Mini-Lecture I.3: Holonomy and Parallel Transport of $D_{\mu }(x)$

We study ordered transport and the hopping identity. The worked calculation and its limitations are the classroom target; the longer source reconstruction supports independent study.

How to use this lecture

Prerequisites: I.2; exponential series and differentiation. Review the relevant prerequisite bridge before starting.

• Classroom core, 90 minutes: question and prerequisites (10); ordered transport and the hopping identity (25); worked calculation or laboratory (30); problems 1–2 (20); written exit check (5).
• Self-study, approximately 2–3 hours: reconstruct the algebra in the main text, work through the laboratory, then solve problems 1–3 before reading the answer checkpoints. The estimate is provisional.
• Research extension: problem 4 states the source-dependent task and what counts as completion. Its open claims are not assumed in the classroom assessment.

The classroom result is the explicitly calculated model or conditional derivation below. Retained sections marked Source reconstruction explain the IAS argument and identify provenance; that label does not certify the argument. The claim map records the research limits.

Reading

• Primary: IAS lectures, pages 7-12. Consult the page images when an equation is clipped or extraction is ambiguous.
• Gentler route: the prerequisite bridges, followed by this lecture’s worked calculation.
• Optional research: the bibliography and version map identifies the primary papers for this module and their role.

Source Pages

Primary source: IASQCDLectures.pdf, pages 7-12.

The question

How can a product of covariant-derivative operators at one coordinate encode transport around a curve? The mechanism is ordinary translation acting on multiplication operators. Tracking two factors explicitly fixes the direction of every shift and prevents a silent reversal of path ordering.

Translation as an operator identity

Let $M_f$ multiply a test function by a matrix-valued $f(x)$ and let $(T_h\psi )(x)=\psi (x+h)$. Then

$$T_h{M}_f=M_{f(x+h)}{T}_h.$$

This equality follows by applying both sides to $\psi$. It does not mean that $f$ commutes with differentiation. Indeed $[{\partial }_{\mu },M_f]=M_{{\partial }_{\mu }f}$.

Use $D_{\mu }={\partial }_{\mu }+{\mathcal{A}}_{\mu }$ with the conventions of I.1. For a small smooth displacement $h$,

$$e^{h\cdot D}=M_{e^{h\cdot \mathcal{A}(x)}}{T}_h+O(|h{|}^2).$$

For bounded matrices the second-order error is governed by the commutator; the laboratory derives it. Differential operators require a common domain and derivative bounds. For a smooth contour with $N$ steps of order $1/N$, uniform estimates make the sum of these errors tend to zero. This argument does not apply unchanged to Brownian increments of order $N^{-1/2}$, where quadratic terms can survive.

Two factors determine the order

Set $V_j(x)=e^{h_j\cdot \mathcal{A}(x)}$ and deliberately write earlier contour segments on the left:

$$E_N=e^{h_1\cdot D}{e}^{h_2\cdot D}\cdots e^{h_N\cdot D}.$$

For two split factors,

$$M_{V_1}{T}_{h_1}{M}_{V_2}{T}_{h_2}=M_{V_1(x)V_2(x+h_1)}{T}_{h_1+h_2}.$$

For three, the same step gives the multiplier

$$V_1(x)V_2(x+h_1)V_3(x+h_1+h_2)$$

and the translation $T_{h_1+h_2+h_3}$. Induction yields

$$E_N\simeq M_{V_1(x)V_2(x+h_1)\cdots V_N(x+h_1+\cdots +h_{N-1})}{T}_{h_1+\cdots +h_N}.$$

The approximation is only the finite-step splitting; moving translations past multiplication operators is exact. The matrix factors retain their order even though the translations commute with one another.

Relation to the parallel-section convention

In the smooth limit the multiplier $V(s)$ obeys

$$\frac{dV}{ds}=V(s){\dot{C}}^{\mu }(s){\mathcal{A}}_{\mu }(C(s)),V(0)=1.$$

By contrast the parallel transport $U$ of I.1 obeys $U^{\prime }=-{\dot{C}}^{\mu }{\mathcal{A}}_{\mu }U$. Differentiating $VU$ shows $(VU{)}^{\prime }=0$, so $V=U^{-1}$. The plus sign together with earlier factors on the left thus represents inverse transport along the stated contour. This is an explicit convention dictionary.

If the derivative factors are written in the reverse order, the accumulated arguments and contour traversal must also be reversed. One cannot keep $C_k=x+{\sum }_{j<k}{h}_j$ while silently reversing the product. The source’s ordered identity on pages 7–11 must be read with its accompanying orientation.

For a closed polygon, ${\sum }_j{h}_j=0$ and the total translation becomes the identity. The remaining operator is multiplication by the matrix holonomy $V_C(x)$. It is not a scalar Wilson expectation times the identity on color space. Taking the color trace and then a gauge expectation are two further operations.

Gauge covariance of the operator presentation

Covariant derivatives satisfy $D_{\mu }^S=M_S{D}_{\mu }{M}_{S^{-1}}$. The adjacent conjugations cancel in a product of exponentials, giving $E_N^S=M_S{E}_N{M}_{S^{-1}}$. If $E_N=M_V{T}_H$, then

$$E_N^S=M_{S(x)V(x)S^{-1}(x+H)}{T}_H.$$

Thus inverse transport carries one transformation at each endpoint. When $H=0$, $V_C^S(x)=S(x)V_C(x)S^{-1}(x)$ and its color trace is gauge invariant.

The combination $D=\partial +\mathcal{A}$ is a covariant differential operator. Its covariance includes the derivative acting on the gauge transformation; it is incorrect to ascribe all of it to treating $\mathcal{A}$ as an ordinary conjugated matrix.

A constant non-Abelian rectangle

Take constant matrices $X,Y$ as the two connection components and traverse small increments $(ϵ,0)$, $(0,ϵ)$, $(-ϵ,0)$, $(0,-ϵ)$ in the earlier-on-left convention. The multiplier is

$$V_{\square }=e^{ϵX}{e}^{ϵY}{e}^{-ϵX}{e}^{-ϵY}=1+{ϵ}^2[X,Y]+O({ϵ}^3).$$

Expanding each factor through second order cancels all linear terms and the individual $X^2,Y^2$ terms, leaving $XY-YX$. Reversing the traversal changes its sign. This is the finite curvature checkpoint for the area insertion in I.4.

For $X=i{\sigma }_1$ and $Y=i{\sigma }_2$, which are anti-Hermitian, $[X,Y]=-2i{\sigma }_3$. The untraced matrix insertion is nonzero even though its trace vanishes at this order. Surrounding transport matters for general traced insertions.

What is established

Translation conjugation, the finite ordered-product rearrangement, gauge covariance, and the rectangle expansion are explicit operator or finite-matrix calculations. The smooth continuum limit needs the stated estimates; unbounded operators and rough paths need additional analysis. These distinctions delimit the holonomy dictionary used in loop calculus.

Worked laboratory: checking the translation and its direction

For $T_a=e^{a{\partial }_x}$, Taylor’s theorem gives $T_{a}h(x)=h(x+a)$ for analytic test functions. Let $M_f$ denote multiplication by $f$. Acting on a test function rather than manipulating symbols alone,

$$(T_a{M}_f{T}_{-a}h)(x)=f(x+a)h(x).$$

Thus $T_a{M}_f{T}_{-a}=M_{f(x+a)}$. For $f(x)=x^2$ and $h(x)=x+1$, both sides give $(x+a{)}^2(x+1)$. Reversing the translation reverses the shift.

For two bounded matrices $X,Y$, expansion through order ${ϵ}^2$ gives

$$e^{ϵX}{e}^{ϵY}-e^{ϵ(X+Y)}=\frac{{ϵ}^2}{2}[X,Y]+O({ϵ}^3).$$

Both small increments enter the commutator. With $N$ slices of size $1/N$, the accumulated first-order splitting error is typically $O(1/N)$ under uniform boundedness assumptions. Differential operators and rough paths require additional domain estimates; the finite-matrix estimate does not prove their continuum limit.

Problem set

1. Classroom core. Apply $e^{a{\partial }_x}$ to $x^2$.

2. Self-study calculation. Derive the second-order difference of the two exponentials in the laboratory.

3. Self-study interpretation. Why must a path-ordering convention be fixed before assigning an area-derivative sign?

4. Research extension. Translate the source’s operator product to the parallel-section convention in I.2.

Answer checkpoints

1. The result is $(x+a{)}^2=x^2+2ax+a^2$.

2. The mixed term in the product is $XY$; that in the single exponential is $(XY+YX)/2$, leaving $[X,Y]/2$.

3. Exchanging the order of two neighboring insertions reverses their commutator and the associated oriented plaquette.

4. Completion: write the transport equation, orientation, endpoint transformation, and a small-plaquette check in both conventions. A dictionary with matched signs is the requested result.

Teaching note

Use the decisive step in problem 2 as the written exit check for I.3; assign problem 3 only after students have named the assumptions. Grade the research extension by its documented method and justified conclusion, including an unresolved result when evidence is missing. The timing above awaits classroom rehearsal.

Previous note · Next note · Course guide