How to Find the Support of a Function

The support of a function records where the function is nonzero, together with the boundary points of that region. To find it, first identify all inputs where the function’s value is not zero, then take the closure of that set.

The answer can depend on the function’s domain and on whether you mean ordinary support or essential support. These distinctions matter especially when a function is defined only on an open interval or is considered up to equality almost everywhere.

The standard definition

Let $f:X\to\mathbb{R}$ (or $\mathbb{C}$), where $X$ is a topological space. The support of $f$ is

$$\operatorname{supp}(f)=\overline{\{x\in X:f(x)\ne 0\}},$$

where the bar means closure in $X$. In plain terms, include every point where the function is nonzero and every point that can be approached arbitrarily closely by such points.

The function does not have to be nonzero at every point in its support. For example, a boundary point where $f$ equals zero can still belong to the support if nonzero values occur arbitrarily close to it.

A step-by-step method

1. Fix the domain and ambient space. Decide whether the function is being considered on an interval, on all of $\mathbb{R}$, or on some other space. The closure is taken in that space.
2. Solve $f(x)\ne 0$. This gives the set of points where the function is nonzero. For a piecewise function, examine each piece and its endpoints.
3. Take the closure. Add any limit points of the nonzero set that lie in the chosen space.
4. Check for a different convention. In measure theory and some applications, “support” may mean essential support instead.

Worked examples

An indicator function

Define $f:\mathbb{R}\to\mathbb{R}$ by

$$f(x)= \begin{cases} 1,&0<x<1,\\ 0,&\text{otherwise}. \end{cases}$$

The nonzero set is $(0,1)$. Its closure in $\mathbb{R}$ is $[0,1]$, so

$$\operatorname{supp}(f)=[0,1].$$

Although $f(0)=f(1)=0$, both endpoints are in the support because points where $f$ is nonzero approach them.

A function that is nonzero everywhere

For $g(x)=e^{-x^2}$ on $\mathbb{R}$, $g(x)>0$ for every real $x$. Its nonzero set is all of $\mathbb{R}$, so its support is $\mathbb{R}$. It is not compactly supported: its values approach zero as $|x|$ grows, but they never become zero outside a bounded interval.

Likewise, a nonzero special function that has only isolated zeros on its domain will generally have support equal to that whole domain. Zeros alone do not remove points from the support; only an open region on which the function vanishes can be excluded.

Why the domain matters

Consider the constant function $h(x)=1$ defined only for $x\in(0,1)$. If its domain is the space $X=(0,1)$, then its support relative to $X$ is $(0,1)$. If instead you extend it by zero to all of $\mathbb{R}$, its support in $\mathbb{R}$ is $[0,1]$. State the domain or zero extension when endpoints affect the answer.

Ordinary support and essential support

For a measurable function on $\mathbb{R}^n$ with Lebesgue measure (or in a setting where the usual essential-support definition is well-defined), the essential support is the complement of the largest open set on which the function is zero almost everywhere. Equivalently, it is unchanged by modifying the function on a set of measure zero.

This can differ from ordinary support. For example, let $q(x)=1$ at $x=0$ and $q(x)=0$ everywhere else on $\mathbb{R}$. Its ordinary support is $\{0\}$. Since it is zero almost everywhere, its essential support is empty. In many analysis problems involving integrals or $L^p$ spaces, essential support is the relevant notion; check how the text defines the term.

Common mistakes

• Reporting only where the function is nonzero. That set need not be closed. The standard support includes its limit points.
• Discarding boundary points where the function equals zero. A zero value at a point does not exclude it from the support if nonzero points approach it.
• Ignoring the domain. The support of a function on an open interval is taken relative to that interval unless an extension or ambient space is specified.
• Confusing small values with zero values. A function that decays toward zero may still be nonzero everywhere and therefore have support across its entire domain.
• Using ordinary support for an almost-everywhere problem. If functions are identified when they agree almost everywhere, essential support is usually the appropriate convention.