Iterated Integrals & Fubini's Theorem — Iterated Integrals and Fubini's Theorem - Calculus 3

SECTION 1

Historical Context & Motivation

The challenge of computing areas, volumes, and accumulated quantities over two-dimensional regions has occupied mathematicians for centuries. While single-variable calculus provides a clean framework for integrating functions of one variable, extending integration to functions of two or more variables introduces fundamental questions about how to organize the computation. The key insight — that a double integral over a planar region can be decomposed into two successive single-variable integrals — evolved gradually through the work of several mathematicians spanning from the 17th to the early 20th century.

1693 • Newton & Leibniz: Foundations of Calculus

The independent development of single-variable calculus by Newton and Leibniz laid the groundwork for all future integration theory. Leibniz's notation for integrals (∫) proved especially influential for generalizing to higher dimensions.

1740s • Euler's Multi-Variable Explorations

Leonhard Euler routinely computed double integrals by performing successive integrations, treating each variable in turn. He used this technique to solve problems in mechanics and probability, though he did not formally justify the interchange of integration order.

1813 • Cauchy's Rigorous Integral

Augustin-Louis Cauchy provided the first rigorous definition of the definite integral as a limit of sums for continuous functions. His framework made it possible to state precise conditions under which double integrals exist and can be evaluated iteratively.

1907 • Fubini's Theorem Published

Guido Fubini, an Italian mathematician, published the definitive theorem establishing conditions under which a double integral equals an iterated integral and the order of integration may be exchanged. His result unified and extended earlier partial results by Lebesgue and Tonelli.

1909 • Tonelli's Complement

Leonida Tonelli extended Fubini's result to non-negative measurable functions, removing the requirement of integrability from the hypothesis. Together, the Fubini–Tonelli theorem became a cornerstone of modern analysis and measure theory.

The central question Fubini's Theorem addresses is deceptively simple: if we want to integrate a function f(x, y) over a rectangle or more general region in ℝ², can we always reduce the problem to performing one integral 'inside' the other? And if so, does the order in which we integrate — first with respect to x then y, or vice versa — affect the result? These questions are not merely computational conveniences; they lie at the heart of how we define and compute multiple integrals across all of applied mathematics, physics, and engineering.

SECTION 2

Core Principles & Definitions

Before stating Fubini's Theorem, we need to establish the foundational concepts that make it meaningful. The transition from single-variable to multi-variable integration requires careful attention to the domain of integration, the meaning of an iterated integral, and the distinction between a double integral defined as a limit of Riemann sums and the sequential evaluation process that Fubini's Theorem justifies.

Double Integral as a Limit

The double integral ∬_R f(x, y) dA is defined as the limit of Riemann sums over partitions of the region R. It represents the signed volume between the surface z = f(x, y) and the xy-plane.

Iterated Integral

An iterated integral evaluates a multi-variable integral by performing one single-variable integration at a time. For a rectangle [a, b] × [c, d], this yields ∫ₐᵇ (∫ᶜᵈ f(x, y) dy) dx, where the inner integral is computed first while holding the outer variable constant.

Rectangular vs. General Regions

On a rectangle R = [a, b] × [c, d], the limits of integration are constants. On Type I regions (bounded by y = g₁(x) and y = g₂(x)) or Type II regions (bounded by x = h₁(y) and x = h₂(y)), the inner limits become functions of the outer variable.

Fubini's Theorem (Statement)

If f is continuous on R = [a, b] × [c, d], then the double integral equals either iterated integral: ∬ₐ f dA = ∫ₐᵇ ∫ᶜᵈ f(x, y) dy dx = ∫ᶜᵈ ∫ₐᵇ f(x, y) dx dy. Continuity is sufficient but not necessary; integrability over R also suffices.

Order of Integration

Fubini's Theorem guarantees that when its hypotheses hold, the order of integration may be switched without changing the result. Strategically choosing the order can dramatically simplify computation — one order may yield an elementary antiderivative while the other does not.

✦

KEY TAKEAWAY

Think of a double integral like computing the total weight of a variable-density sheet of metal. You could slice the sheet into thin vertical strips, weigh each strip by integrating along its height (inner integral), and then sum the strip weights across the width (outer integral). Alternatively, you could slice horizontally. Fubini's Theorem guarantees that both slicing strategies yield the same total weight — provided the density function is well-behaved (continuous or integrable) over the region.

SECTION 3

Visual Explanation

The geometric intuition behind iterated integrals is best understood by visualizing how a two-dimensional region is swept out by one-dimensional slices. The following diagram illustrates the two possible orders of integration over a rectangular region, showing how the inner integral computes the 'slice area' and the outer integral accumulates those slices into a total volume.

Left: vertical slices where the inner integral runs over y for each fixed x, then the outer integral accumulates across x. Right: horizontal slices where the inner integral runs over x for each fixed y, then the outer integral accumulates along y. Fubini's Theorem guarantees equality of both approaches when f is continuous (or integrable) on the rectangle.

In the left diagram, each vertical strip at position x has a width dx and a 'height' determined by the inner integral ∫ᶜᵈ f(x, y) dy. This inner integral is a function of x alone, since y has been integrated out. The outer integral then sweeps this function across [a, b]. The right diagram reverses the roles: horizontal strips of height dy have widths determined by ∫ₐᵇ f(x, y) dx, and the outer integral sweeps from c to d. The visual symmetry between the two approaches reflects the algebraic content of Fubini's Theorem — neither slicing direction is privileged, and both produce the same numerical result.

SECTION 4

Mathematical Framework

We now present the formal statements and key formulas that govern iterated integrals and the application of Fubini's Theorem. The precision of these formulations matters: understanding when and why the theorem applies protects against subtle errors that arise in more advanced settings.

DOUBLE INTEGRAL OVER A RECTANGLE

∬_R f(x, y) dA = lim_{m,n→∞} Σᵢ₌₁ᵐ Σⱼ₌₁ⁿ f(xᵢ*, yⱼ*) ΔA

R = [a, b] × [c, d] is a closed rectangle; (xᵢ*, yⱼ*) is a sample point in the (i, j)-th subrectangle; ΔA = Δx · Δy is the area of each subrectangle. The double sum converges to the double integral when f is integrable on R.

FUBINI'S THEOREM (RECTANGULAR DOMAIN)

∬_R f(x, y) dA = ∫ₐᵇ [ ∫ᶜᵈ f(x, y) dy ] dx = ∫ᶜᵈ [ ∫ₐᵇ f(x, y) dx ] dy

If f is continuous on R = [a, b] × [c, d], then both iterated integrals exist and equal the double integral. More generally, it suffices that f be integrable on R and that the inner integrals exist for almost every value of the outer variable.

TYPE I REGION (VERTICAL SLICES)

∬_D f(x, y) dA = ∫ₐᵇ ∫_{g₁(x)}^{g₂(x)} f(x, y) dy dx

D = {(x, y) : a ≤ x ≤ b, g₁(x) ≤ y ≤ g₂(x)} where g₁ and g₂ are continuous on [a, b]. The inner limits are now functions of x.

TYPE II REGION (HORIZONTAL SLICES)

∬_D f(x, y) dA = ∫ᶜᵈ ∫_{h₁(y)}^{h₂(y)} f(x, y) dx dy

D = {(x, y) : c ≤ y ≤ d, h₁(y) ≤ x ≤ h₂(y)} where h₁ and h₂ are continuous on [c, d]. The inner limits are now functions of y.

⚠

When Does Fubini Fail?

Fubini's Theorem requires that either f be integrable over R (in the Riemann or Lebesgue sense), or — via Tonelli's extension — that f be non-negative and measurable. Classic counterexamples involve functions that are not absolutely integrable, where the two iterated integrals can yield different values. For functions encountered in standard Calculus 3, continuity on a closed bounded region is the simplest sufficient condition, and it is satisfied in virtually all applied problems.

SECTION 5

Detailed Breakdown — Region Types & Changing Order

In practice, the most important skill associated with Fubini's Theorem is correctly setting up iterated integrals over non-rectangular regions and, when advantageous, reversing the order of integration. This requires classifying the domain as a Type I or Type II region (or sometimes decomposing it into sub-regions of each type), expressing the boundary curves appropriately, and determining which order leads to a tractable inner antiderivative. The diagram below illustrates a region that can be described in both ways, highlighting how the limits of integration change when the order is reversed.

The same region D is expressed using vertical slices (Type I, left) with outer variable x and inner limits y = g₁(x) to y = g₂(x), and using horizontal slices (Type II, right) with outer variable y and inner limits x = h₁(y) to x = h₂(y). Reversing the order of integration requires solving the boundary curves for the opposite variable.

Steps for Reversing the Order of Integration

Sketch the region: Draw the curves given by the original limits of integration and shade the enclosed region D.
Identify the boundary curves: Determine which curves form the lower/upper boundaries (for dy dx) or left/right boundaries (for dx dy).
Solve for the opposite variable: If the original inner variable is y, solve each boundary equation for x as a function of y, and vice versa.
Determine new outer limits: Find the range of the new outer variable by identifying where the region starts and ends in that coordinate direction.
Write the new iterated integral: Assemble the new limits into the reversed iterated integral and verify by checking a sample point.

SECTION 6

Worked Example

Let us evaluate the double integral ∬_R (x²y + 3xy²) dA where R = [0, 2] × [1, 3], using both orders of integration to verify Fubini's Theorem.

Evaluate ∬_R (x²y + 3xy²) dA over R = [0, 2] × [1, 3]

1
Step 1 — Set Up the Iterated Integral (dy dx order)

Since R is a rectangle, Fubini's Theorem applies directly (the integrand is a polynomial, hence continuous). We write the iterated integral with the inner variable y and outer variable x: ∫₀² ∫₁³ (x²y + 3xy²) dy dx
2
Step 2 — Evaluate the Inner Integral (integrate with respect to y)

Treating x as a constant, we integrate term by term with respect to y: ∫₁³ (x²y + 3xy²) dy = [x²y²/2 + xy³]₁³. Substituting the limits: at y = 3 we get x²(9)/2 + x(27) = 9x²/2 + 27x; at y = 1 we get x²(1)/2 + x(1) = x²/2 + x. The difference is (9x²/2 + 27x) − (x²/2 + x) = 4x² + 26x.

Inner integral = 4x² + 26x
3
Step 3 — Evaluate the Outer Integral (integrate with respect to x)

Now integrate the result from Step 2 over x ∈ [0, 2]: ∫₀² (4x² + 26x) dx = [4x³/3 + 13x²]₀². Evaluating at x = 2: 4(8)/3 + 13(4) = 32/3 + 52 = 32/3 + 156/3 = 188/3. At x = 0 both terms vanish.

∬_R (x²y + 3xy²) dA = 188/3 ≈ 62.67
4
Step 4 — Verify by Reversing the Order (dx dy)

We write ∫₁³ ∫₀² (x²y + 3xy²) dx dy. Inner integral: ∫₀² (x²y + 3xy²) dx = [x³y/3 + 3x²y²/2]₀² = 8y/3 + 6y². Outer integral: ∫₁³ (8y/3 + 6y²) dy = [4y²/3 + 2y³]₁³ = (36/3 + 54) − (4/3 + 2) = (12 + 54) − (4/3 + 2) = 66 − 10/3 = 198/3 − 10/3 = 188/3. ✓

Both orders yield 188/3, confirming Fubini's Theorem.

SECTION 7

Strategies, Strengths & Common Pitfalls

While Fubini's Theorem is conceptually elegant, applying it in practice requires attention to several strategic and technical considerations. The following table compares the two integration orders and highlights situations where one is clearly preferable.

Consideration	dy dx Order (Type I)	dx dy Order (Type II)
Natural for...	Regions bounded above/below by functions of x	Regions bounded left/right by functions of y
Inner limits	y from g₁(x) to g₂(x)	x from h₁(y) to h₂(y)
Preferred when...	f(x, y) has a simpler antiderivative in y	f(x, y) has a simpler antiderivative in x, or dy dx leads to non-elementary inner integrals
Classic example	∫₀¹ ∫₀ˣ f dy dx — triangle below y = x	∫₀¹ ∫ᵧ¹ f dx dy — same triangle, reversed
Common pitfall	Forgetting that inner limits depend on the outer variable, not both	Incorrectly inverting boundary functions when switching order

Comparison of dy dx vs. dx dy integration orders

💡

STRATEGIC INSIGHT

The classic example that motivates order reversal is ∫₀¹ ∫ₓ¹

e^{y²}

dy dx. The inner integral ∫ₓ¹

e^{y²}

dy has no elementary antiderivative in y. But reversing the order to ∫₀¹ ∫₀ʸ

e^{y²}

dx dy makes the inner integral trivial: ∫₀ʸ

e^{y²}

dx = y·

e^{y²}

, which is then easily integrated with a u-substitution. Choosing the right order of integration can be the difference between an impossible integral and a routine one.

SECTION 8

Connection to Advanced Theory

Fubini's Theorem for double integrals is the entry point to a family of powerful results in multivariable and abstract analysis. Understanding these connections enriches your grasp of the theorem's role and prepares you for more advanced coursework in real analysis, probability, and mathematical physics.

Concept	Fubini for Double Integrals	Generalization / Extension
Dimension	Two variables (x, y) over ℝ²	n variables over ℝⁿ — triple integrals, etc. Fubini extends: an n-fold integral can be reduced to n successive single integrals.
Integration framework	Riemann integral (sufficient for continuous functions)	Lebesgue integral — Fubini-Tonelli theorem handles measurable functions and is essential in probability theory and functional analysis.
Coordinate systems	Cartesian (x, y)	Polar, cylindrical, spherical — iterated integrals combined with the change of variables (Jacobian) formula.
Applications	Area, volume, mass, center of mass	Moment of inertia, probability distributions (joint → marginal via iterated integration), Fourier transforms, convolutions.

Fubini's Theorem in context: from double integrals to broader mathematics

In your next encounters with triple integrals, you will apply Fubini's Theorem iteratively — reducing a triple integral to a double integral and then to a single integral. When you study polar coordinates, the iterated-integral structure remains the same, but the area element dA = r dr dθ introduces a Jacobian factor. In probability, computing E[X] for a jointly distributed pair (X, Y) amounts to applying Fubini to exchange the order of integration over the joint density — a technique used constantly in Bayesian inference and statistical mechanics. The theorem you learn here is, in a very real sense, the computational engine behind all of multivariable integration.

SECTION 9

Practice Problems

Question 1: CONCEPTUAL

Explain in your own words why Fubini's Theorem requires some hypothesis on f (such as continuity) rather than holding for every function. What could go wrong if f were 'badly behaved' on R?

Show answer

Without a regularity condition, the two iterated integrals might not even exist, or they might exist but yield different values. The classic counterexample involves a function on [0,1]² defined so that integrating in one order gives +1 and in the other gives −1; this occurs because the function fails to be absolutely integrable over the rectangle. Continuity ensures that all Riemann sums converge to the same limit regardless of partition, which in turn guarantees that the double integral and both iterated integrals agree. In the Lebesgue setting, absolute integrability (∬|f| dA < ∞) is the key sufficient condition.

Question 2: BASIC CALCULATION

Evaluate ∫₀³ ∫₀² (2x + 4y) dx dy.

Show answer

First, compute the inner integral with respect to x: ∫₀² (2x + 4y) dx = [x² + 4xy]₀² = (4 + 8y) − 0 = 4 + 8y. Next, integrate with respect to y: ∫₀³ (4 + 8y) dy = [4y + 4y²]₀³ = 12 + 36 = 48. The answer is 48. You can verify by reversing the order: ∫₀² ∫₀³ (2x + 4y) dy dx = ∫₀² [2xy + 2y²]₀³ dx = ∫₀² (6x + 18) dx = [3x² + 18x]₀² = 12 + 36 = 48. ✓

Question 3: INTERMEDIATE

Evaluate the integral ∫₀¹ ∫ₓ¹ sin(y²) dy dx by first reversing the order of integration.

Show answer

The original integral has inner limits y from x to 1 and outer limits x from 0 to 1. The region D is the triangle where 0 ≤ x ≤ y and 0 ≤ y ≤ 1. Reversing: ∫₀¹ ∫₀ʸ sin(y²) dx dy. The inner integral is ∫₀ʸ sin(y²) dx = y sin(y²) (since sin(y²) is constant with respect to x). Now integrate: ∫₀¹ y sin(y²) dy. Let u = y², du = 2y dy, so y dy = du/2. When y = 0, u = 0; when y = 1, u = 1. We get (1/2) ∫₀¹ sin(u) du = (1/2)[−cos(u)]₀¹ = (1/2)(−cos 1 + 1) = (1 − cos 1)/2 ≈ 0.2298.

Question 4: APPLIED

A thin metal plate occupies the region D bounded by y = 0, y = x, and x = 2. Its density at any point is ρ(x, y) = x + y (in g/cm²). Set up and evaluate an iterated integral to find the total mass of the plate.

Show answer

The region D is the triangle with vertices (0,0), (2,0), (2,2). As a Type I region: 0 ≤ x ≤ 2, 0 ≤ y ≤ x. Mass M = ∫₀² ∫₀ˣ (x + y) dy dx. Inner integral: ∫₀ˣ (x + y) dy = [xy + y²/2]₀ˣ = x² + x²/2 = 3x²/2. Outer integral: ∫₀² (3x²/2) dx = [x³/2]₀² = 8/2 = 4. The total mass is 4 grams. The physical interpretation is that we first compute the mass of each thin vertical strip (inner integral), then sum all strips from x = 0 to x = 2 (outer integral), which is exactly the iterated-integral strategy justified by Fubini's Theorem.

Question 5: CRITICAL THINKING

Consider the iterated integral ∫₀¹ ∫₀^(√y)

e^{x³}

dx dy. Explain why evaluating this integral in the given order is problematic, reverse the order of integration, and evaluate the result.

Show answer

The inner integral ∫₀^(√y)

e^{x³}

dx has no closed-form antiderivative in x — there is no elementary function whose derivative with respect to x is

e^{x³}

. To proceed, we reverse the order. The region D satisfies 0 ≤ y ≤ 1 and 0 ≤ x ≤ √y, equivalently 0 ≤ x ≤ 1 and x² ≤ y ≤ 1 (since x ≤ √y ⟺ y ≥ x²). The reversed integral is ∫₀¹ ∫_{x²}¹

e^{x³}

dy dx. Inner integral: ∫_{x²}¹

e^{x³}

dy = (1 − x²)

e^{x³}

. Outer integral: ∫₀¹ (1 − x²)

e^{x³}

dx = ∫₀¹

e^{x³}

dx − ∫₀¹ x²

e^{x³}

dx. For the second integral, let u = x³, du = 3x² dx, giving (1/3)∫₀¹ eᵘ du = (1/3)(e − 1). The first integral ∫₀¹

e^{x³}

dx does not have an elementary form. However, if we reconsider the factored integrand (1 − x²)

e^{x³}

and try integration by parts or re-grouping: note that d/dx[x·

e^{x³}

·(1/3)] would require careful handling. A cleaner approach: rewrite (1 − x²)

e^{x³}

e^{x³}

− x²

e^{x³}

. Let u = x³ in the second term: (1/3)(e − 1). The first term has no elementary form, so the complete closed form is ∫₀¹

e^{x³}

dx − (e − 1)/3, which can be expressed in terms of the incomplete gamma function. The key pedagogical point is that reversing the order was essential to making progress, and Fubini's Theorem guaranteed the validity of this reversal.

SUMMARY

Summary

A double integral ∬_R f(x, y) dA represents the signed volume under a surface over a region R. Fubini's Theorem provides the bridge between this abstract definition (as a limit of Riemann sums) and practical computation: when f is continuous (or more generally integrable) on R, the double integral equals either iterated integral, and the order of integration may be freely exchanged. On rectangular domains R = [a, b] × [c, d], the limits are constants; on Type I and Type II regions, the inner limits become functions of the outer variable.

Strategically choosing or reversing the order of integration can convert an intractable integral into a routine one — this is one of the most powerful computational techniques in multivariable calculus. The theorem extends naturally to triple integrals and higher dimensions, to non-Cartesian coordinate systems (with Jacobian factors), and to the Lebesgue framework via the Fubini–Tonelli theorem, making it a cornerstone of modern analysis, probability, and mathematical physics.