Calculus › Section P · Vector analysis › Topic P.1

Vector fields

Attach a vector to every point in space and you get a wind map, a flow, a force. Two measurements — divergence and curl — then extract what a field does at a point: how much it spreads and how much it swirls.

A vector at every point

A vector field attaches one vector to every point of a region — a function whose output is a vector. In the plane, \(\mathbf{F}(x, y) = \langle M(x, y), N(x, y)\rangle\); in space, \(\mathbf{F}(x, y, z) = \langle M, N, P\rangle\) with three component functions. Weather maps show wind fields; fluid dynamics lives in velocity fields; gravitation and electrostatics are force fields. To sketch a planar field by hand, evaluate \(\mathbf{F}\) at grid points and draw short arrows: \(\langle -y, x\rangle\) produces arrows that circulate counter-clockwise around the origin — a swirl this topic will measure with the curl. Fields are the setting for the whole chapter: line integrals weigh a field along a path, and the big theorems at the end compare a field's divergence and curl with its behaviour on boundary curves and surfaces.

Gradient fields

The most important fields are built from scalar ones: if \(f(x, y, z)\) is smooth, its gradient \(\nabla f\) is a vector field, called a gradient field, and \(f\) is its potential function. Gradient fields are also called conservative fields, because — as the path-independence topic proves — the work they do along a path depends only on the endpoints, so a particle moving under their influence conserves energy. As a quick example, \(f = xyz\) has \(\nabla f = \langle yz, xz, xy\rangle\). Two sanity checks precede that proof: the gradient topic showed \(\nabla f\) is perpendicular to level surfaces, so a gradient field always crosses its potential's level sets at right angles; and the curl below detects most impostors on sight.

Divergence: how much flows out

The divergence of \(\mathbf{F} = \langle M, N, P\rangle\) is the dot product of the del operator with the field:

\operatorname{div}\mathbf{F} = \nabla \cdot \mathbf{F} = \frac{\partial M}{\partial x} + \frac{\partial N}{\partial y} + \frac{\partial P}{\partial z}

— a scalar field. Picture \(\mathbf{F}\) as a fluid velocity: where \(\operatorname{div}\mathbf{F} > 0\) more fluid leaves a tiny box than enters (a source); where it is negative, more enters than leaves (a sink); zero divergence means the flow is locally incompressible. Example: \(\mathbf{F} = \langle x, y, z\rangle\) has divergence 3 everywhere — uniform expansion. For a gradient field, the divergence is the Laplacian \(\nabla^2 f = f_{xx} + f_{yy} + f_{zz}\), the operator behind heat flow: heat travels along \(-\nabla T\), and where \(\nabla^2 T \ne 0\) the temperature is locally changing.

Curl: how much it swirls

The curl of \(\mathbf{F} = \langle M, N, P\rangle\) is the cross product of del with the field:

\operatorname{curl}\mathbf{F} = \nabla \times \mathbf{F} = \Big\langle \frac{\partial P}{\partial y} - \frac{\partial N}{\partial z},\ \frac{\partial M}{\partial z} - \frac{\partial P}{\partial x},\ \frac{\partial N}{\partial x} - \frac{\partial M}{\partial y} \Big\rangle

— a vector field. Its direction gives the axis of local rotation and its magnitude the strength: a paddle wheel dropped into the flow spins fastest when its axle aligns with \(\operatorname{curl}\mathbf{F}\). The planar swirl \(\langle -y, x\rangle\) embeds in space as \(\langle -y, x, 0\rangle\), and its curl is \(\langle 0, 0, 2\rangle\): pure counter-clockwise spin about the z-axis, confirming what the arrow sketch showed. Curl, unlike divergence, is only defined in three dimensions — a planar field is curled by embedding it with third component 0 and reading the z-component.

Worked example: divergence and curl together

Compute the divergence and the curl of \(\mathbf{F} = \langle xy, yz, zx\rangle\). Divergence: \(\frac{\partial(xy)}{\partial x} + \frac{\partial(yz)}{\partial y} + \frac{\partial(zx)}{\partial z} = y + z + x\). Curl, component by component: the \(\mathbf{i}\)-piece is \(\frac{\partial(zx)}{\partial y} - \frac{\partial(yz)}{\partial z} = 0 - y\); the \(\mathbf{j}\)-piece is \(\frac{\partial(xy)}{\partial z} - \frac{\partial(zx)}{\partial x} = 0 - z\); the \(\mathbf{k}\)-piece is \(\frac{\partial(yz)}{\partial x} - \frac{\partial(xy)}{\partial y} = 0 - x\). So \(\operatorname{div}\mathbf{F} = x + y + z\) and \(\operatorname{curl}\mathbf{F} = \langle -y, -z, -x\rangle\). The nonzero curl answers a side question: \(\mathbf{F}\) is not a gradient field, because every gradient field has zero curl — the identity below is an impostor detector, and here it detected one.

Worked example: auditing a gradient field

Take the potential \(f = x^2 + y^2\) and audit its gradient field \(\nabla f = \langle 2x, 2y\rangle\). Divergence: \(\frac{\partial(2x)}{\partial x} + \frac{\partial(2y)}{\partial y} = 4\), which matches the Laplacian \(f_{xx} + f_{yy} = 2 + 2\) — as it must, since \(\operatorname{div}\nabla f = \nabla^2 f\) identically. Curl: the only live component is \(\mathbf{k}\), namely \(\frac{\partial(2y)}{\partial x} - \frac{\partial(2x)}{\partial y} = 0 - 0 = 0\), so \(\operatorname{curl}\nabla f = \mathbf{0}\). The geometry agrees: the field radiates straight outward through the circles \(x^2 + y^2 = c\) at right angles, never circulating along them — precisely the perpendicularity rule from the gradient topic.

Two identities worth memorising

Two identities follow from the equality of mixed partials, and both earn their memory space. First, \(\operatorname{curl}\nabla f = \mathbf{0}\) for every smooth \(f\): expand the curl of \(\langle f_x, f_y, f_z\rangle\) and the mixed partials cancel in pairs. Second, \(\operatorname{div}(\operatorname{curl}\mathbf{F}) = 0\) for every smooth \(\mathbf{F}\): a swirl has no net source. Each identity buys a free test — a field with nonzero curl cannot be a gradient field, and a field that is a curl has zero divergence. Beyond testing, these identities power the rest of the chapter: Green's theorem, Stokes's theorem and the divergence theorem all arrange curl and divergence against boundary curves and surfaces, and the identities guarantee the pieces fit.

Common mistakes to avoid

First, type confusion: divergence is a scalar, curl is a vector — writing \(\nabla \cdot \mathbf{F}\) with three components, or \(\nabla \times \mathbf{F}\) as one number, reveals the mix-up. Second, curl needs three dimensions: the cross product with del exists only in 3D, so planar fields are embedded as \(\langle M, N, 0\rangle\). Third, sign slips in the middle curl component: it is \(\frac{\partial M}{\partial z} - \frac{\partial P}{\partial x}\), the reverse of the intuitive reading — memorise the determinant layout rather than a verbal recipe. Fourth, do not read too much into zero: \(\operatorname{div}\mathbf{F} = 0\) does not mean \(\mathbf{F}\) is constant (the swirl \(\langle -y, x\rangle\) varies everywhere yet has zero divergence), and \(\operatorname{curl}\mathbf{F} = \mathbf{0}\) only certifies a potential when the region has no holes — the fine print the conservative-fields topic handles.

Key takeaways

A vector field \(\mathbf{F} = \langle M, N, P\rangle\) attaches a vector to every point; the gradient of any smooth function is a field, called conservative, with \(f\) its potential. Divergence \(\nabla \cdot \mathbf{F} = M_x + N_y + P_z\) is the scalar measure of net outflow — sources positive, sinks negative — and for a gradient field equals the Laplacian of the potential. Curl \(\nabla \times \mathbf{F} = \langle P_y - N_z, M_z - P_x, N_x - M_y\rangle\) is the vector measure of local rotation, defined only in three dimensions. The identities \(\operatorname{curl}\nabla f = \mathbf{0}\) and \(\operatorname{div}\operatorname{curl}\mathbf{F} = 0\) are free impostor tests and the machinery behind every theorem in the rest of this chapter.

Flashcards

Flashcard

Vector field (plane)

F(x, y) = ⟨M(x, y), N(x, y)⟩: one vector per point

Wind maps and flow maps are fields.

Test your understanding

Question 1 of 5

What is the divergence of F = ⟨x, y, z⟩?

Question 2 of 5

What is the curl of F = ⟨xy, yz, zx⟩?

Question 3 of 5

Which statement is true of every gradient field F = ∇f?

Question 4 of 5

What is the divergence of the gradient field ∇f when f = x² + y²?

Question 5 of 5

The curl of a vector field measures a point's...

Standards: Beyond AP

← Previous topicNext topic →