What Are Limits?
A limit describes the value a function approaches as the input gets close to a chosen number. If \(x\) moves closer and closer to \(a\), and the function values \(f(x)\) move closer and closer to one value \(L\), we write:
$$
\lim_{x \to a} f(x) = L
$$
The key idea is “approaches.” A limit is about the behavior of a function near a point, not necessarily the value of the function exactly at that point. A function can have a limit at \(x=a\) even if \(f(a)\) is undefined, or even if \(f(a)\) is defined as a different value.
Limits are useful because many important calculus ideas depend on what happens near a point. Continuity, derivatives, tangent lines, vertical asymptotes, and removable discontinuities all rely on the idea of approaching a value rather than simply substituting a value.
For example, the function
$$
f(x)=\frac{x^2-9}{x-3}
$$
is undefined at \(x=3\) because direct substitution gives \(\frac{0}{0}\). But for values of \(x\) near \(3\), the function behaves like \(x+3\), so the function values approach \(6\). The limit exists even though the original function is not defined at that exact input.
Why Limits Matter
Limits help answer questions that ordinary substitution cannot answer. They let you describe what a function is trying to do near a point, especially when direct substitution fails.
Limits matter when you want to:
-
Understand whether a function is continuous.
-
Study a hole, jump, or vertical asymptote in a graph.
-
Estimate behavior from a table of nearby values.
-
Build the idea of a derivative as an instantaneous rate of change.
-
Check whether left-side and right-side behavior agree.
-
Separate the value of a function at a point from the value it approaches near that point.
This is why limits appear early in calculus. They are the bridge between algebraic function values and the “nearby behavior” that calculus studies.
Key Terms to Know
-
Function value: The actual value \(f(a)\) obtained by substituting \(a\) into the function, when that value is defined.
-
Limit: The value \(f(x)\) approaches as \(x\) gets close to \(a\).
-
Left-hand limit: The value approached as \(x\) gets close to \(a\) from values less than \(a\), written as \(\lim_{x \to a^-} f(x)\).
-
Right-hand limit: The value approached as \(x\) gets close to \(a\) from values greater than \(a\), written as \(\lim_{x \to a^+} f(x)\).
-
Two-sided limit: The ordinary limit \(\lim_{x \to a} f(x)\), which exists only when the left-hand and right-hand limits both exist and are equal.
-
Direct substitution: Evaluating \(f(a)\) directly. This works for many continuous functions, but it does not always determine the limit.
-
Removable discontinuity: A “hole” in a graph where the limit exists but the function value is missing or different.
-
Vertical asymptote: A vertical line near which function values grow without bound in at least one direction.
-
Numerical approximation: An estimate based on sampled function values rather than an exact symbolic proof.
How Limits Work
A two-sided limit exists only when the function approaches the same value from both directions. In notation, the limit exists when:
$$
\lim_{x \to a^-} f(x) = L
\quad \text{and} \quad
\lim_{x \to a^+} f(x) = L
$$
Then the two-sided limit is:
$$
\lim_{x \to a} f(x)=L
$$
If the left-hand and right-hand limits approach different finite values, the two-sided limit does not exist. If one side grows without bound, oscillates, or fails to settle toward a value, the two-sided limit also does not exist.
Numerical Sampling Near a Point
A numerical limit estimate uses nearby inputs on both sides of \(a\). For a small positive distance \(h\), the left and right samples are:
$$
\text{left sample}=f(a-h)
$$
$$
\text{right sample}=f(a+h)
$$
By making \(h\) smaller, the sampled \(x\)-values move closer to \(a\). A common numerical approach is to compare values such as:
$$
h=10^{-1},\ 10^{-2},\ 10^{-3},\ 10^{-4},\ 10^{-5}
$$
If the left and right samples settle toward the same number, that number is a good numerical estimate of the two-sided limit. If they settle toward different values, the two-sided limit does not exist.
The calculator parses expressions with its restricted math-expression parser; it does not execute user text as JavaScript. For forms outside its bounded exact analysis, it uses adaptive, scale-aware samples on each requested side. It requires several of the closest valid samples to stabilize before displaying a numerical estimate. If they do not stabilize, the result is reported as inconclusive rather than forcing an answer.
Exact-analysis badges identify results obtained from continuity, algebraic cancellation, degree comparison, domain analysis, or a supported standard limit. For numerical two-sided checks, both tails must stabilize and agree within a relative tolerance with only a small floating-point roundoff guard, so a genuine small jump is not averaged into a false limit. Numerical estimates remain evidence rather than proof.
Examples of Limits in Practice
Example 1: A Removable Discontinuity
Consider:
$$
f(x)=\frac{x^2-9}{x-3}
$$
At \(x=3\), direct substitution gives:
$$
f(3)=\frac{3^2-9}{3-3}=\frac{0}{0}
$$
So \(f(3)\) is undefined. But the numerator factors:
$$
x^2-9=(x-3)(x+3)
$$
For \(x \ne 3\), the function simplifies to:
$$
\frac{x^2-9}{x-3}=\frac{(x-3)(x+3)}{x-3}=x+3
$$
Now values near \(3\) are easy to understand:
| \(h\) |
\(f(3-h)\) |
\(f(3+h)\) |
| \(10^{-1}\) |
\(5.9\) |
\(6.1\) |
| \(10^{-2}\) |
\(5.99\) |
\(6.01\) |
| \(10^{-3}\) |
\(5.999\) |
\(6.001\) |
| \(10^{-4}\) |
\(5.9999\) |
\(6.0001\) |
Both sides approach \(6\), so:
$$
\lim_{x \to 3}\frac{x^2-9}{x-3}=6
$$
This example shows why direct substitution and limits are not always the same thing. The function is undefined at \(x=3\), but the nearby behavior is clear.
Example 2: A Vertical Asymptote
Consider:
$$
f(x)=\frac{1}{x-2}
$$
At \(x=2\), the denominator is zero. Sampling near \(2\) gives very different behavior on the two sides:
| \(h\) |
\(f(2-h)\) |
\(f(2+h)\) |
| \(10^{-1}\) |
\(-10\) |
\(10\) |
| \(10^{-2}\) |
\(-100\) |
\(100\) |
| \(10^{-3}\) |
\(-1000\) |
\(1000\) |
| \(10^{-4}\) |
\(-10000\) |
\(10000\) |
From the left, the values decrease without bound. From the right, the values increase without bound. The two-sided finite limit does not exist.
The exact rational analyzer determines that the left-hand limit is \(-\infty\) and the right-hand limit is \(+\infty\), so the two-sided limit does not exist. The nearby table still shows the sampled growth behind that classification.
Example 3: An Oscillating Function
Consider:
$$
f(x)=\sin\left(\frac{1}{x}\right)
$$
as \(x \to 0\). The expression \(\frac{1}{x}\) grows rapidly in magnitude near zero, so the sine function oscillates faster and faster. Some sampled values from the right are:
| \(x\) |
\(\sin\left(\frac{1}{x}\right)\) |
| \(10^{-1}\) |
\(\approx -0.544\) |
| \(10^{-2}\) |
\(\approx -0.506\) |
| \(10^{-3}\) |
\(\approx 0.827\) |
| \(10^{-4}\) |
\(\approx -0.306\) |
| \(10^{-5}\) |
\(\approx 0.036\) |
These values do not settle toward one number. The limit does not exist.
Oscillatory functions are one reason numerical limit estimates need caution. The exact analyzer recognizes common unbounded trigonometric phases; otherwise adaptive sampling deliberately returns an inconclusive result when values do not stabilize.
How to Interpret the Result
The main result identifies whether it came from exact analysis, domain analysis, or adaptive numerical sampling. Exact results follow a supported algebraic or continuity rule; numerical results should be read as evidence about nearby behavior.
A proven finite result follows from a supported exact rule. A numerical estimate means several adaptive samples appear to stabilize; for a two-sided estimate, both sides must stabilize near the same number.
A left-hand estimate comes from inputs just less than the approach value. A right-hand estimate comes from inputs just greater than the approach value. If these two estimates differ, the two-sided limit should not be treated as a single finite number.
A direct substitution value shows \(f(a)\), the value at the exact input. This may match the limit for continuous functions, but it may also be undefined or different from the limit. A removable discontinuity is the classic case where direct substitution fails but the limit still exists.
A DNE classification means a supported analysis proves incompatible one-sided behavior, detects an absent real-domain side, or recognizes unbounded oscillation. Unstable evidence that cannot justify one of these conclusions is labeled inconclusive instead.
The near-\(x\) table is often the most useful part of the result. It lets you see whether the function values are stabilizing, diverging, or behaving differently on the left and right. The graph preview can help you see the same behavior visually, but the graph is still only a finite-window preview.
Common Mistakes and Misconceptions
Confusing direct substitution with the limit. Direct substitution gives \(f(a)\), while the limit describes what \(f(x)\) approaches near \(a\). These are the same for many continuous functions, but not for all functions.
Assuming a numerical estimate is a proof. A table of nearby values can strongly suggest a limit, but it does not prove the limit in the same way that algebraic simplification or a formal argument can.
Ignoring one-sided behavior. A two-sided limit exists only when both sides approach the same value. Checking only one side can hide a jump, vertical asymptote, or domain restriction.
Reading large values as a confirmed infinite limit. A value such as \(100000\) or \(-100000\) is still a finite sampled value. It may suggest infinite behavior, but the trend should be checked carefully.
Misreading implicit multiplication. Both 2*x and 2x are accepted, as are 3*(x+1) and 3(x+1); 2*pi*x is also accepted. Explicit multiplication can still make a long expression easier to read.
Using the wrong variable. The calculator is designed for functions of \(x\). Expressions using other variable names are not supported.
Using degrees in trigonometric functions. Trigonometric inputs such as sin(x), cos(x), and tan(x) use radians. For example, sin(pi/2) represents \(\sin\left(\frac{\pi}{2}\right)\) and gives \(1\).
Using unsupported syntax. Stick to standard arithmetic, parentheses, powers, constants such as pi and e, and supported functions such as sqrt, cbrt, abs, exp, trigonometric functions, and logarithms.
When to Use Limit Estimates
Use numerical limit estimates when you want a quick way to explore function behavior near a point.
They are especially helpful for:
-
Checking whether direct substitution seems reliable.
-
Investigating holes and removable discontinuities.
-
Comparing left-hand and right-hand behavior.
-
Previewing vertical asymptotes or very large outputs.
-
Building intuition before solving a limit algebraically.
-
Teaching or learning how tables and graphs reveal nearby behavior.
A numerical estimate is most useful as a first look. For homework, exams, proofs, or technical work, use it alongside exact methods whenever exact reasoning is expected.
Limitations and Things to Keep in Mind
The calculator first applies a bounded exact analysis to continuous elementary expressions, polynomial and rational functions, selected standard limits, and recognizable oscillatory forms. Expressions outside that subset use adaptive numerical sampling, which can only inspect selected points and may return an inconclusive result.
The result can be misleading for functions with:
-
Rapid oscillation near the approach value.
-
Narrow spikes or dips between sampled points.
-
Very slow convergence.
-
Vertical asymptotes that produce large but finite sampled values.
-
Piecewise behavior that cannot be entered directly.
-
Domain restrictions on only one side of the approach value.
Displayed finite values are rounded for readability, with up to \(10\) decimal places. Very small differences can be hidden by formatting, and values that appear equal after rounding may not be exactly equal.
The graph preview is also limited. It uses a fixed window around the approach value and omits non-finite values and extremely large vertical values. A graph can reveal useful patterns, but it should not be treated as a complete proof of limiting behavior.
For important academic, scientific, engineering, financial, or safety-related work, double-check the result using exact mathematics, appropriate software, or guidance from a qualified instructor or professional.
How to Use This Calculator
-
Enter a function using \(x\) as the variable.
-
Enter arithmetic with explicit or implicit multiplication and supported functions such as
sqrt, abs, exp, sin, cos, tan, ln, or log.
-
Enter the value that \(x\) should approach. Use a finite number,
infinity, or -infinity.
-
Choose the direction: left, right, or both.
-
Review the main result and badges to see whether the estimate is finite, one-sided, two-sided, or classified as not existing.
-
Compare direct substitution, the left estimate, the right estimate, and the near-\(x\) table to understand the result.
-
Use the graph preview as a visual check, and download the graph if you need to save it.
Frequently Asked Questions
Does a limit have to equal the function value?
No. A limit describes what \(f(x)\) approaches as \(x\) gets close to \(a\). The function value \(f(a)\) may be the same as the limit, different from the limit, or undefined.
When does a two-sided limit exist?
A two-sided limit exists when the left-hand and right-hand limits both exist and are equal. If the two sides approach different values, the two-sided limit does not exist.
Why can direct substitution be undefined when the limit exists?
Direct substitution can produce an undefined form such as \(\frac{0}{0}\) even when nearby values approach a clear number. This often happens with removable discontinuities, where algebraic simplification reveals the limiting behavior.
Can this calculator prove an exact limit?
It can prove limits inside its documented bounded subset, including continuous elementary expressions, polynomial and rational functions, selected standard forms, and common oscillatory cases. Other expressions use conservative adaptive sampling and may be reported as inconclusive.
What does it mean if the result says the limit does not exist?
It means the sampled one-sided behavior did not support a single finite two-sided limit. The left and right estimates may differ, one side may be non-finite, or the nearby function behavior may fail to settle.
Why do trigonometric results look different from degree-based calculations?
Trigonometric functions use radians. For degree-based angles, convert degrees to radians first. For example, \(90^\circ=\frac{\pi}{2}\) radians.
Sources and References
Books
-
Gilbert Strang and Edwin “Jed” Herman. Calculus Volume 1. OpenStax, Rice University, 2016. Chapter 2, especially sections 2.2 “The Limit of a Function” and 2.4 “Continuity.” OpenStax Calculus Volume 1, Section 2.2; OpenStax Calculus Volume 1, Section 2.4.
-
Matthew Boelkins, David Austin, Christina Safranski, and Steven Schlicker. Active Calculus: Single Variable, 2nd Edition. Grand Valley State University Libraries, 2025. Section 1.7 “Limits, continuity, and differentiability.” Active Calculus Section 1.7; ScholarWorks@GVSU book record.
Online and Official Sources
-
MIT OpenCourseWare. “Session 4: Limits and Continuity.” Single Variable Calculus, Fall 2010. Accessed July 4, 2026. MIT OpenCourseWare Session 4.
