Skip to main content

Section 7.3 Integrals of a Scalar Function

We may consider a curve as a deformed interval. We know how to integrate a function over an interval. In this section we want to learn how to integrate functions defined on a deformed interval, that is, on a curve. Denote the curve by C, and assume that a function, f, with values in R is defined on C. If P is a point on the curve, denote by s=s(P) the distance measured along the curve from one end of the curve to P. Suppose that the total length of the curve is L. The function f is now only a function of s. Hence we may look at f as a function on the interval [0,L] and define
Cfds:=0Lf(s)ds.
Now suppose that C is a smooth curve, and that γ(t), t[a,b], is a regular parametrisation of C. As γ is a regular parametrisation, every point, PC, has a representation γ(t) for a unique t[a,b] (recall that a parametrisation is one-to-one by definition). According to Proposition 7.8 the length of the curve between γ(a) and γ(t)=P is given by
s(t)=atγ(τ)dτ.
By the fundamental theorem of calculus it follows that
s(t)=γ(t),
and thus by the substitution formula (5.5) we see that
0Lf(s)ds=abf(s(t))s(t)dt=abf(γ(t))γ(t)dt.
Hence we make the following definition.

Definition 7.10. Line integral of a scalar function.

Suppose that C is a smooth positively oriented curve, and that γ(t), t[a,b], a regular parametrisation of C. Then for every continuous scalar function on C we define
Cfds:=abf(γ(t))γ(t)dt.
The integral over a piecewise smooth curve is defined to be the sum over the integrals over the smooth parts of the curve.
One can show that the above definition of line integral is independent of the particular parametrisation γ choosen.

Example 7.11.

Compute C(x2y)ds if C is given by γ1(t)=t and γ2(t)=cosht with t[0,1].
Solution.
We first compute γ(t)=(γ1(t))2+(γ2(t))2. As γ1(t)=1 and γ2(t)=sinht we get
γ(t)=1+sinh2t=cosh2t=cosht.
Hence,
C(x2y)ds=01((γ1(t))2γ2(t))(γ1(t))2+(γ2(t))2)dt=01(t2cosht)coshtdt=01t2coshtdt01cosh2tdt
We now compute the last two integrals separately. For the first we integrate by parts twice:
01t2coshtdt=[t2sinht]01201tsinhtdt=sinh12([tcosht]0101coshtdt)=sinh12cosh1+2[sinht]01=sinh12cosh1+2sinh1=3sinh12cosh1.
Using a well known identity and integration of parts we have
cosh2tdt=sinhtcoshtsinh2tdt=sinhtcosht(cosh2t1)dt=sinhtcosht+tcosh2tdt.
Hence,
01cosh2tdt=12[t+sinhtcosht]01=1+sinh1cosh12,
and therefore
C(x2y)ds=3sinh12cosh11+sinh1cosh12.
This is the required result.

Insight 7.12. Physical Interpretation of Line Integrals.

We finish this section by giving some physical interpretations of the line integral. First consider the case N=2. We think of the curve, C, in the plane to be the boundary of a block of land. We want to put up a fence along that boundary. Its height at a point (x,y)R2 is supposed to be h(x,y), and may vary along different parts. (The fence to the bad neighbours is a bit higher than the fence to the good neighbours.) Before we go to the hardware store we want to know how much fencing we need to buy, so we want to compute the surface area of the fence. This is exactly what the line integral gives us, so the area is
Chds.
Another possible physical interpretation is the following. Suppose the curve C has the shape of a bent wire. The wire is made of different materials, or its thickness slightly varies. Fix a point, P, on C and let S be a piece of length Δs of C containing P. If Δm is the mass of S then Δm/Δs is the average mass density of S. Shrinking Δs to zero we get the mass density of the wire at P, that is, the mass per length. Denote that mass density at (x,y,z)C by ϱ(x,y,z). Then the total mass of the wire is given by
Cϱds.
In the next section we motivate and define line integrals of vector fields. The definition will be reduced to the one given here.