Calculus IV: Conservative Fields and Green's Theorem

In Part 2, we defined two types of line integrals: the work integral $\int_C \mathbf{F} \cdot \mathrm{d}\mathbf{r}$ and the flux integral $\int_C \mathbf{F} \cdot \mathbf{n}\,\mathrm{d}s$. We noticed that for certain “nice” vector fields, the work integral depends only on the endpoints — not on the path. This post is about understanding why, and about the crown jewel of two-dimensional vector calculus: Green’s theorem, which connects line integrals to double integrals.

The thread running through everything is the idea that local properties of a vector field (how much it spins or expands at a point) determine global properties (the total circulation or flux around a curve). This is the first instance of a pattern that culminates in the general Stokes’ theorem.

This is Part 3 of a four-part series on Calculus IV.

What This Post Covers

• Path Independence — When does the value of a line integral depend only on endpoints?
• The Fundamental Theorem of Line Integrals — The multivariable analogue of $\int_a^b f’(x)\,\mathrm{d}x = f(b) - f(a)$
• Finding Potential Functions — How to recover $f$ from $\nabla f$
• The Component Test — A quick check for whether a vector field could be conservative
• The Tricky $\mathrm{d}\theta$ Form — A cautionary tale about domains and topology
• Curl and Divergence — Local measures of rotation and expansion
• Green’s Theorem — The bridge between line integrals and double integrals
• Applications — Computing area via line integrals, the shoelace formula, and regions with holes

Path Independence and Conservative Fields

We observed in Part 2 that the vector field $\mathbf{F} = x\,\mathbf{i} + y\,\mathbf{j} + z\,\mathbf{k}$ gave the same work integral $\frac{3}{2}$ along two very different paths from $(0,0,0)$ to $(1,1,1)$. This is called path independence: the integral $\int_C \mathbf{F} \cdot \mathrm{d}\mathbf{r}$ depends only on where $C$ starts and ends, not on the route it takes.

A vector field with this property is called conservative. The name comes from physics: for a conservative force, the total energy (kinetic + potential) is conserved.

Path independence has an immediate consequence for closed curves: if $C$ is any closed loop (starting and ending at the same point), then

for a conservative field. This is because we can think of a closed loop as a path from $A$ to $B$ followed by a path from $B$ back to $A$; the second path gives the negative of the first.

The fundamental theorem of line integrals

Why was $\mathbf{F} = x\,\mathbf{i} + y\,\mathbf{j} + z\,\mathbf{k}$ path-independent? Because it is the gradient of $f(x,y,z) = \frac{1}{2}(x^2 + y^2 + z^2)$:

This is the key to the entire theory.

Theorem (Fundamental Theorem of Line Integrals). Let $f\colon \mathbb{R}^n \to \mathbb{R}$ be differentiable with continuous gradient $\mathbf{F} = \nabla f$. Let $C$ be a smooth curve from point $A$ to point $B$. Then

Proof. Using the chain rule, $\frac{\mathrm{d}}{\mathrm{d}t}f(\mathbf{r}(t)) = \nabla f(\mathbf{r}(t)) \cdot \frac{\mathrm{d}\mathbf{r}}{\mathrm{d}t}$. So

The converse is also true: every conservative field is the gradient of some scalar function $f$, called a potential function.

In the language of differential forms, this theorem says $\int_C \mathrm{d}f = f(B) - f(A)$ — the exact analogue of the one-variable fundamental theorem of calculus.

Finding Potential Functions

Given a vector field $\mathbf{F} = M\,\mathbf{i} + N\,\mathbf{j}$ that we believe to be conservative, how do we find the potential function $f$ with $\nabla f = \mathbf{F}$?

The idea is like finding an antiderivative, but with multiple variables. The $+C$ from single-variable calculus becomes much richer: a “constant” with respect to $x$ can be any function of $y$.

Example in $\mathbb{R}^2$

Let $\mathbf{F} = 2xy^3\,\mathbf{i} + 3(x^2 - 1)y^2\,\mathbf{j}$. We want $f$ with $\frac{\partial f}{\partial x} = 2xy^3$ and $\frac{\partial f}{\partial y} = 3(x^2 - 1)y^2$.

Step 1. Integrate $\frac{\partial f}{\partial x} = 2xy^3$ with respect to $x$:

where $g(y)$ is an unknown function of $y$ (the “constant” of integration).

Step 2. Differentiate our expression with respect to $y$ and set it equal to $N$:

The crucial check: $g’(y)$ depends only on $y$, not on $x$. If it still contained $x$, then no potential function would exist.

Step 3. Integrate: $g(y) = -y^3 + C$. The potential function is

Example in $\mathbb{R}^3$

For $\mathbf{F} = (2xy + 3z^2)\,\mathbf{i} + (x^2 + 4z^2)\,\mathbf{j} + (6xz + 8yz)\,\mathbf{k}$:

1. Integrate $\frac{\partial f}{\partial x} = 2xy + 3z^2$ to get $f = x^2 y + 3xz^2 + g(y,z)$.
2. Differentiate with respect to $y$: $x^2 + \frac{\partial g}{\partial y} = x^2 + 4z^2$, so $\frac{\partial g}{\partial y} = 4z^2$, giving $g = 4yz^2 + h(z)$.
3. Differentiate with respect to $z$: $6xz + 8yz + h’(z) = 6xz + 8yz$, so $h’(z) = 0$.

Therefore $f(x,y,z) = x^2 y + 3xz^2 + 4yz^2 + C$.

The Component Test

Before going through the work of finding a potential function, it is useful to have a quick test for whether one could exist at all.

Theorem (Component Test in $\mathbb{R}^2$). If $\mathbf{F} = M\,\mathbf{i} + N\,\mathbf{j}$ is a gradient field, then $\frac{\partial M}{\partial y} = \frac{\partial N}{\partial x}$.

This follows immediately from Clairaut’s theorem: if $f$ exists with $\frac{\partial f}{\partial x} = M$ and $\frac{\partial f}{\partial y} = N$, then $\frac{\partial M}{\partial y} = \frac{\partial^2 f}{\partial y \partial x} = \frac{\partial^2 f}{\partial x \partial y} = \frac{\partial N}{\partial x}$.

In $\mathbb{R}^3$, the test becomes three conditions:

If any one of these fails, $\mathbf{F}$ is definitely not conservative. If all three hold, then $\mathbf{F}$ is usually conservative — but there is a subtle catch.

The Tricky $\mathrm{d}\theta$ Form

Consider the vector field

This is the gradient of $\theta(x,y) = \arctan(y/x)$ — the polar angle. And indeed, $\mathbf{F}$ passes the component test: $\frac{\partial M}{\partial y} = \frac{\partial N}{\partial x}$ everywhere that $\mathbf{F}$ is defined. So is it conservative?

The answer is: it depends on the domain.

The problem is that $\theta$ is not a continuous function on all of $\mathbb{R}^2 \setminus {(0,0)}$. As you walk counterclockwise around the origin, $\theta$ increases from $0$ toward $2\pi$, then abruptly jumps back to $0$. No matter where you place the “cut” (the ray where $\theta$ is discontinuous), the function $\theta$ cannot be made continuous on any domain that encircles the origin.

Computing directly on the unit circle $\mathbf{r}(t) = (\cos t, \sin t)$:

A conservative field cannot have a nonzero integral around a closed loop. So $\mathbf{F}$ is not conservative on $\mathbb{R}^2 \setminus {(0,0)}$.

However, on any simply connected domain that avoids the origin (like a half-plane), $\mathbf{F}$ is conservative: the component test works, and a potential function (some branch of $\theta$) exists.

Theorem. If $D$ is an open, simply connected domain (no holes), and $\mathbf{F}$ is defined on all of $D$ with continuous partial derivatives and passes the component test, then $\mathbf{F}$ is conservative on $D$.

The lesson: the component test is necessary but not quite sufficient. Topology matters — specifically, whether the domain has “holes” that a closed curve could wrap around.

Curl and Divergence

We now develop two local measurements of a vector field’s behavior. Both are motivated by thinking about what a tiny closed curve “sees.”

Curl (circulation density)

Imagine tossing a small chip of wood into a flowing stream. Two things happen: it floats downstream, and it spins. The spinning is caused by different parts of the chip experiencing different velocities.

The curl of $\mathbf{F} = M\,\mathbf{i} + N\,\mathbf{j}$ at a point $(x,y)$ quantifies this spinning tendency:

This formula emerges naturally from computing the circulation $\oint_{\square} \mathbf{F} \cdot \mathrm{d}\mathbf{r}$ around a tiny square of side $2h$ centered at $(x,y)$, then dividing by the area $4h^2$:

Notice: $\operatorname{curl}\mathbf{F} = 0$ is exactly the component test. Conservative fields have zero curl everywhere — they produce no local spinning.

Divergence (flux density)

Similarly, the divergence measures the tendency of a vector field to expand at a point — like an air current flowing outward from a high-pressure region.

This comes from computing the outward flux across the boundary of a tiny square and dividing by the area. A positive divergence means the field is “creating” flow at that point (a source); a negative divergence means it is absorbing flow (a sink).

Curl measures local rotation; divergence measures local expansion.

Connection to differential forms

Both curl and divergence arise from the exterior derivative of a 1-form $\phi = M\,\mathrm{d}x + N\,\mathrm{d}y$:

The coefficient is exactly the curl! And the divergence version comes from the flux form of the same integral. This unification is a hint that differential forms are the “right” language for all of this.

Green’s Theorem

We have all the pieces. Green’s theorem is the bridge that connects line integrals (around a boundary) to double integrals (over a region).

Green’s theorem for circulation

Theorem (Green’s Theorem). Let $R$ be a bounded, simply connected region in $\mathbb{R}^2$ with boundary $C$ oriented counterclockwise. Let $\mathbf{F} = M\,\mathbf{i} + N\,\mathbf{j}$ be a vector field with continuous partial derivatives on an open set containing $R$. Then:

In words: the circulation of $\mathbf{F}$ around $C$ equals the integral of the circulation density over $R$.

The proof idea

The proof is a beautiful argument about cancellation.

The proof of Green's theorem: subdivide the region, circulate around each cell, and interior boundaries cancel.

1. Subdivide $R$ into many tiny rectangular cells.
2. The double integral $\iint_R \operatorname{curl}\mathbf{F}\,\mathrm{d}A$ is approximated by summing $\operatorname{curl}\mathbf{F}(x_i, y_i) \cdot \Delta A$ over all cells.
3. By the definition of curl, each term is approximately the circulation $\oint_{\square_i} \mathbf{F} \cdot \mathrm{d}\mathbf{r}$ around the $i$-th cell.
4. When we add up these tiny circulations, each interior boundary segment appears twice — once for each adjacent cell — in opposite directions. They cancel.
5. The only boundary segments that don’t cancel are those on the outer boundary $C$.
6. Therefore: $\iint_R \operatorname{curl}\mathbf{F}\,\mathrm{d}A \approx \sum_i \oint_{\square_i} \mathbf{F} \cdot \mathrm{d}\mathbf{r} \approx \oint_C \mathbf{F} \cdot \mathrm{d}\mathbf{r}$.

Green’s theorem for flux

There is a companion version for flux:

The proof is essentially identical, but with outward flux across each cell boundary instead of circulation around it.

The differential forms perspective

From the viewpoint of differential forms, both versions are the same theorem. If $\phi = M\,\mathrm{d}x + N\,\mathrm{d}y$ is a 1-form, then Green’s theorem says

where $\mathrm{d}\phi = \left(\frac{\partial N}{\partial x} - \frac{\partial M}{\partial y}\right)\mathrm{d}x \wedge \mathrm{d}y$ is the exterior derivative. The circulation version uses the vector line integral form; the flux version uses the flux integral form. Both reduce to integrating the exterior derivative over the region.

Example: circulation via Green’s theorem

Let $\mathbf{F} = x\,\mathbf{i} + xy\,\mathbf{j}$ and let $C$ be the boundary of the region ${(x,y) : -2 \leq x \leq 2,\ -2 \leq y \leq 2 - x^2}$, oriented counterclockwise.

Instead of parameterizing $C$ (which has a straight-line piece and a parabolic piece), we use Green’s theorem:

Setting up the integral:

Since $\frac{x^4}{2}$ and $-2x^2$ are both even functions:

Applications of Green’s Theorem

Area by line integral

Green’s theorem can run “backwards”: instead of simplifying a line integral into a double integral, we can compute a double integral via a line integral. The most common application is area.

The area of a region $R$ is $\iint_R 1\,\mathrm{d}A$. We need a vector field $\mathbf{F}$ with $\operatorname{curl}\mathbf{F} = 1$. A convenient choice is $\mathbf{F} = -\frac{1}{2}y\,\mathbf{i} + \frac{1}{2}x\,\mathbf{j}$, which gives:

The area of an ellipse. For the ellipse $\frac{x^2}{a^2} + \frac{y^2}{b^2} = 1$ with parameterization $\mathbf{r}(t) = (a\cos t, b\sin t)$:

The shoelace formula

For a polygon with vertices $(x_1, y_1), (x_2, y_2), \ldots, (x_k, y_k)$ listed counterclockwise, applying the same area formula to each line segment and simplifying gives:

This is the shoelace formula — named for the criss-cross pattern of the terms.

Regions with holes

Green’s theorem, as stated, requires $R$ to be simply connected. But we can extend it to regions with holes by carefully orienting the boundaries.

If $R$ is the region between an outer boundary $C_2$ and an inner boundary $C_1$ (like an annulus), then the “boundary” of $R$ consists of $C_2$ oriented counterclockwise and $C_1$ oriented clockwise. Green’s theorem then gives:

where both $C_1$ and $C_2$ are traversed counterclockwise (the minus sign on $C_1$ accounts for the clockwise orientation of the inner boundary).

This extension is essential for handling vector fields with singularities. For the gravity field $\mathbf{F} = -\frac{x\,\mathbf{i} + y\,\mathbf{j}}{(x^2+y^2)^{3/2}}$, which has $\operatorname{curl}\mathbf{F} = 0$ everywhere except the origin, we can show that $\oint_C \mathbf{F} \cdot \mathrm{d}\mathbf{r} = 0$ for any closed curve not passing through the origin — either by finding a potential function, or by deforming the curve to a circle where the integral can be computed directly.

Looking Ahead

Green’s theorem is the first of several powerful results that relate a boundary integral to an interior integral. In Part 4, we move from curves to surfaces: we learn to parameterize surfaces in $\mathbb{R}^3$, compute surface area via cross products, and set the stage for the surface integrals and Stokes’ theorem that complete the story.