Changing Integration Order — Changing the Order of Integration - Calculus 3

SECTION 1

Historical Context & Motivation

The idea that the order of integration in a multiple integral can be reversed without changing the result has deep roots in the development of analysis. As mathematicians of the 18th and 19th centuries formalized the calculus of several variables, they confronted situations where evaluating a double integral in one order led to an antiderivative that was elementary, while the other order produced an integrand with no closed-form antiderivative at all. This practical difficulty motivated a rigorous investigation into iterated integrals and the conditions under which their order could be exchanged. The resulting theory not only streamlined computations but also deepened understanding of continuity, measurability, and convergence in higher dimensions.

1740s • Euler's Double Integrals

Leonhard Euler routinely swapped integration order in double integrals while solving problems in mechanics and series summation, often without formal justification, trusting the consistency of his results.

1813 • Cauchy's Rigorous Analysis

Augustin-Louis Cauchy began formalizing the conditions under which iterated integrals could be interchanged, identifying continuity of the integrand over a closed, bounded region as a sufficient condition.

1907 • Fubini's Theorem

Guido Fubini published his celebrated theorem giving precise conditions—integrability of the absolute value over the product region—under which the order of integration in a double (or higher) integral can be reversed, completing the rigorous foundation.

1930s • Tonelli's Extension

Leonida Tonelli extended Fubini's result to non-negative measurable functions, eliminating the need to verify integrability in advance and greatly expanding the practical reach of the interchange theorem.

The central question this lesson addresses is deceptively simple: given a double integral written as an iterated integral in one order, how do you rewrite it in the other order, and why would you want to? In many textbook and real-world problems, one order of integration yields an integrand whose antiderivative is unknown or extremely complicated, while the reversed order produces a straightforward computation. Mastering this technique requires understanding the geometry of the integration region and translating between descriptions of that region as a Type I (vertical slices) or Type II (horizontal slices) domain.

SECTION 2

Core Principles & Definitions

Before diving into computation, it is essential to establish the foundational ideas that govern when and how you may change the order of integration. These principles connect the algebraic expression of an iterated integral to the geometric shape of the region over which the integration is performed. A firm grasp of these ideas ensures that you can always set up the reversed integral correctly, regardless of the complexity of the boundary curves.

Fubini's Theorem

If f(x, y) is continuous on a closed, bounded region R, then the double integral over R equals both iterated integrals: ∫∫ f dA = ∫∫ f dy dx = ∫∫ f dx dy. The value is the same regardless of order.

Type I Region (Vertical Slices)

A region R is Type I if it can be described as a ≤ x ≤ b, g₁(x) ≤ y ≤ g₂(x). Here x is the outer variable and y is bounded by functions of x. You integrate with respect to y first.

Type II Region (Horizontal Slices)

A region R is Type II if it can be described as c ≤ y ≤ d, h₁(y) ≤ x ≤ h₂(y). Here y is the outer variable and x is bounded by functions of y. You integrate with respect to x first.

Sketch the Region First

The single most important step when changing integration order is to sketch the region R in the xy-plane. The limits of the original integral encode boundary curves; plotting them reveals how to re-describe R for the new order.

Solve Boundaries for the New Variable

After sketching, express each boundary curve in terms of the new outer variable. This algebraic step—solving y = g(x) for x = h(y) or vice versa—produces the new inner limits of integration.

✦

KEY TAKEAWAY

Think of a double integral as summing the values of a function over a 2D region. You can sweep that region with vertical strips (fixing x, scanning y) or with horizontal strips (fixing y, scanning x). Changing integration order is like choosing whether to mow a lawn in north–south rows or east–west rows—the total area mowed is identical, but one direction may be far easier to execute if the yard has an irregular shape.

SECTION 3

Visual Explanation — Type I vs. Type II Regions

The diagram below illustrates the same triangular region R described in two equivalent ways. On the left, the region is expressed as a Type I region with vertical slices: for each fixed x between 0 and 1, y ranges from 0 up to the line y = x. On the right, the same region is expressed as a Type II region with horizontal slices: for each fixed y between 0 and 1, x ranges from y to 1. Both descriptions cover exactly the same set of points in the plane, but the limits of integration look different.

Left: the triangular region bounded by y = 0, x = 1, and y = x, described with vertical slices (Type I). Right: the same triangle described with horizontal slices (Type II). The dashed lines show representative slices in each direction.

Notice how the vertical dashed lines on the left run from the bottom boundary y = 0 up to the line y = x, giving inner limits 0 ≤ y ≤ x. Meanwhile, the horizontal dashed lines on the right run from the left boundary x = y to the right boundary x = 1, giving inner limits y ≤ x ≤ 1. The outer limits in both cases run from 0 to 1. This geometric correspondence is the heart of changing the integration order: the boundary curves remain the same, but you describe them from a different perspective.

SECTION 4

Mathematical Framework

The formal justification for changing integration order rests on Fubini's Theorem. We state the version most commonly used in a multivariable calculus course and then show how the limits transform when you switch between Type I and Type II descriptions of the region.

FUBINI'S THEOREM

∫∫_R f(x, y) dA = ∫_a^b ∫_{g₁(x)}^{g₂(x)} f(x, y) dy dx = ∫_c^d ∫_{h₁(y)}^{h₂(y)} f(x, y) dx dy

f(x, y) is continuous on the closed, bounded region R. The first form uses Type I limits (outer in x, inner in y); the second uses Type II limits (outer in y, inner in x).

TYPE I → TYPE II CONVERSION

∫_a^b ∫_{g₁(x)}^{g₂(x)} f(x, y) dy dx → ∫_c^d ∫_{h₁(y)}^{h₂(y)} f(x, y) dx dy

Step 1: Identify boundary curves y = g₁(x) and y = g₂(x). Step 2: Solve each for x to get x = h₁(y) and x = h₂(y). Step 3: Determine the new outer limits c and d from the range of y over R.

GENERAL DOUBLE INTEGRAL OVER TYPE I

∫_a^b ∫_{g₁(x)}^{g₂(x)} f(x, y) dy dx

a, b are constants (the x-bounds of R). g₁(x) is the lower boundary curve, g₂(x) is the upper boundary curve. The inner integral is computed first with x treated as a constant.

⚠

Common Pitfall

When reversing the order, do not simply swap x and y in the limits. You must sketch the region, identify all boundary curves, and re-derive the limits from the geometry. Blindly swapping symbols produces incorrect integrals whenever the boundary curves are not symmetric in x and y.

In practice the procedure always follows the same logical sequence. First, read the original limits to identify the boundary curves of R. Second, sketch those curves in the xy-plane and shade the enclosed region. Third, determine the full range of the new outer variable across R. Fourth, for a generic value of that outer variable, determine where a horizontal (or vertical) slice enters and exits R—these intersection points, expressed as functions of the outer variable, become the new inner limits. Fifth, write the new iterated integral with the correct differential order.

SECTION 5

Detailed Breakdown — When and How to Split Regions

Not every region can be described with a single pair of inner limits after reversing the order. When the region's boundary changes character—say the upper boundary switches from one curve to another—you must split the region into sub-regions, each of which has a simple description, and write the reversed integral as a sum of integrals over those sub-regions. The diagram below illustrates a region where this splitting is necessary.

The region bounded by y = x², y = 4, and x = 4. In Type I (dy dx), the upper boundary changes from y = x² to y = 4 at x = 2, so the original integral may already be split. After reversing to dx dy, for each fixed y between 0 and 4 the horizontal slice runs from x = √y to x = 4, giving a single pair of limits and thus a simpler integral.

This example highlights an important phenomenon: sometimes reversing the order of integration actually reduces the number of sub-integrals you need. In the original dy dx form, you might need to split the x-interval at x = 2 because the upper boundary switches from y = x² to y = 4. But in the reversed dx dy form, every horizontal slice from y = 0 to y = 4 enters R at x = √y and exits at x = 4, requiring no split at all. Conversely, there are regions where the original order requires no split but the reversed order does. The key diagnostic is to check, for the new outer variable, whether the inner boundaries change their functional form partway through the range.

Step 1: Read the original limits and list every boundary curve.
Step 2: Sketch the region, marking intersection points of boundary curves.
Step 3: Sweep the region with slices in the new direction. If the left/right (or top/bottom) boundary changes form, note the y-value (or x-value) where the change occurs.
Step 4: Split the outer integral at each such transition value and write separate inner limits for each sub-region.

SECTION 6

Worked Example

Consider the integral ∫₀¹ ∫ₓ¹ sin(y²) dy dx. Attempting to evaluate this directly requires finding an antiderivative of sin(y²) with respect to y, which cannot be expressed in terms of elementary functions. By changing the order of integration, we transform the problem into one that is readily solvable.

Evaluate ∫₀¹ ∫ₓ¹ sin(y²) dy dx by reversing the integration order

1
Step 1 — Identify the region R from the given limits

The outer integral has x from 0 to 1. The inner integral has y from x to 1. This tells us R is the set of points (x, y) with 0 ≤ x ≤ 1 and x ≤ y ≤ 1. The boundary curves are: y = x (lower boundary of inner integral), y = 1 (upper boundary), x = 0 (left boundary from outer integral), and x = 1 (implicitly from y = 1 and y = x intersecting).

R: 0 ≤ x ≤ 1, x ≤ y ≤ 1 (a triangle with vertices (0,0), (0,1), (1,1))
2
Step 2 — Sketch the region and determine Type II limits

Plotting the triangle, we see that for each fixed y between 0 and 1, a horizontal slice enters R at x = 0 and exits at x = y (since y = x rearranged gives x = y as the right boundary). The outer variable y runs from 0 to 1.

Reversed limits: 0 ≤ y ≤ 1, 0 ≤ x ≤ y
3
Step 3 — Write the reversed integral

Applying Fubini's Theorem, we write the integral in dx dy order. The integrand sin(y²) does not change—only the limits and the differential order change.

∫₀¹ ∫₀ʸ sin(y²) dx dy
4
Step 4 — Evaluate the inner integral (with respect to x)

Since sin(y²) is independent of x, the inner integral is straightforward: ∫₀ʸ sin(y²) dx = sin(y²) · [x]₀ʸ = y sin(y²).

Inner integral = y sin(y²)
5
Step 5 — Evaluate the outer integral

Now compute ∫₀¹ y sin(y²) dy. Use the substitution u = y², so du = 2y dy, meaning y dy = du/2. When y = 0, u = 0; when y = 1, u = 1. The integral becomes (1/2) ∫₀¹ sin(u) du = (1/2)[−cos(u)]₀¹ = (1/2)(−cos 1 + cos 0) = (1/2)(1 − cos 1).

(1 − cos 1)/2 ≈ 0.2298

💡

Why This Works

The original order forced us to integrate sin(y²) with respect to y—an impossible task with elementary functions. After reversing, the inner variable became x, and since sin(y²) has no x-dependence, the inner integral reduced to a simple multiplication by y. The resulting outer integrand y sin(y²) is solvable by a basic u-substitution. This is the prototypical example of why changing integration order is such a powerful technique.

SECTION 7

When to Change Order — Strengths & Limitations

Changing the order of integration is not always necessary or advantageous. The decision depends on the integrand, the shape of the region, and whether one order produces a simpler inner antiderivative. The table below summarizes common scenarios and their outcomes.

Scenario	Original Order	Reversed Order
Inner integrand has no elementary antiderivative (e.g., sin(y²), $e^{x²}$ )	Stuck — cannot proceed analytically	Often the troublesome variable becomes the outer variable, simplifying the inner integral
Region is simpler as Type II	May require splitting into multiple sub-integrals	Single integral with clean limits
Both orders yield elementary antiderivatives	Solvable directly	Also solvable — choose whichever is simpler
Boundary curves are difficult to invert	May be the easier choice if limits are already simple	Inverting curves (e.g., solving x = arcsin(y)) can introduce complexity
Integrand factors as g(x)·h(y) over a rectangle	Separates into a product of single integrals	Same advantage — order is irrelevant over rectangles

Comparison of scenarios for choosing integration order

✦

KEY TAKEAWAY

Changing integration order is a strategic tool, not a mandatory step. Think of it like choosing a route on a road trip: sometimes the scenic route (one order) has an impassable bridge, and you must take the highway (the other order) to reach your destination. Before committing to a computation, scan both orders and pick the one where every antiderivative is within reach.

SECTION 8

Connection to Triple Integrals & Advanced Theory

The technique of changing integration order extends naturally to triple integrals and beyond. For a triple integral ∫∫∫ f(x, y, z) dV, there are six possible orders of integration (any permutation of dx, dy, dz). Fubini's theorem guarantees that all six produce the same result when f is continuous on a closed, bounded region, but the practical difficulty of computing the inner integrals can vary wildly. In many applications—computing volumes, moments of inertia, or centers of mass—choosing the optimal order can reduce a multi-page computation to a few lines.

Feature	Double Integrals (2D)	Triple Integrals (3D)
Number of possible orders	2 (dy dx or dx dy)	6 (all permutations of dx, dy, dz)
Region description	Type I or Type II in the plane	Projections onto coordinate planes define slicing strategy
Splitting complexity	Moderate — 2–3 sub-regions at most	Can be high — may need multiple sub-regions in two variables
Key theorem	Fubini's Theorem (2D)	Fubini's Theorem (general version for ℝⁿ)

Double vs. triple integral order-changing comparison

Beyond Cartesian coordinates, changing the order of integration connects to the broader idea of choosing the right coordinate system. Converting to polar, cylindrical, or spherical coordinates is conceptually the same strategy: you re-parameterize the region so that the limits and integrand become more tractable. The Jacobian determinant that appears in a coordinate transformation plays the same role as the implicit rearrangement of boundary curves when you reverse order in Cartesian coordinates. Mastering order reversal in double integrals therefore builds the geometric intuition you will rely on when tackling changes of variables, surface integrals, and the theorems of Green, Stokes, and Gauss later in multivariable calculus.

SECTION 9

Practice Problems

Question 1: CONCEPTUAL

Explain in your own words why the integral ∫₀¹ ∫ₓ¹

e^{y²}

dy dx cannot be evaluated directly by performing the inner integration first. What property of the integrand

e^{y²}

causes the difficulty, and how does reversing the order resolve it?

Show answer

The function

e^{y²}

has no antiderivative expressible in terms of elementary functions (polynomials, exponentials, trigonometric functions, and their compositions). This means you cannot find a closed-form expression for ∫

e^{y²}

dy. When you reverse the order, y becomes the outer variable and x becomes the inner variable. Since

e^{y²}

does not depend on x, the inner integral ∫₀ʸ

e^{y²}

dx simply yields y·

e^{y²}

, which is integrable with respect to y via the substitution u = y². The reversal moves the intractable integrand outside the inner integral, turning the inner computation into a trivial multiplication.

Question 2: BASIC CALCULATION

Reverse the order of integration for ∫₀² ∫₀ˣ f(x, y) dy dx and write the equivalent integral in dx dy form. (You need not evaluate; just provide the new limits.)

Show answer

The original limits describe the region R where 0 ≤ x ≤ 2 and 0 ≤ y ≤ x, which is a triangle with vertices at (0,0), (2,0), and (2,2). To reverse: for each fixed y between 0 and 2, a horizontal line enters R at x = y (from the boundary y = x, solved for x) and exits at x = 2 (the rightmost boundary). Thus the reversed integral is ∫₀² ∫ᵧ² f(x, y) dx dy. Note that the inner limits are from x = y to x = 2, and the outer limits run from y = 0 to y = 2.

Question 3: INTERMEDIATE

Evaluate ∫₀⁴ ∫_{√y}²

e^{x³}

dx dy by changing the order of integration.

Show answer

First identify R: 0 ≤ y ≤ 4 and √y ≤ x ≤ 2. The boundary x = √y is equivalent to y = x². At x = 2, y = 4. So R is bounded by y = x², y = 0 (implicit since √y ≥ 0 requires y ≥ 0), x = 2, and x = √y. Sketching, for each fixed x from 0 to 2, y runs from 0 to x². The reversed integral is ∫₀² ∫₀^{x²}

e^{x³}

dy dx. The inner integral gives

e^{x³}

· x². The outer integral is ∫₀² x²

e^{x³}

dx. Use u = x³, du = 3x² dx, so x² dx = du/3. Bounds: u = 0 to u = 8. This gives (1/3)∫₀⁸ eᵘ du = (1/3)(e⁸ − 1) ≈ 993.24.

Question 4: APPLIED

A lamina occupies the region bounded by y = ln(x), y = 0, and x = e, with density ρ(x, y) = 1. Set up and evaluate the integral for the area of the lamina by integrating in the order dx dy. (Hint: you will need to reverse from the natural dy dx form.)

Show answer

In dy dx form the region is 1 ≤ x ≤ e, 0 ≤ y ≤ ln(x), and the area is ∫₁ᵉ ∫₀^{ln x} dy dx. To convert to dx dy: the curve y = ln(x) gives x = eʸ. The maximum y is ln(e) = 1. For each y from 0 to 1, x ranges from eʸ (left boundary) to e (right boundary). The reversed integral is ∫₀¹ ∫_{eʸ}ᵉ dx dy. The inner integral gives e − eʸ. The outer integral is ∫₀¹ (e − eʸ) dy = [ey − eʸ]₀¹ = (e·1 − e¹) − (e·0 − e⁰) = (e − e) − (0 − 1) = 0 − (−1) = 1. The area is 1 square unit, which can be verified by the dy dx form: ∫₁ᵉ ln(x) dx = [x ln x − x]₁ᵉ = (e − e) − (0 − 1) = 1.

Question 5: CRITICAL THINKING

Consider ∫₀¹ ∫_{y}^{√y} f(x, y) dx dy. (a) Sketch the region and reverse the order of integration. Note that the reversed form requires splitting the integral. (b) Explain geometrically why one order requires a split and the other does not.

Show answer

(a) The region R is bounded by x = y (a line) and x = √y (a parabola, equivalently y = x²). These curves intersect at (0,0) and (1,1). For the given dx dy form, each horizontal slice has x from y to √y, with no split needed because both boundary functions are valid for all y in [0,1]. To reverse: project R onto the x-axis, giving 0 ≤ x ≤ 1. For a given x, the region is bounded below by y = x² (from x = √y solved for y) and above by y = x (from x = y). So the reversed integral is ∫₀¹ ∫_{x²}^{x} f(x, y) dy dx. In this case, no split is needed because the lower and upper curves do not change character over 0 ≤ x ≤ 1. (b) A split is necessary when, as the outer variable sweeps across its range, the inner boundary transitions from one curve to another. Here, both orders avoid a split because the region is bounded by exactly two curves that form a complete closed boundary. Splits arise in regions bounded by three or more curves when different pairs of curves serve as the inner boundaries over different sub-intervals of the outer variable.

SUMMARY

Lesson Summary

Summary

Changing the order of integration is a technique that rewrites a double integral ∫∫ f dA by swapping the roles of the inner and outer variables while preserving the value of the integral. The theoretical foundation is Fubini's Theorem, which guarantees equality of the two iterated integrals when f is continuous on a closed, bounded region R. The practical procedure requires you to sketch the region, identify all boundary curves, solve those curves for the new inner variable, and determine whether the region must be split into sub-regions where the boundary character changes.

The most common motivation is that one order produces an integrand with no elementary antiderivative (such as sin(y²) or $e^{x³}$ ), while the reversed order renders the inner integral trivial. Regions can be classified as Type I (vertical slices, dy dx) or Type II (horizontal slices, dx dy), and every double integral over a well-behaved region can be expressed in either form. This technique extends to triple integrals, where six orders are possible, and connects to the broader strategy of choosing optimal coordinates for integration.