Notes on Line Integrals

Suppose F ~ = hF1 , F2 , F3 i is a vector field and C is an oriented curve given by a position

vector ~r. We can think of the vector field as “pushing” something along the curve. Examples are
a force field, in which case the total amount of “push” is called work, and a velocity field with a
closed curve, in which the total amount of “push” is called circulation. In these notes, we’ll use
the generic term push and the symbol P ; while not a recognized term in either mathematics or
physics, the term push has the advantage of indicating meaning, without the baggage of using
a physical term (work, flow, or circulation) to mean something different from what it means in
standard English.
In the simplest case, the vector field is constant and the curve is a straight line in the x
direction. The total push is simply the product of the horizontal component F1 of the vector
field and the change in position ∆x: P = F1 ∆x. For easy generalization, we can write this as
a dot product:
~ · ∆~r.
P = F1 ∆x = hF1 , F2 , F3 i · h∆x, 0, 0i = F
This dot product formula holds for any straight line, regardless of the direction, but only when
the vector field is the same everywhere on the curve.
Now suppose the vector field is different at different points along the curve and the curve is
not necessarily a straight line. To calculate the total amount of push, it is necessary to calculate
the amount of push for a little bit of the curve and then add the totals for all the little bits. For
~
an arbitrarily small bit of curve, we can take the vector field to be the constant value F(x, y, z)
and the change in position to be d~r. Since we are dealing with a vector force and displacement,
it is only the component of the field that is directed along the curve that counts toward the
push; hence we need the dot product rather than some other type of multiplication. Hence, the
little bit of push at this point of the curve is
~ · d~r.
dP = F

Summing the infinitesimal contributions at each point, the total push is given by
Z
P = ~ · d~r.
F
C

Evaluating Line Integrals

The evaluation of definite integrals has a common mathematical structure that is independent
of the geometry. The idea is to set up an iterated integral using as many integration variables
as the dimension of the region; that is, we need one parameter for a curve, two for a surface,
and three for a solid region. Integrals along lines that can be described using only one spatial
coordinate are trivial in that they almost immediately reduce to a simple one-variable integral.
Other curves must be parameterized, and the parameter used for that parameterization serves as
the variable for the integration. Hence, we have the following general plan for direct calculation
of any definite integral:
1. Choose the integration variable(s). An n-dimensional region requires n variables. Any
spatial variable other than an integration variable needs to be written in terms of inte-
gration variables.

2. Write the differential in terms of the integration variable(s).

1
 3. Write the integrand in terms of the integration variables.

4. Identify the ranges of the integration variable(s) using a nested set of inequalities, and
use these inequalities to set the integral up as an iterated integral.

5. Compute the integral from the inside out.

In the case of line integrals, there are also indirect methods that can sometimes be used instead
of the direct method.

Method 1 — Direct evaluation by parameterization

Consider first the special case where the curve is parallel to a coordinate axis. As an example,
suppose the curve is the portion of the line x = 2 that runs from y = 0 to y = 4 and suppose
~ = hx, 2x + yi. In this case, any little bit of the curve runs only in the
the vector field is F
y direction, so we simply have d~r = ~j dy. Similarly, every point on the curve has x = 2, so
~ = h2, 4 + yi. Hence,
F
Z Z 4 Z 4
P = ~ · d~r =
F ~ · ~j) dy =
(F (4 + y) dy = · · · = 24.
C y=0 0

In the more general case, we can compute d~r for a curve given as a ~r(t) = hx(t), y(t), z(t)i
using a simple multivariable substitution formula:
d~r
d~r = dt.
dt
With this formula, line integrals become ordinary one-variable integrals in the parameter t:
Z b !
~ · d~r
Z
~ · d~r =
F F dt. (1)
C a dt

As an example, suppose F ~ = hx, 2x + yi as before, but C is the top half of the unit circle,
traveled clockwise. We can use the curve −C, parameterized by

x = cos t, y = sin t, 0 ≤ t ≤ π,

instead of the original curve C. Using this parameterization, we have

* +
~ = hcos t, 2 cos t + sin ti, d~r dx dy
F = , = h− sin t, cos ti,
dt dt dt
so
~ · d~r = (cos t)(− sin t) + (2 cos t + sin t)(cos t) = 2 cos2 t = 1 + cos 2t,
F
dt
and
Z π ! Z π
~ · d~r
Z Z
~ · d~r = −
F ~ · d~r = −
F F dt = − (1 + cos 2t) dt = · · · = − (π + 0) = −π.
C −C 0 dt 0

Note that the direct method makes sense in either two or three dimensions.

2
Method 2 — The fundamental theorem for line integrals

R
~ · d~r can sometimes be evaluated with a potential function.1 The following
The line integral C F
conditions must hold:
~ = ∇f
1. The vector field must be a gradient field: F ~ .

2. The curve C must lie in a domain in which the vector field is continuous.

3. The curve C must be piecewise smooth.

When these conditions are met, the fundamental theorem of line integrals says
Z Z
~ · d~r =
F ~ · d~r = f (B) − f (A),
∇f (2)
C C

where A and B are the initial and terminal points, respectively, of the curve C. The funda-
mental theorem for line integrals is analogous to the fundamental theorem of calculus, but it
is applicable in a much narrower range of problems. Whereas any continuous function of one
variable is the derivative of some function, an arbitrarily chosen continuous function of several
variables is almost certainly not the gradient of some function.

Method 2.1 – Reparameterization for line integrals

~ is known to be a gradient field, then the line integral from A to B of F

If F ~ is independent of
path. In addition to allowing for computation using the fundamental theorem for line integrals,
this property allows the integral to be computed by the direct method applied to any curve
from A to B. This may be convenient in cases where it is difficult to compute a potential
function.

Method 3 — Green’s theorem

Green’s theorem can be used for line integrals in the plane, provided the curve and vector field
satisfy several conditions:

1. The curve must be the boundary of a plane region R, traveled counterclockwise.

~ = hF1 , F2 i must have continuous first partial deriva-
2. The components of the vector field F
tives in a connected open region that contains the curve.

If these conditions are met, then Green’s theorem asserts the equivalence of integrals of F1 and
F2 over the curve with integrals of derivatives of F1 and F2 over the region inside the curve:
!
I ZZ
∂F2 ∂F1
F1 dx + F2 dy = − dA. (3)
C R ∂x ∂y
1
See the appendix.

3
Green’s theorem can also be written in vector form as
!
I
~ · d~r =
I ZZ
∂F2 ∂F1 ZZ  
~ · ~k dA.
~ ×F
F F1 dx + F2 dy = − dA = ∇ (4)
C C R ∂x ∂y R

Before using Green’s theorem, it is important to check the necessary conditions for the curve
and vector field, as they are not always met.

Method 4 — Stokes’ theorem

Stokes’ Theorem is the three-dimensional generalization of Green’s Theorem. The requirements
are a little more general:

1. The curve must be the boundary of a surface S, traveled counterclockwise as viewed from
above, with “above” defined by the particular normal vector chosen to represent S.

2. The components of the vector field F ~ = hF1 , F2 , F3 i must have continuous first partial
derivatives in a connected open region that contains the curve.

If these conditions are met, then Stokes’s theorem asserts the equivalence of the line integral
around the closed curve C with an integral taken over the enclosed surface S:2
I ZZ
~ · d~r =
F ~ dA.
~ ×F
∇ ~ (5)
C S

2
Surface integrals are discussed later in the course. Indeed, even the notation for the integral on the right
side will likely be unfamiliar at this point. Stokes’ Theorem is mentioned here for completeness only.

4
Appendix – Finding Potential Functions
Suppose we want to use the fundamental theorem for line integrals (method 2) to evaluate
Z
~ · d~r,
F ~ = h2x, 1 + 2yz 3 , 3y 2 z 2 i
F (6)
C

for some complicated curve C. This will only work if there is a potential function f for which
~
~ = F.
∇f

If there is such a function (if not, we’ll be able to tell), it has to satisfy three requirements:

fx = 2x, fy = 1 + 2yz 3 , fz = 3x2 z 2 .

The first requirement seems to mean f = x2 , but actually the most general possibility is that
f is x2 plus additional terms that might include y and z, but not x. In other words,

f = x2 + f1 (y, z).

The same analysis can be applied to the other requirements, giving us

f = y + y 2 z 3 + f2 (x, z)

and
f = y 2 z 3 + f3 (x, y).
We now need to determine if there is a function f that satisfies all three of these formulas. Such
a function would have to include the terms x2 , y and y 2 z 3 . Careful checking shows that

f = x2 + y + y 2 z 3

works with f1 (y, z) = y + y 2 z 3 , f2 (x, z) = x2 , f3 (x, y) = x2 + y.

Note that this procedure fails if there is no potential function. For example, suppose
~ = h2xy + y, x2 i.
F

~ we get
~ = F,
From ∇f
fx = 2xy + y, fy = x2
and then
f = x2 y + xy + f1 (y), f = x2 y + f2 (x).
These results cannot be reconciled because the term xy in the first formula cannot be f2 (x) in
the second.

03/27/2019 Glenn Ledder