Vector Fields
Some quantities in the world are perfectly described by a single number at every point —
temperature in a room, say, or air pressure. But others need more: the wind at a spot isn't just
"20 kilometres an hour," it also blows in a particular direction; gravity at a spot doesn't just
pull "with a certain strength," it pulls toward the centre of the Earth. Whenever a
quantity carries both a size and a direction at every point of space, one number per point isn't
enough — you need a whole arrow at every point. That is a vector field,
and you have been reading them your whole life without knowing the name: the arrows on a weather
map, the streamlines painted onto a river-flow diagram, even the delicate curved lines that leap
into view when iron filings are sprinkled near a magnet.
A
gradient
attaches an arrow to each point of the plane. Loosen that idea — let the arrow be
anything, not necessarily a gradient — and you have a vector field:
a rule that assigns to every point (x, y) a vector
\mathbf{F}(x, y) = \big(P(x, y),\; Q(x, y)\big),
where P and Q are ordinary scalar
functions, the field's two components. Picture the plane carpeted with little arrows: the
wind over a map, water swirling in a basin, the pull of gravity around a planet. At a point
the arrow tells you the direction and strength of the flow or force living there.
Evaluating and sketching a field
Reading a field is just plugging in a point. Take the rotation field
\mathbf{F}(x, y) = (-y,\; x) — the swirl you get when stirring a
cup of coffee. We sketch it by evaluating at a few points.
Step 1 — evaluate at the point. Substitute the coordinates into each
component. At (1, 0):
\mathbf{F}(1, 0) = (-0,\; 1) = (0, 1).
Step 2 — read it as an arrow. Draw the vector (0, 1)
with its tail at the point (1, 0): it points straight up, length
1. The field is turning the point counterclockwise.
Step 3 — repeat at more points. At (0, 1) we get
(-1, 0) (pointing left); at (-1, 0) we get
(0, -1) (down); at (0, -1) we get
(1, 0) (right). Each arrow is perpendicular to the spoke from the
origin — the hallmark of pure rotation.
Step 4 — note the magnitude. The arrow's length is
\|\mathbf{F}\| = \sqrt{(-y)^2 + x^2} = \sqrt{x^2 + y^2}, the
distance from the origin: the field spins faster the farther out you go, like a rigid record
on a turntable.
A curve that is everywhere tangent to the arrows is a field line (a
streamline). For the rotation field the field lines are circles about the origin — follow the
arrows and you go round and round.
Now contrast it with the source field
\mathbf{F}(x,y) = (x, y), sketched the very same way — plug in points
first, look for the pattern second. At (1,0) we get
(1,0): pointing away from the origin, the same direction as the point
itself. At (0,1) we get (0,1); at
(2,0) we get (2,0) — twice as long as the
arrow at (1,0). Every arrow points straight outward from the
origin, growing longer the farther out you go — the pattern a splash makes bursting outward,
rather than a swirl going round. Two fields, evaluated with exactly the same recipe, sketch two
completely different pictures: one curls, the other bursts.
Gradient fields: when a field is a slope
A special, beautifully behaved family of fields are the gradient fields: a
field \mathbf{F} is a gradient field when it is the gradient of
some scalar potential f(x, y),
\mathbf{F} = \nabla f = \left( \frac{\partial f}{\partial x},\; \frac{\partial f}{\partial y} \right).
Then the arrows are the steepest-uphill directions of the surface
z = f(x, y) — the field "flows downhill" off a landscape. How do
we test whether a given \mathbf{F} = (P, Q) arises this way? Compare
mixed partials.
Step 1 — suppose it is a gradient. If
\mathbf{F} = \nabla f then
P = f_x and Q = f_y.
Step 2 — differentiate across. Using
partial derivatives,
differentiate P by y and
Q by x:
\frac{\partial P}{\partial y} = f_{xy}, \qquad \frac{\partial Q}{\partial x} = f_{yx}.
Step 3 — invoke equality of mixed partials. For a function with continuous
second derivatives, f_{xy} = f_{yx} (Clairaut's theorem). Therefore
\frac{\partial P}{\partial y} = \frac{\partial Q}{\partial x}.
Step 4 — apply the test. For the rotation field
\mathbf{F} = (-y, x) we have
P_y = -1 but Q_x = +1. Since
-1 \ne 1, the swirl is not a gradient field — no
landscape has it as its slope. By contrast the source field
\mathbf{F} = (x, y) has P_y = 0 = Q_x, and
indeed \mathbf{F} = \nabla\!\left(\tfrac12(x^2 + y^2)\right).
-
A vector field on a region of the plane is a map
\mathbf{F}(x, y) = (P, Q) assigning a vector to each point; its
magnitude is \|\mathbf{F}\| = \sqrt{P^2 + Q^2} and its
field lines are the curves everywhere tangent to it.
-
\mathbf{F} is a gradient field (or conservative)
when \mathbf{F} = \nabla f for some scalar potential
f; then the arrows point steepest-uphill on the surface
z = f.
-
A necessary condition is
\frac{\partial P}{\partial y} = \frac{\partial Q}{\partial x},
from equality of mixed partials. The rotation field (-y, x) fails
it; the source field (x, y) = \nabla\!\big(\tfrac12(x^2+y^2)\big)
passes. (When this condition is also sufficient is the subject of
conservative fields.)
Vector fields are the native language of physics.
-
Velocity fields. In a flowing fluid,
\mathbf{F}(x, y) is the velocity of the fluid at that point. The
field lines are the paths a speck of dust would trace — exactly the streamlines an
engineer photographs in a wind tunnel.
-
Electric fields. A point charge at the origin produces the field
\mathbf{E} = \dfrac{k\,(x, y)}{(x^2 + y^2)^{3/2}}, arrows
radiating outward (positive charge) and weakening with the square of the distance.
-
Gravitational fields. The pull of a planet at the origin is
\mathbf{g} = -\dfrac{GM\,(x, y)}{(x^2 + y^2)^{3/2}}, the same
shape pointing inward. Both the electric and gravitational fields are gradient
fields — they descend from a potential — which is exactly why "potential energy" is a
thing.
Three more worked examples
Reading a field well takes practice at three different jobs: plugging in a point, spotting a
gradient, and measuring speed. Here is one worked example of each.
Example A — evaluate and draw. Let
\mathbf{F}(x, y) = (x^2, xy).
At (2, 1): \mathbf{F}(2,1) = (2^2, 2\cdot 1) = (4, 2)
— an arrow pointing up and to the right, longer sideways than up.
At (-1, 3): \mathbf{F}(-1,3) = ((-1)^2, (-1)\cdot 3) = (1, -3)
— short to the right, long downward.
At (0, 2): \mathbf{F}(0,2) = (0^2, 0\cdot 2) = (0, 0)
— no arrow at all. The whole y-axis is still for this
field, since both components carry a factor of x.
Example B — spot the gradient. Let
f(x, y) = x^2y - \tfrac{1}{3}y^3. Its gradient field is built one
partial derivative at a time:
f_x = 2xy, \qquad f_y = x^2 - y^2,
so \nabla f = (2xy,\; x^2 - y^2) is the gradient field of
f — every arrow of this field points in the steepest-uphill direction
of the surface z = f(x,y).
Example C — magnitude as speed. A river's velocity field is measured as
\mathbf{F}(x, y) = (3, 2y) (metres per second): a steady rightward
current of 3, plus a sideways drift that grows with
y. At the point (4, -1), the speed of the
water is the magnitude of the arrow there:
\|\mathbf{F}(4,-1)\| = \sqrt{3^2 + (2\cdot(-1))^2} = \sqrt{9+4} = \sqrt{13} \approx 3.6 \text{ m/s}.
Notice the point's x-coordinate never entered the calculation — this
particular field only changes with y, so the current is exactly as fast
anywhere along a horizontal line.
-
A vector field is not a single vector.
\mathbf{F}(x,y) = (-y, x) is not "the vector
(-y,x)" sitting somewhere on its own — it is a whole
assignment, one arrow at every single point of the plane, a different vector at
every different (x,y). Asking "what is the vector field?" without
naming a point is like asking "what is the temperature?" without saying where — the field only
becomes a vector once you plug in a point.
-
Direction and magnitude both carry information. A longer arrow means the flow
is faster (or the force is stronger) there — it does not mean that point is "more important" or
"more correct." When comparing two points of a fluid-flow field, the one with the longer arrow is
simply where the water moves quickest, nothing more.
A weather map bristling with wind barbs is a vector field frozen at one instant: each barb's
direction shows which way the wind blows there, and the little flags on its tail count the speed —
more feathers, faster wind. Follow the arrows with your eye and you can trace out field lines just
like the ones you sketched by hand: they spiral into a low-pressure system (counterclockwise
in the Northern Hemisphere) and spiral out of a high-pressure system, the very same curling
and bursting patterns as the rotation and source fields above — just borrowed from the sky. Seen
from a weather satellite, a hurricane is nothing more than a rotation field with real wind roaring
through it.
Explore the carpet of arrows
Pick a field and watch its arrows fill the plane — each scaled and coloured by its magnitude
(short and cool near the origin, long and warm far out). The rotation field swirls, the
source field bursts outward, and the gradient field
\nabla\!\big(\tfrac12(x^2+y^2)\big) = (x, y) coincides with the
source — a slope made visible.
Fields fill space, not just the plane
A vector field pins an arrow to every point of space, not just a flat page. Here is a
three-dimensional field — a vortex that swirls around the vertical axis while
drifting upward, like a rising whirlwind. Drag to rotate it and look down the axis
to see the swirl, or from the side to see the upward climb.
See it explained