Subsection 10.1.1 Motivation for Line Integrals
Recall that we have considered various types of integrals.
-
The definite integral
\(\int_a^b f(x) \,dx\) is an integral over an interval
\([a,b]\) in
\(\mathbb{R}\)
-
A double integral
\(\iint_D f(x,y) \,dA\) is an integral over a region
\(D\) in
\(\mathbb{R}^2\)
-
A triple integral
\(\iiint_D f(x,y,z) \,dV\) is an integral over a region
\(D\) in
\(\mathbb{R}^3\text{.}\)
-
The integral
\(\int_a^b \,dx\) (where
\(f(x) = 1\)) can be thought of as summing up small changes in
\(x\text{,}\) over the interval
\([a,b]\text{.}\)
-
More generally, the integral
\(\int_a^b f(x) \,dx\) can be thought of as summing up the products
\(f(x)\) and a small change in
\(x\text{,}\) over
\([a,b]\text{.}\)
Physically, this can be thought of as a quantity distributed along the
\(x\)-axis between
\(a\) and
\(b\text{,}\) with line density
\(f(x)\) at each point
\(x\text{,}\) and finding the total quantity by summing up, over
\([a,b]\text{,}\) the products,
\begin{equation*}
\text{quantity} = \text{density} \times \text{width} = f(x) \times \,dx
\end{equation*}
-
Double integrals
\(\iint_R f(x,y) \,dA\) can be thought of as summing up the products of function values
\(f(x,y)\) and small pieces of area
\(\,dA\text{,}\) over the region
\(R\text{.}\)
-
Triple integrals
\(\iiint_R f(x,y,z) \,dV\) can be thought of as summing up the products of function values
\(f(x,y,z)\) and small pieces of volume
\(\,dV\text{,}\) over the region
\(R\text{.}\)
In fact, to be consistent with the double and triple integral notation, we could write
\(\int_a^b f(x) \,dx\) as
\(\int_I f(x) \,dx\text{,}\) where
\(I = [a,b]\) is the “interval of integration”.
All integrals share the theme of being a sum, over a domain, of a product of a function with a small piece of that domain. Here, we will consider a
line integral, which is an integral over a curve.
Subsection 10.1.2 Line Integrals
Consider a mass distributed along a curve in the plane or 3-space, with mass density
\(f\) at each point along the curve. For example, a wire with varying mass density along its length. We want to find the total mass of the wire. To do this, we can find the sum, along the curve, of the product of the function value
\(f\) and a small segment of the curve.
Consider a curve
\(\mathcal{C}\) in the plane, with parametrization
\(\vec{r}(t)\text{,}\) \(a \leq t \leq b\text{,}\) and let
\(f(x,y)\) be the density at each point
\((x,y)\text{.}\) Partition the interval
\([a,b]\) into
\(a = t_0 \lt t_1 \lt \dots \lt t_n = b\text{.}\) This partition divides the curve into
\(n\) small arcs (or “subarcs”). On each arc, choose a sample point
\((x(t_k^*), y(t_k^*))\) and let
\(\Delta s_k\) denote the arc length of the
\(k\)th piece, where
\(k=1,2, \dots, n\text{.}\)
Then, the mass of the
\(k\)th piece of the wire is approximately,
\begin{align*}
\text{mass} \amp\approx \text{density} \times \text{length}\\
\amp= f(x(t_k^{*}), y(t_k^{*})) \cdot \Delta s_k
\end{align*}
Then, the total mass of the wire is approximated by the sum of all the pieces,
\begin{equation*}
S_n = \sum_{k=1}^n f(x(t_k^{*}), y(t_k^{*})) \, \Delta s_k
\end{equation*}
This is a Riemann sum of
\(f(x,y)\) with respect to arc length, so as
\(n \to \infty\) (and the maximum
\(\Delta s_k \to 0\)), the sum intuitively approaches the exact mass of the wire.
Definition 10.1.1. Line Integral.
The
line integral of
\(f\) along
\(\mathcal{C}\) is given by,
\begin{equation*}
\boxed{\int_{\mathcal{C}} f(x,y) \,ds = \lim_{n \to \infty} \sum_{k=1}^n f(x(t_k^{*}), y(t_k^{*})) \, \Delta s_k}
\end{equation*}
If
\(f\) is the linear mass density (mass per unit length), then,
\begin{equation*}
\text{total mass of the wire} = \int_{\mathcal{C}} f(x,y) \,ds
\end{equation*}
More generally, a line integral can represent any quantity distributed along a curve, with density
\(f\) at each point. For example,
-
Total charge along a wire with charge density
\(f\)
-
Total amount of some substance distributed along a curve with density
\(f\text{.}\)
If
\(\mathcal{C}\) is a closed curve, then the line integral is often written with a circle on the integral sign, as
\begin{equation*}
\oint_{\mathcal{C}} f(x,y,z) \,ds
\end{equation*}
This is just for emphasis, and does not change the meaning of the integral.
Line integrals are actually better described as “curve integrals”, because they are integrals over a curve, which is not necessarily a line.
Subsection 10.1.3 Geometric Interpretation of Line Integral
The line integral
\(\int_C f(x,y)\,ds\) also has a geometric interpretation: it is the area of the vertical, curtain-like surface between the curve
\(C\) in the
\(xy\)-plane and the curve
\(z = f(x,y)\) above it. Intuitively, imagine a vertical “wall” or “fence” standing on the curve
\(C\text{,}\) whose height at each point
\((x,y)\) is
\(f(x,y)\text{.}\) The line integral gives the total area of one side of this wall. It’s like a lateral surface area.
Each thin vertical strip of the curtain has approximate area
\(f(x(t_k^*), y(t_k^*)) \cdot \Delta s_k\) (height times width). Summing all the strips and taking the limit gives the line integral
\(\int_C f(x,y)\,ds\text{.}\)
This works as long as
\(f(x,y) \geq 0\text{,}\) otherwise the line integral can be thought of as a signed area (similar to the regular definite integral).
Here is a Desmos link, where you can visualize the line integral as a curtain, for any function and any curve:
Line Integral Visualization.
Subsection 10.1.4 Evaluating Line Integrals
The line integral of
\(f\) along
\(\mathcal{C}\) is not a typical integral, because the variable of integration is the arc length parameter
\(s\text{.}\) To evaluate it, we need to express
\(ds\) in terms of a parametrization
\(t\) of the curve.
Let
\(\mathcal{C}\) be a curve with parametrization
\(\vec{r}(t) = \vecii{x(t)}{y(t)}\text{,}\) \(a \leq t \leq b\text{.}\)
Recall that the arc length of the curve
\(\mathcal{C}\) is given by,
\begin{equation*}
\boxed{\int_{\mathcal{C}} \,ds = \int_a^b \sqrt{(x'(t))^2 + (y'(t))^2} \,dt = \int_a^b \abs{\vec{r}'(t)} \,dt}
\end{equation*}
In particular, the arc length element
\(ds\) is related to
\(dt\) by,
\begin{equation*}
ds = \abs{\vec{r}'(t)} \,dt = \sqrt{(x'(t))^2 + (y'(t))^2} \,dt
\end{equation*}
In a similar way, the line integral of
\(f\) over
\(\mathcal{C}\) can be expressed in terms of the parameter
\(t\text{.}\)
Theorem 10.1.5.
Let the density function
\(f(x,y)\) be continuous on
\(\mathcal{C}\text{.}\) Then, the line integral of
\(f\) over
\(\mathcal{C}\) is given by,
\begin{equation*}
\boxed{\int_{\mathcal{C}} f(x,y) \,ds = \int_a^b f(x(t), y(t)) \sqrt{(x'(t))^2 + (y'(t))^2} \,dt = \int_a^b f(\vec{r}(t)) \abs{\vec{r}'(t)} \,dt}
\end{equation*}
From this perspective, arc length is a special case of the more general line integral. If the density function is
\(f(x,y)=1\text{,}\) the line integral reduces to the arc length of the curve. All of this works as long as
\(\mathcal{C}\) is smooth, or at least piecewise smooth, and
\(f\) is continuous on
\(\mathcal{C}\text{.}\)
When setting up a line integral, often the most difficult part is finding a parametrization of the curve
\(\mathcal{C}\text{.}\)
Subsection 10.1.5 Line Integrals in Space
Line integrals can be extended to curves in 3-space. Again, consider a mass distributed along a curve
\(\mathcal{C}\) in 3-space (like a wire), with mass density
\(f(x,y,z)\) at each point
\((x,y,z)\text{.}\) Then, the total mass of the wire can be found by summing up, along the curve, the products of the density
\(f\) and small segments of the curve. Let
\(\mathcal{C}\) have parametrization
\(\vec{r}(t) = \veciii{x(t)}{y(t)}{z(t)}\text{,}\) \(a \leq t \leq b\text{.}\) Then,
\begin{equation*}
\int_{\mathcal{C}} f(x,y,z) \,ds = \lim_{n \to \infty} \sum_{k=1}^n f(x(t_k^{*}), y(t_k^{*}), z(t_k^{*})) \, \Delta s_k
\end{equation*}
Similarly, to evaluate this integral,
\begin{equation*}
\boxed{\int_{\mathcal{C}} f(x,y,z) \,ds = \int_a^b f(x(t), y(t), z(t)) \sqrt{(x'(t))^2 + (y'(t))^2 + (z'(t))^2} \,dt = \int_a^b f(\vec{r}(t)) \cdot \abs{\vec{r}'(t)} \,dt}
\end{equation*}
The integrals in the plane and in space can both be written in the compact form with vector notation,
\begin{equation*}
\int_a^b f(\vec{r}(t)) \cdot \abs{\vec{r}'(t)} \,dt
\end{equation*}