Simply connected domain

A region $\Omega$ in ${\mathbb{R}}^2$ or ${\mathbb{R}}^3$ is simply connected if it is connected (one piece) and every closed curve in $\Omega$ can be continuously contracted to a point without leaving $\Omega$. No holes.

Examples

• ${\mathbb{R}}^2$, ${\mathbb{R}}^3$: simply connected.
• Any open disk, rectangle, ball: simply connected.
• Any half-plane or half-space: simply connected.
• ${\mathbb{R}}^2\setminus \{(0,0)\}$ (plane with origin removed): not simply connected. A loop around the origin cannot shrink to a point without crossing the missing origin.
• An annulus (washer-shaped region in 2D): not simply connected. Same reason.
• ${\mathbb{R}}^3\setminus \{0\}$: simply connected (in 3D you can route around a point, so the missing point doesn’t obstruct loop contraction).
• ${\mathbb{R}}^3$ minus a line (e.g., the $z$-axis): not simply connected. A loop linking the line cannot be contracted without crossing it.

The 3D case trips people up. A single missing point is not an obstruction in 3D (loops can detour around it through the third dimension), but a missing line is. The general rule for ${\mathbb{R}}^n$: removing an $(n-2)$-dimensional set creates a non-simply-connected hole.

Why it matters in vector calculus

The cross-partial test for conservativeness is local: $\nabla \times \mathbf{F}=0$ at every point. Conservativeness itself is global: zero circulation around every closed loop.

On simply connected domains the two coincide: local zero curl implies global zero circulation. Why: every closed loop in a simply connected region is the boundary of a 2D surface inside the region (the loop’s “interior”), so by Stokes’ theorem (or Green’s theorem in 2D) the circulation equals the flux of curl through that surface, which is zero if curl is zero.

On non-simply-connected domains the implication fails. A loop encircling a hole has no interior in $\Omega$ (any candidate interior would cross the hole), and Stokes’ / Green’s doesn’t apply.

The cautionary 2D field

$\mathbf{F}=⟨-y/(x^2+y^2),x/(x^2+y^2)⟩$ on ${\mathbb{R}}^2\setminus \{(0,0)\}$ satisfies the cross-partial test everywhere but has circulation $2\pi$ around the unit circle. Not conservative on its natural domain.

Restrict to a simply connected sub-domain (e.g., the right half-plane $\{x>0\}$) and the same $\mathbf{F}$ is conservative there, with potential $\arctan (y/x)$. The simply-connected restriction is what makes the local test sufficient.

In complex analysis

Same phenomenon, sharper, for contour integrals. ${\int }_{C}dz/(z-z_0)$ is $2\pi i$ if $C$ encloses $z_0$, and $0$ if it doesn’t. The integrand $1/(z-z_0)$ is analytic on $\mathbb{C}\setminus \{z_0\}$, which is not simply connected, so the missing point obstructs loop contraction. The Residue theorem is the systematic extraction of these non-contractible contributions.

Cauchy’s theorem requires the integrand to be analytic on a simply connected domain. Drop simply connectedness and contour integrals stop vanishing.

Theorem-application checklist

Always check:

• Cross-partial test for conservativeness: requires simply connected domain for the converse.
• Green’s theorem: requires the bounded planar region to have a simple closed boundary (special case of simply connected: a region homeomorphic to a disk).
• Stokes’ theorem: the oriented surface must have boundary equal to the given curve — no holes that the curve encircles.
• Cauchy’s theorem in complex analysis: requires $f$ analytic on a simply connected domain containing the contour.

Skipping these checks is where most errors come from. The textbook 2D rotational field is the standard reminder.