Gauge transformations and connection moduli
Separate frame changes from geometry and compute a moduli space where large gauges matter.
On this page
The question for today
What you’ll understand
- Apply one consistent gauge action on connections.
- Identify stabilizers and the limitations of a naive quotient.
- Compute flat U(1) connection classes on a circle.
A connection has many coordinate descriptions, and a fixed bundle can support many geometrically distinct connections. Moduli questions ask which differences survive the allowed gauge transformations. The word gauge can describe a change of local frame or an active bundle automorphism; these viewpoints induce related formulas, but their inverses must be tracked. Here we explicitly choose the action on local matrix potentials. All quotient statements refer to this convention and to smooth gauge maps on the specified bundle, including maps not connected to the identity.
The action and its invariants
A gauge automorphism of a principal bundle covers the identity of the base and commutes with its right action. It can be represented on the total space by an equivariant group-valued function, or locally by compatible maps into the structure group. The displayed action is the pullback convention for connections. On a nontrivial bundle, arbitrary unrelated maps on overlapping charts are not a gauge transformation: their compatibility is inherited from the bundle transitions. A global map from the base to the group describes gauges only after a trivialization has been chosen.
Curvature changes by conjugation, so invariant polynomials and traces of holonomy are unchanged. For a loop based at , the holonomy matrix transforms by . An open-path transport matrix has different gauge factors at its endpoints and is not a scalar gauge invariant by itself. These facts distinguish quantities that can label a moduli problem from quantities that merely encode a chosen frame. They also give direct algebraic checks on a proposed transformation law.
A quotient need not be a manifold
The space of smooth connections on a fixed bundle is affine, modeled on adjoint-valued one-forms. Dividing by the gauge group produces a moduli set; imposing an equation such as flatness gives a smaller moduli problem. A stabilizer consists of gauges fixing a connection and can vary between points. For this reason the naive quotient need not be a smooth manifold. Analytic constructions often use Sobolev completions and slices, together with hypotheses excluding or controlling reducible connections. Those are additional results, not automatic consequences of writing a quotient symbol.
For a connected base, a covariantly constant gauge transformation is determined by its value at one point, and that value must commute with the holonomy group. This relates stabilizers to centralizers. Using gauges fixed at a base point removes this freedom in many transport descriptions, but changes the equivalence relation. Flat connections on bundles over suitable connected manifold bases correspond to holonomy representations up to conjugacy when one allows the corresponding bundle constructions; prescribing one underlying bundle may select only part of that representation space.
Large gauges on a circle
Every smooth connection on the trivial bundle over a circle is flat. Write its imaginary potential as . A periodic gauge can remove the zero-average part of , leaving the constant average . Gauge maps are single-valued precisely for integer and shift this constant by under our action. Thus retaining only gauges homotopic to the identity would produce a different quotient from retaining the full smooth gauge group.
The resulting full moduli set is , detected by the phase . Constant gauges stabilize every connection because the group is abelian; this does not erase the distinction between different phases. The parameter circle is an unusually simple example. For nonabelian groups and more complicated bases, conjugacy classes, singular strata, and compatibility relations among several loops appear. The elementary circle calculation is a model for checking conventions, not a general regularity theorem for connection moduli.
Work it through
Two potentials with the same loop phase
Compare circle potentials and . Find a globally defined gauge relating them, and test whether is equivalent.
Choose , a smooth single-valued circle gauge of winding minus one. Then .
Our action gives . Their positive-loop phases are both .
The half-integer potential has phase . Since , its holonomy invariant excludes gauge equivalence to the first two.
The first pair belongs to one full-gauge class. A fractional parameter change of one third cannot be implemented by a single-valued circle gauge.
The living notebook
Explore the model
Interactive lab · quantum
Relative phase remains observable
Pause/Reset, set mixing angle 90 degrees and relative phase 60 degrees. Predict basis probabilities 1/2,1/2 and Bloch coordinates (1/2,√3/2,0). Reset and set both controls to 90 degrees and −120 degrees: probabilities are unchanged but Bloch coordinates become (−1/2,−√3/2,0). Playback adds phase precession and parameter changes retain time. Relative phase here is observable; it is not an unobservable common U(1) gauge phase or a computation of bundle moduli.
- Probability |0⟩
- 71.1%
- Probability |1⟩
- 28.9%
- Bloch (x, y, z)
- 0.78, 0.45, 0.42
A normalized two-level pure state, not an interacting field simulation. Playback adds one radian per second to the relative phase; basis probabilities stay fixed. Parameter changes pause playback and recompute the model at the current elapsed time. Reset restores both parameters and returns time to zero. Playback starts paused. Adjust either parameter, then choose Play or Step.
Look forUnchanged probabilities in one basis do not imply every invariant or comparison is unchanged.
Pause & reason
Check your understanding
Your turn
Try it yourself
Reduce to a constant representative, and decide whether a gauge connected to the identity can then reduce that constant to zero.
Take a moment to make a prediction or work it out on paper. Open a hint when you need one.
Give me a hint
Differentiate a periodic sine to remove the oscillating term. The remaining average is tested by loop holonomy.
Compare with the solution
The periodic gauge has , so the transformed potential is . This gauge has winding zero. The remaining holonomy is , rather than one, so no globally defined gauge can reduce it to zero. In particular no identity-component gauge can do so. Removing a local oscillation and removing a global loop phase are different operations, even on a topologically trivial underlying bundle.
Keep the idea
Take it with you
- A gauge action requires a declared pullback or inverse convention.
- Stabilizers can obstruct smooth manifold descriptions of a quotient.
- Large circle gauges identify constant potentials modulo integers.
Read further
Original sources and deeper explanations to follow at your own pace.
Optional progress, saved in this browser when available. Revisit any lesson at any time.
Lesson not marked complete.