

The useful converse boundary is equally sharp: zero $c_1$ guarantees topological triviality for $U(1)$ bundles on $S^2$, but vanishing of one characteristic class is not a universal triviality criterion for arbitrary groups and bases.

Exercises

Verify the overlap relation, flux, and impossibility of a global potential for the charge-$n$ monopole bundle.

Solution

Subtracting gives $A_N-A_S=n\,\mathrm d\phi$, corresponding to $g_{NS}=e^{in\phi}$. Differentiation gives $F=(n/2)\sin\theta\,\mathrm d\theta\wedge\mathrm d\phi$, so $\int_{S^2}F=(n/2)(2)(2\pi)=2\pi n$. If $A$ were global, Stokes would give $\int_{S^2}\mathrm dA=0$. Hence a global potential exists only in the $n=0$ sector.

References

• Atiyah, Michael F., and Raoul Bott. “The Yang–Mills Equations over Riemann Surfaces.” Philosophical Transactions of the Royal Society of London A 308 (1983): 523–615. DOI; Open PDF.
• Wu, Tai Tsun, and Chen Ning Yang. “Concept of Nonintegrable Phase Factors and Global Formulation of Gauge Fields.” Physical Review D 12 (1975): 3845–3857. DOI.