Tutorials › Measure Theory › Measurable Level and Equality Sets

Measurable Functions · Tutorial 182 of 798

Measurable Level and Equality Sets

Use rational thresholds and sigma-algebra operations to describe and verify measurable level, equality, and comparison sets.

Intermediate 10 min read

What You'll Learn

  • Express strict and non-strict level sets using half-line preimages and countable intersections or unions
  • Distinguish the endpoint behavior of less-than, less-than-or-equal, and equality sets
  • Prove that comparisons between two measurable real-valued functions are measurable
  • Represent equality sets as complements of strict comparison sets
  • Check exact level and comparison sets in examples, including a rational-versus-irrational example
  • Avoid confusing set equality, almost-everywhere equality, and the measures of complementary sets

From Measurability to Measurable Sets

Measurability of a real-valued function is defined by the measurability of its preimages. Many sets that arise in analysis are precisely such preimages: a level set records where a function lies above or below a fixed number, while an equality set records where two functions take the same value. Comparisons between two functions require a little more care, because the functions need not be combined into a new function to describe the set.

Throughout, let \((X,\mathcal{M},\mu)\) be an arbitrary measure space; no finiteness, sigma-finiteness, or completeness assumption is imposed. Functions in this tutorial are finite, real-valued functions. A function \(f:X\to\mathbb{R}\) is measurable if \(f^{-1}(B)\in\mathcal{M}\) for every Borel set \(B\subseteq\mathbb{R}\). We use the half-line tests for real-valued measurability: in particular, if \(f\) is measurable, then \(\{x:f(x)<a\}\) and \(\{x:f(x)\le a\}\) belong to \(\mathcal{M}\) for every \(a\in\mathbb{R}\). A sigma-algebra is closed under complements and countable unions, and therefore also under countable intersections.

Definition: For a measurable \(f:X\to\mathbb{R}\) and \(a\in\mathbb{R}\), the strict and non-strict level sets are \(\{f<a\}\), \(\{f\le a\}\), \(\{f>a\}\), and \(\{f\ge a\}\). For measurable \(f,g:X\to\mathbb{R}\), the comparison sets are \(\{f<g\}\), \(\{f\le g\}\), \(\{f>g\}\), and \(\{f\ge g\}\); the equality set is \(\{f=g\}\). In each case, for example, \(\{f<a\}\) means \(\{x\in X:f(x)<a\}\).

Level Sets and Their Endpoints

Taking complements changes a strict inequality into the opposite non-strict inequality. For example, \(\{f>a\}=\{f\le a\}^{c}\), not \(\{f<a\}^{c}\). The latter complement is \(\{f\ge a\}\). This distinction matters at points where \(f(x)=a\).

Proposition: If \(f:X\to\mathbb{R}\) is measurable on \((X,\mathcal{M},\mu)\), then each of its four level sets at every real number \(a\) belongs to \(\mathcal{M}\). More explicitly, $$ \{f<a\}=\bigcup_{\substack{q\in\mathbb{Q}\\q<a}}\{f<q\},\qquad \{f\le a\}=\bigcap_{n=1}^{\infty}\{f<a+1/n\},\qquad \{f>a\}=\{f\le a\}^{c},\qquad \{f\ge a\}=\{f<a\}^{c}. $$

Proof. The half-line test gives \(\{f<q\}\in\mathcal{M}\) for each rational \(q\), so the displayed countable union is measurable. To verify its identity, if \(f(x)<a\), density of \(\mathbb{Q}\) in \(\mathbb{R}\) supplies a rational \(q\) with \(f(x)<q<a\), and hence \(x\) belongs to that union. Conversely, membership in one of its sets gives \(f(x)<q<a\), so \(f(x)<a\). For the intersection identity, \(f(x)\le a\) implies \(f(x)<a+1/n\) for every positive integer \(n\). If \(f(x)>a\), choose \(n\) large enough that \(1/n<f(x)-a\). Then \(f(x)>a+1/n\), so \(x\notin\{f<a+1/n\}\). This proves the identity. Finally, the last two identities follow by taking complements, and all four sets are measurable by closure of \(\mathcal{M}\) under countable unions, intersections, and complements. \(\square\)

The rational thresholds in the first identity are countable, which is essential: a sigma-algebra is guaranteed to be closed under countable unions, not arbitrary unions. The intersection identity also displays why the non-strict endpoint is retained: when \(f(x)=a\), the inequalities \(f(x)<a+1/n\) hold for every \(n\).

Comparisons and Equality of Two Functions

For comparisons between measurable functions, rational numbers again provide the needed countable description. A rational number strictly between two distinct real numbers can witness which one is smaller. This avoids assuming at this stage that a difference or any other algebraic combination of measurable functions is measurable.

Theorem: Let \(f,g:X\to\mathbb{R}\) be measurable on \((X,\mathcal{M},\mu)\). Then all four comparison sets and the equality set belong to \(\mathcal{M}\). In particular, $$ \{f<g\}=\bigcup_{q\in\mathbb{Q}}\bigl(\{f<q\}\cap\{g>q\}\bigr), \qquad \{f=g\}=X\setminus\bigl(\{f<g\}\cup\{g<f\}\bigr). $$ The remaining comparison sets are \(\{f\le g\}=\{g<f\}^{c}\), \(\{f>g\}=\{g<f\}\), and \(\{f\ge g\}=\{f<g\}^{c}\).

Proof. First, for each rational \(q\), \(\{f<q\}\) and \(\{g>q\}\) are measurable by the level-set result, so their intersection is measurable. There are countably many rational numbers, and therefore the union in the formula is measurable. If \(f(x)<g(x)\), density of the rationals gives \(q\in\mathbb{Q}\) with \(f(x)<q<g(x)\), placing \(x\) in the union. Conversely, membership in the union means there is a rational \(q\) such that \(f(x)<q\) and \(g(x)>q\); these inequalities imply \(f(x)<g(x)\). Thus the first identity holds. Interchanging \(f\) and \(g\) proves measurability of \(\{g<f\}\). At each \(x\), exactly one of \(f(x)<g(x)\), \(f(x)=g(x)\), or \(g(x)<f(x)\) holds, because \(f(x)\) and \(g(x)\) are real numbers. Hence the equality-set identity follows. The other identities follow directly from the definitions and complements; for example, \(f(x)\le g(x)\) holds exactly when \(g(x)<f(x)\) does not. All these sets are in \(\mathcal{M}\). \(\square\)

These are pointwise set identities. They do not say merely that the relations hold almost everywhere, and they do not require \(\mu(X)\) to be finite. If two functions agree outside a null set, their equality set contains the complement of that null set; that is a different statement from equality at every point.

Worked Example: Strict and Non-Strict Levels of a Square

On \(\mathbb{R}\) with its Borel sigma-algebra and Lebesgue measure \(m\), let \(f(x)=x^2\). This is a continuous, hence Borel-measurable, real-valued function. Find the level sets at \(a=4\), and check their measures.

The inequality \(x^2<4\) is equivalent to \(-2<x<2\), so \(\{f<4\}=(-2,2)\). The inequality \(x^2\le4\) is equivalent to \(-2\le x\le2\), so \(\{f\le4\}=[-2,2]\). Taking complements gives \(\{f>4\}=(-\infty,-2)\cup(2,\infty)\) and \(\{f\ge4\}=(-\infty,-2]\cup[2,\infty)\). Equality requires \(x^2=4\), which factors as \((x-2)(x+2)=0\); thus \(\{f=4\}=\{-2,2\}\).

Check: the interval \((-2,2)\) has length \(2-(-2)=4\), so \(m(\{f<4\})=4\). The closed interval has the same measure, \(m(\{f\le4\})=4\), since adding its two endpoints does not change Lebesgue measure. Each singleton has Lebesgue measure zero, so \(m(\{f=4\})=0\). The complements are consistent: the two tails in \(\{f>4\}\) exclude the endpoints, whereas \(\{f\ge4\}\) includes them.

Worked Example: Comparing Two Affine Functions

On \(\mathbb{R}\) with Lebesgue measure, let \(f(x)=x\) and \(g(x)=2-x\). Find the strict and non-strict comparison sets and the equality set. Both functions are continuous and therefore measurable.

The comparison \(f(x)<g(x)\) is \(x<2-x\). Adding \(x\) to both sides gives \(2x<2\), and dividing by the positive number \(2\) gives \(x<1\). Thus \(\{f<g\}=(-\infty,1)\). Replacing the strict inequality by a non-strict one gives \(x\le2-x\), hence \(2x\le2\) and \(x\le1\). Therefore \(\{f\le g\}=(-\infty,1]\). Reversing the comparison gives \(\{f>g\}=(1,\infty)\) and \(\{f\ge g\}=[1,\infty)\).

For equality, solve \(x=2-x\): adding \(x\) yields \(2x=2\), so \(x=1\) and \(\{f=g\}=\{1\}\). Check: at \(x=0\), \(f(0)=0<2=g(0)\); at \(x=1\), \(f(1)=1=g(1)\); and at \(x=2\), \(f(2)=2>0=g(2)\). These test points confirm the direction of each interval and that the boundary point belongs to both non-strict comparison sets, but neither strict comparison set.

Worked Example: An Equality Set Equal to the Rationals

On \(\mathbb{R}\) with its Borel sigma-algebra and Lebesgue measure, define \(f(x)=0\) for rational \(x\) and \(f(x)=1\) for irrational \(x\), and let \(g(x)=0\) for every \(x\). Verify measurability and find the comparison and equality sets.

For any open set \(U\subseteq\mathbb{R}\), the preimage \(f^{-1}(U)\) is one of \(\varnothing\), \(\mathbb{Q}\), \(\mathbb{R}\setminus\mathbb{Q}\), or \(\mathbb{R}\): the possibilities depend on whether \(U\) contains \(0\), \(1\), both, or neither. The rationals are Borel because \(\mathbb{Q}\) is a countable union of closed singleton sets. Its complement is Borel as well. Thus these possible preimages are Borel, so \(f\) is measurable; the constant function \(g\) is measurable too.

On rational inputs, \(f(x)=g(x)=0\); on irrational inputs, \(f(x)=1>0=g(x)\). Consequently \(\{f=g\}=\mathbb{Q}\), \(\{f<g\}=\varnothing\), and \(\{f>g\}=\mathbb{R}\setminus\mathbb{Q}\). Check: \(\mathbb{Q}\) is countable and every singleton has Lebesgue measure zero, so countable subadditivity gives \(m(\mathbb{Q})=0\). Since \(\mathbb{R}=\mathbb{Q}\,\dot\cup\,(\mathbb{R}\setminus\mathbb{Q})\), with both sets measurable, and \(m(\mathbb{R})=\infty\), additivity gives \(m(\mathbb{R}\setminus\mathbb{Q})=\infty\). Their Lebesgue measures are not both infinite: \(m(\mathbb{Q})=0\), while \(m(\mathbb{R}\setminus\mathbb{Q})=\infty\).

Common Endpoint and Measure Pitfalls

A common endpoint error is to solve a non-strict inequality as though it were strict. For example, the incorrect calculation “\(x^2\le4\) gives \(-2<x<2\)” omits the solutions \(x=-2\) and \(x=2\). Substitution checks the error: \((-2)^2=4\) and \(2^2=4\), so both points satisfy \(x^2\le4\). The correction is \(\{x:x^2\le4\}=[-2,2]\), while \(\{x:x^2<4\}=(-2,2)\). Equalities at endpoints can leave the Lebesgue measure unchanged, as in this example, but they still change the set.

A separate mistake is to infer that a measurable set and its complement have the same measure, or that both must have infinite measure. The rational example gives a direct correction: the set and its complement have measures \(0\) and \(\infty\), respectively. Measurability is a statement about belonging to the sigma-algebra; it does not imply finite measure, positive measure, or any symmetry between a set and its complement.

The countable formulas above are useful whenever a proof needs to restrict attention to where a function satisfies a numerical condition. For example, a claim about \(f(x)<g(x)\) can be handled as a measurable-set question without first constructing or proving measurability of a new function. The rational separator is the key: each possible threshold gives a measurable intersection, and the union over rational thresholds remains countable.

Check Your Understanding

Use the set identities and check strictness at any boundary points.

  1. For measurable real-valued \(f\), write \(\{f\ge a\}\) as a complement of a strict level set. Explain which points with \(f(x)=a\) it contains.
  2. Prove directly from density of \(\mathbb{Q}\) that \(\{f<g\}=\bigcup_{q\in\mathbb{Q}}(\{f<q\}\cap\{g>q\})\).
  3. For \(f(x)=x^2\) and \(g(x)=1\) on \(\mathbb{R}\), find \(\{f=g\}\), \(\{f<g\}\), and \(\{f\le g\}\).
  4. Explain why the equality set of two measurable real-valued functions is measurable, using only the two strict comparison sets and sigma-algebra operations.
  5. For the rational-versus-irrational function in the worked example, state the measures of its equality set with the zero function and of the strict greater-than set.