Week 2 Discussion Notes

Table of Contents

Double Integrals

In 1D, the integral represented a (signed) area:

abf(x)dx=area under f(x) over [a,b]\int_a^b f\p{x} \,\diff{x} = \text{area under } f\p{x} \text{ over } \br{a, b}

This interpretation of double integrals is basically the same. Instead of area, we have volume:

Df(x,y)dA=volume under f(x,y) over D\iint_\mathscr{D} f\p{x, y} \,\diff{A} = \text{volume under } f\p{x, y} \text{ over } \mathscr{D}

You can already see that unlike in 1D, the region you integrate over can get very complicated. In 1D, you mostly integrated over intervals, but in 2D, D\mathscr{D} can be a rectangle, a disk, a star, etc. However, despite this, Fubini's theorem tells you that a double integral can be (in a nutshell) calculated as two 1D integrals:

Theorem (Fubini)
Df(x,y)dA=Df(x,y)dxdy=Df(x,y)dydx\begin{aligned} \iint_\mathscr{D} f\p{x, y} \,\diff{A} &= \iint_\mathscr{D} f\p{x, y} \,\diff{x} \,\diff{y} \\ &= \iint_\mathscr{D} f\p{x, y} \,\diff{y} \,\diff{x} \end{aligned}

In other words, Fubini's theorem tells you that you can calculate a double integral as an iterated integral, and that the order of the variables you integrate in doesn't matter.

Example 1.

Integrate f(x,y)=(x+y)2f\p{x, y} = \p{x + y}^2 over R=[0,2]×[1,1]\mathscr{R} = \br{0, 2} \times \br{-1, 1}.

Solution.

When given a rectangle, the first interval always corresponds to xx and the second interval always corresponds to yy. This gives us the following bounds:

Rf(x,y)dA=1102(x+y)2dxdy=11(x+y)33x=0x=2dy=11(y+2)3y33dy=(y+2)4y41211=203.\begin{aligned} \iint_\mathscr{R} f\p{x, y} \,\diff{A} &= \int_{-1}^1 \int_0^2 \p{x + y}^2 \,\diff{x} \,\diff{y} \\ &= \int_{-1}^1 \left. \frac{\p{x + y}^3}{3} \right\rvert_{x=0}^{x=2} \,\diff{y} \\ &= \int_{-1}^1 \frac{\p{y + 2}^3 - y^3}{3} \,\diff{y} \\ &= \left. \frac{\p{y + 2}^4 - y^4}{12} \right\rvert_{-1}^1 \\ &= \boxed{\frac{20}{3}}. \end{aligned}

To illustrate Fubini's theorem, I'll calculate the integral in the other order too:

Rf(x,y)dA=0211(x+y)2dydx=02(x+y)33y=1y=1dx=02(x+1)3(x1)33dx=(x+1)4(x1)41202=203.\begin{aligned} \iint_\mathscr{R} f\p{x, y} \,\diff{A} &= \int_0^2 \int_{-1}^1 \p{x + y}^2 \,\diff{y} \,\diff{x} \\ &= \int_0^2 \left. \frac{\p{x + y}^3}{3} \right\rvert_{y=-1}^{y=1} \,\diff{x} \\ &= \int_0^2 \frac{\p{x + 1}^3 - \p{x - 1}^3}{3} \,\diff{x} \\ &= \left. \frac{\p{x + 1}^4 - \p{x - 1}^4}{12} \right\rvert_0^2 \\ &= \frac{20}{3}. \end{aligned}
Example 2.

Integrate f(x,y)=exf\p{x, y} = e^x over the region D\mathscr{D} bounded by y=x+1y = x + 1, y=xy = x, x=0x = 0, and x=1x = 1.

Solution.

Unlike the previous example, D\mathscr{D} is slightly more complicated than a rectangle. For example, for each xx in the interval [0,1]\br{0, 1}, the interval that yy lives in changes:

x=0    0y1x=1    1y2.\begin{aligned} x = 0 &\implies 0 \leq y \leq 1 \\ x = 1 &\implies 1 \leq y \leq 2. \end{aligned}

This means that the bounds of yy are a function of xx:

For each xx in [0,1]\br{0, 1}, yy will range over [x,x+1]\br{x, x+1}, so the integral is

DexdA=01xx+1exdydx=01yexy=xy=x+1dx=01exdx=e1.\begin{aligned} \iint_\mathscr{D} e^x \,\diff{A} &= \int_0^1 \int_x^{x+1} e^x \,\diff{y} \,\diff{x} \\ &= \int_0^1 \Bigl. ye^x \Bigr\rvert_{y=x}^{y=x+1} \,\diff{x} \\ &= \int_0^1 e^x \,\diff{x} \\ &= \boxed{e - 1}. \end{aligned}