Ma · Mathematics

Why Do We Need the Formal Definition of a Limit?

Published By the 9to5Science editorial team

At a glance

  1. 1. Start with intuition

    A limit describes values approaching a target, not necessarily reaching it.

  2. 2. Identify the ambiguity

    Phrases such as “arbitrarily close” need precise meanings about distance and order.

  3. 3. Use epsilon and delta

    The formal definition turns closeness into a statement that can be tested and proved.

  4. 4. Apply it to examples

    Work through a removable discontinuity and a limit that does not exist.

A limit is often introduced informally by saying that f(x)f(x) approaches LL as xx approaches aa. This picture is useful, but it is not precise enough for every mathematical argument. The formal definition explains exactly what “approaches” means and gives you a method for proving that a proposed limit is correct.

The definition is not an arbitrary complication added to calculus. It records the logical content of the intuitive idea: by restricting xx sufficiently close to aa, you can force f(x)f(x) to be as close to LL as desired.

The intuitive idea and its limits

Suppose

lim⁡x→2(3x+1)=7.\lim_{x\to 2} (3x+1)=7.

A graph suggests that when xx is near 22, the value 3x+13x+1 is near 77. But several questions remain:

  • What does “near” mean numerically?
  • How near must xx be to 22 to make the output close to 77?
  • Does the required closeness depend on the desired accuracy?
  • Must x=2x=2 itself be considered?

The last question matters especially when the function is undefined at aa. For example, define

f(x)=x2−1x−1(x≠1).f(x)=\frac{x^2-1}{x-1}\quad (x\ne 1).

For every x≠1x\ne 1, this simplifies to f(x)=x+1f(x)=x+1, so the values approach 22 as xx approaches 11. The function need not be defined at 11 for the limit to exist.

A graph or table can suggest this conclusion, but neither one proves it. A table samples only finitely many points, while a graph has limited visual resolution. A formal definition covers every allowed value of xx simultaneously.

The epsilon-delta definition

Assume that aa is an accumulation point of the domain of ff, so the domain contains points other than aa arbitrarily close to aa. We write

lim⁡x→af(x)=L\lim_{x\to a}f(x)=L

if, for every ε>0\varepsilon>0, there exists a δ>0\delta>0 such that, for every xx in the domain of ff,

0<∣x−a∣<δ⟹∣f(x)−L∣<ε.0<|x-a|<\delta \quad\Longrightarrow\quad |f(x)-L|<\varepsilon.

Here:

  • ε\varepsilon is the permitted error in the output;
  • δ\delta is the required error in the input;
  • 0<∣x−a∣0<|x-a| excludes the point x=ax=a itself.

The order of the quantifiers is essential. First, someone specifies how accurate the output must be by choosing ε\varepsilon. Then you choose a suitable δ\delta. The value of δ\delta may depend on ε\varepsilon, but it cannot depend on the particular xx chosen afterward.

In plain language: no matter how small a tolerance is demanded around LL, there is a neighborhood around aa—possibly very small—such that every input in that neighborhood produces an output within the requested tolerance.

A complete example

Prove that

lim⁡x→2(3x+1)=7.\lim_{x\to 2}(3x+1)=7.

Start with the output error:

∣f(x)−7∣=∣(3x+1)−7∣=∣3x−6∣=3∣x−2∣.|f(x)-7|=|(3x+1)-7|=|3x-6|=3|x-2|.

To make this less than ε\varepsilon, it is enough to require

3∣x−2∣<ε.3|x-2|<\varepsilon.

This will happen whenever

∣x−2∣<ε3.|x-2|<\frac{\varepsilon}{3}.

Therefore choose δ=ε/3\delta=\varepsilon/3. If 0<∣x−2∣<δ0<|x-2|<\delta, then

∣(3x+1)−7∣=3∣x−2∣<3δ=ε.|(3x+1)-7|=3|x-2|<3\delta=\varepsilon.

This verifies the definition, so the limit is 77.

The proof does more than confirm a numerical answer. It gives a precise conversion: an input accuracy of less than ε/3\varepsilon/3 guarantees an output accuracy of less than ε\varepsilon.

Why informal reasoning can fail

Consider the expression

lim⁡x→0∣x∣x.\lim_{x\to 0}\frac{|x|}{x}.

From the right, x>0x>0, so ∣x∣/x=1|x|/x=1. From the left, x<0x<0, so ∣x∣/x=−1|x|/x=-1. The function therefore approaches different values from the two sides, and the two-sided limit does not exist.

A few positive and negative values in a table might be overlooked, and a graph might be misread near the origin. The formal definition exposes the problem: no single number LL can satisfy the required condition for all sufficiently close inputs on both sides.

The definition also prevents a common mistake: substituting x=ax=a and treating the result as the limit. Substitution works for many continuous functions, but a limit concerns nearby points and may exist even when the function is undefined at the point.

What the definition makes possible

The formal definition is needed because it supports general theorems rather than isolated calculations.

Uniqueness

A function cannot have two different limits at the same accumulation point of its domain. If both LL and MM were limits as xx approaches aa, choose an accuracy ε<∣L−M∣/2\varepsilon<|L-M|/2 for each limit and then choose a domain point xx in the intersection of the two sufficiently close punctured neighborhoods, which is possible because aa is an accumulation point of the domain. The triangle inequality gives

∣L−M∣≤∣L−f(x)∣+∣f(x)−M∣<2ε<∣L−M∣,|L-M|\le |L-f(x)|+|f(x)-M|<2\varepsilon<|L-M|,

which is a contradiction unless L=ML=M.

Limit laws

The familiar rules

lim⁡x→a(f(x)+g(x))=lim⁡x→af(x)+lim⁡x→ag(x)\lim_{x\to a}(f(x)+g(x)) =\lim_{x\to a}f(x)+\lim_{x\to a}g(x)

and corresponding product and quotient rules require proof. Epsilon-delta arguments provide that proof and state the conditions under which the rules are valid, especially the condition that a denominator’s limit is nonzero for a quotient.

Continuity

A function is continuous at aa precisely when it is defined there and

lim⁡x→af(x)=f(a).\lim_{x\to a}f(x)=f(a).

Thus the definition separates three related but distinct ideas: the function’s value at a point, its nearby behavior, and whether those agree.

Derivatives and integration

The derivative is itself a limit:

f′(a)=lim⁡h→0f(a+h)−f(a)h. f'(a)=\lim_{h\to 0}\frac{f(a+h)-f(a)}{h}.

The rigorous construction of the definite integral also depends on limiting processes. Without a precise definition of limits, these central ideas would rest only on diagrams and plausibility.

Common mistakes

  1. Requiring f(a)f(a) to exist. A limit can exist even if the function is undefined at aa.
  2. Using x=ax=a in the condition. The definition uses 0<∣x−a∣0<|x-a| because the nearby behavior is the subject.
  3. Choosing δ\delta before seeing ε\varepsilon. The input tolerance generally depends on the demanded output tolerance.
  4. Checking only a few values. Numerical evidence can suggest a limit but cannot establish a statement about every sufficiently close input.
  5. Confusing one-sided and two-sided limits. The left and right limits must agree for the ordinary two-sided limit to exist.

The formal definition is therefore a precise version of an intuitive idea, not a replacement for intuition. Intuition helps you guess the limit; the epsilon-delta definition tells you exactly what must be true and supplies the standard framework for proving it.

References

  1. Understanding Analysis — Stephen Abbott
  2. Principles of Mathematical Analysis — Walter Rudin

Spotted an error? Send a correction. Read how we write and check articles in our editorial policy.