[Intermediate → Advanced] — Accumulating over a region rather than an interval, and why changing the order of integration is sometimes the whole solution.

Before you start

What you will be able to do after this lesson:

  1. Evaluate a double integral as an iterated integral.
  2. Set up the limits for a non-rectangular region.
  3. Change the order of integration and say when it is necessary.
  4. Convert to polar coordinates and explain the Jacobian factor.

1. From intervals to regions

A single integral accumulates along a line: adds up strips of width .

A double integral accumulates over an area:

One slice is now a tiny rectangle of area , and is its height — so the whole thing is a volume under a surface, just as the single integral was an area under a curve.

2. Iterated integration

You never evaluate a double integral directly. You do two ordinary integrals, one inside the other.

The inner integral treats as a constant — exactly like a partial derivative in reverse. The result is a function of , which the outer integral then handles.

Worked, over the rectangle , :

Over a rectangle with fixed limits, the order does not matter — Fubini’s theorem. Over anything else, it can matter a great deal.

3. Non-rectangular regions

When the region is bounded by curves, the inner limits become functions.

For the region under , between and :

Read it inside out: for each fixed , runs from up to ; then runs across .

The rule that prevents most errors: the outer limits must be constants. If your outer integral still has a variable in its limits, the setup is wrong — there would be nothing left to substitute it with.

NOTE

Always sketch the region. Nearly every mistake in multiple integration is a limits mistake, and limits are geometry — they are far easier to see than to derive algebraically.

Sketch it, decide whether to slice vertically or horizontally, and read the limits off the picture.

4. Changing the order

Sometimes reversing the order turns an impossible integral into an easy one.

The inner integral cannot be done — has no elementary antiderivative, as noted earlier.

Sketch the region: from 0 to 1, and from to 1. That is the triangle above the line inside the unit square.

Describe the same triangle the other way: from 0 to 1, and for each , runs from 0 to :

Now it is a substitution. With , :

The integral was always solvable — the first ordering just hid it. That extra factor of , which appeared from integrating over , is exactly what the substitution needed.

5. Polar coordinates

Circular regions are painful in and — the limits involve and similar.

In polar coordinates , , and:

That extra is not optional, and it is not a convention. A small polar rectangle spanning and is not a rectangle of area — it is a curved patch whose arc length grows with radius. Its area is , because the arc subtended by at radius has length .

        the patch at large r is WIDER
        than the patch at small r
        for the same dθ

           ,--.___
         ,'   |   `--.
        /  dr |         <- arc length = r dθ
       |______|_____
        small r   large r

Forgetting the is the single most common error in polar integration, and it produces answers that are wrong by a factor that varies with the region.

Example — area of a circle of radius :

And the Gaussian integral, which cannot be done in one dimension, becomes possible in two:

so the original integral is . The from the Jacobian is precisely what makes the substitution work — this is the classic demonstration that the factor is real.

6. Practice — problems

  1. Evaluate .
  2. Evaluate . Sketch the region first.
  3. Set up (do not evaluate) the double integral for the area between and , both orders.
  4. Reverse the order: , then evaluate.
  5. Use polar coordinates to find over the disc .
  6. Explain why rather than , in one sentence about arc length.
Answers — open only after an attempt
  1. Inner: . Outer: .
  2. Inner: . Outer: . The region is the triangle below over .
  3. The curves meet at and , with between. : . : — note the limits swap roles because the same region is now sliced horizontally.
  4. The region is , — the triangle above . Reversed: .
  5. , so .
  6. Because the patch swept by a fixed angle has arc length , which grows with radius — so a patch far from the origin covers more area than one near it, and the accounts for that stretching.

Before moving on

  • Evaluate a double integral as two nested ordinary integrals.
  • Set up limits for a non-rectangular region, keeping the outer limits constant.
  • Reverse the order of integration and explain when it helps.
  • Explain the in geometrically.

Recap: A double integral accumulates over a region, and is evaluated as two nested single integrals with the inner one treating the outer variable as constant. Over a rectangle the order is free; over a curved region the inner limits become functions and the outer limits must remain constants. Reversing the order can turn an impossible inner integral into a routine substitution. Polar coordinates simplify circular regions, and the Jacobian factor is required because a patch’s area grows with its distance from the origin.

Next: Optimisation — finding extrema of a function of several variables, with and without constraints.

↑ Back to Mathematics