Double integrals over general regions — Type I and II: Question bank

Before we start, let us pin down every symbol these traps use so nothing is ambiguous.

Two pictures anchor almost every trap on this page — keep them in view.

Figure — Double integrals over general regions — Type I and II

True or false — justify

True or false: The value of

\iint_{D} f d A

changes if you switch from Type I to Type II.

False — provided

f

is integrable on

D

(e.g. continuous, or absolutely convergent), Fubini's Theorem guarantees the value is identical; only the difficulty of the two iterated integrals differs, not the number they produce.

True or false: Any region can be described as a single Type I region.

False — an annulus (a washer between two circles) forces a vertical strip to leave the region and re-enter it (see Figure 2), so no single pair of top/bottom curves

g_{1}, g_{2}

works and it must be split into several Type I pieces.

True or false: If a region is both Type I and Type II, you are free to pick whichever gives the easier integral.

True — such a region (e.g. a triangle or a disc) can be swept either way, so you choose the direction whose inner antiderivative is elementary, exactly like the

e^{y^{2}}

rescue in the "Spot the error" section below.

True or false: Setting

f \equiv 1

turns the double integral into the area of

D

True —

\iint_{D} 1 d A = \int_{a}^{b} (g_{2} (x) - g_{1} (x)) d x

, which is precisely the Area between two curves formula from single-variable calculus.

True or false: In a Type I integral the outer limits may depend on

y

False — the outer limits are the pure constants

a, b

; the outer integral must finish as a number, so it can contain no variable at all.

True or false: If

g_{1} (x) = g_{2} (x)

at the endpoints, the region is invalid.

False — the two boundary curves touching at endpoints just means the strip has zero height there (a pointed tip), which is perfectly normal for triangles and lens-shaped regions.

True or false: A region where the top boundary is a single formula must be Type I.

False — having one clean top curve

g_{2}

is convenient for Type I, but the same region might also be Type II if its left and right boundaries

h_{1}, h_{2}

are each single functions of

y

; being "nice for Type I" does not forbid Type II.

Spot the error

Someone writes

\int_{0}^{x} \int_{0}^{x} f d y d x

for the triangle under

y = x

. What is wrong?

The outer limit is

x

, which is illegal — the outer integral must evaluate to a number, so its limits (

a, b

) must be constants; only the inner limit (

g_{2} (x)

) may carry the outer variable.

For the region between

y = x^{2}

and

y = x

[0, 1]

, a student writes inner limits from

x

down to

x^{2}

. What breaks?

0 < x < 1

we have

x^{2} < x

, so

x

is the top curve

g_{2}

— integrating from the higher curve down to the lower one reverses orientation and negates the answer; always test a sample point like

x = \frac{1}{2}

A student swaps

d y d x

d x d y

but keeps the same numeric limits. Why is this wrong?

The value is order-independent, but the limits are not — horizontal and vertical strips hit different boundaries, so you must redraw

D

and re-read the limits, as stressed in Changing the order of integration.

Someone claims

\int_{0}^{1} \int_{x}^{1} e^{y^{2}} d y d x

is impossible to evaluate, full stop. Correct them.

It is impossible in that order because

e^{y^{2}}

has no elementary antiderivative in

y

; switching to Type II (

x

from

0

y

) makes it

\int_{0}^{1} y e^{y^{2}} d y = \frac{e - 1}{2}

, so it is very much doable.

A worked solution has an inner integral

\int_{g_{1} (x)}^{g_{2} (x)}

whose result still contains

y

. Diagnose it.

The inner integral in

y

must eliminate

y

entirely by substituting both limits — leftover

y

means the bounds were plugged in wrong (a forgotten evaluation) and the outer

x

-integral cannot proceed.

A student integrates

f (x, y) = x + 2 y

over the triangle but uses limits

y

from

0

1

for every

x

. Why is that the rectangle, not the triangle?

Constant

y

-limits

[0, 1]

describe the enclosing square, which wrongly includes the wedge above the line

y = x

; the triangle needs the sloped top limit

g_{2} (x) = x

so that extra wedge is excluded and never contributes.

For a region bounded left by

x = y^{2}

and right by

x = 2 - y

, someone integrates

y

on the outside with constant limits but forgets to find where the curves meet. What goes wrong?

Without the intersection points (

y = 1

and

y = - 2

) the outer

y

-limits

c, d

are unknown, so the region is undefined — you must solve

y^{2} = 2 - y

first to fix those bottom and top constants

c = - 2, d = 1

Why questions

Why must the inner limits, not the outer, be allowed to depend on the other variable?

The inner integral is done first while the outer variable is temporarily held fixed, so its limits may reference that frozen value; once it finishes, only the outer variable survives and must be integrated between fixed numbers

a, b

(or

c, d

Why does "extend by zero" let us pretend the region is a rectangle?

Defining the integrand to be

0

outside

D

means the area beyond the curved boundary adds no volume, so the rectangle integral equals the region integral — this is what lets Fubini's Theorem on the rectangle collapse into curve-dependent limits.

Why is Type I derived from Double integrals over rectangles rather than proved directly?

We only know how to iterate with constant limits over a rectangle; enclosing

D

R

and zero-extending is the bridge that converts the hard curved-region case into the solved rectangular one.

Why might switching the order of integration turn an impossible integral into an easy one?

The inner integrand can change from something with no elementary antiderivative into something that does, or gain a helpful factor (like the

y

y e^{y^{2}}

) that unlocks a substitution.

Why does choosing horizontal versus vertical strips sometimes cut the number of sub-regions from three to one?

If sweeping one direction makes a strip enter and leave

D

across a single pair of curves, but the other direction crosses a corner that swaps which curve bounds it, the smart direction avoids splitting

D

into pieces.

Why is the area formula a "sanity check" and not a coincidence?

Because

f \equiv 1

makes the inner integral just the strip height

g_{2} (x) - g_{1} (x)

; summing heights over

x

is literally the single-variable Area between two curves integral, so the two frameworks must agree.

Edge cases

What happens to a Type I integral if

g_{1} (x) = g_{2} (x)

over the whole interval

[a, b]

Every strip has zero height, so the region is a curve of zero area and the integral is

0

regardless of

f

— a degenerate but valid limiting case.

For an annulus

1 \leq x^{2} + y^{2} \leq 4

, why does a single Type I description fail, and what fixes it?

A vertical strip through the hole enters the outer circle, exits into the empty centre, then re-enters (Figure 2) — two disjoint

y

-ranges — so no single top/bottom pair works; splitting into pieces (or using Polar coordinates double integrals) is the clean fix.

If the two bounding curves cross more than twice, how must you set up the integral?

You split

D

at each crossing into sub-regions where the top/bottom (or left/right) ordering is consistent, then sum the separate integrals — a single limit pair cannot span a swap in which curve is on top.

What is the integral over a region that is a single point or a line segment?

Zero — a set of zero area contributes no volume no matter how large

f

is there, since the "thickness" of the strips shrinks to nothing.

When a region is unbounded (e.g.

x \geq 0

0 \leq y \leq e^{- x}

), can Type I still apply?

Yes, but the outer integral becomes improper (

\int_{0}^{\infty}

) and the answer is only trustworthy when the integral converges absolutely (

\iint_{D} ∣ f ∣ d A < \infty

); with that condition you keep the same strip logic and take a limit, and the boundary curve

g_{2} (x) = e^{- x}

still gives the inner limit.

f

changes sign over

D

, does the double integral still represent "volume"?

It represents signed volume — parts where

f < 0

subtract — so the number can be zero or negative even over a nonempty region; true geometric volume would need

\iint_{D} ∣ f ∣ d A

Recall One-line summary of every trap

Almost every mistake is one of three: a variable snuck into an outer limit ( $a, b, c, d$ must be constants), the top/bottom curves ( $g_{1}, g_{2}$ ) got swapped, or the region was not redrawn when the strip direction changed. Sketch first, sample-point second, integrate last.

← Back to parent topic