Overview
The epsilon-delta definition makes the intuitive notion of a limit precise. We say lim_{x->a} f(x) = L if for every epsilon > 0 (no matter how small), there exists delta > 0 such that if 0 < |x - a| < delta then |f(x) - L| < epsilon. This formalizes 'f(x) gets arbitrarily close to L as x approaches a'. The epsilon-delta approach, developed by Cauchy and Weierstrass in the 19th century, replaced vague geometric intuition with rigorous inequalities and is the foundation of real analysis.
Intuition
Challenge-response game: your opponent picks epsilon > 0 (a small tolerance around L). You win if you can find delta > 0 such that whenever x is within delta of a (but not equal to a), f(x) is within epsilon of L. You must win for every possible epsilon your opponent chooses. If you can always win, the limit exists and equals L. The key: 0 < |x-a| < delta (x is close to a but not equal) implies |f(x)-L| < epsilon (f(x) is close to L).
Formal Definition
lim_{x->a} f(x) = L means: for every real epsilon > 0, there exists delta > 0 such that for all x in the domain of f: 0 < |x - a| < delta implies |f(x) - L| < epsilon. The condition 0 < |x-a| says x != a (we don't require f(a) = L or even f(a) to be defined). To prove a limit, one typically: (1) guess L; (2) start with |f(x)-L| < epsilon; (3) work backwards to find delta in terms of epsilon; (4) write the proof forward: assume 0 < |x-a| < delta, then derive |f(x)-L| < epsilon.
Notation
| Notation | Meaning |
|---|---|
| Output tolerance (how close f(x) must be to L) | |
| Input restriction (how close x must be to a) | |
| Distance from x to a | |
| x is near a but not equal to a |
Theorems
Worked Examples
- 1
We need: for every epsilon > 0, find delta > 0 such that 0 < |x-2| < delta implies |(3x-1) - 5| < epsilon.
- 2
Simplify: |(3x-1) - 5| = |3x - 6| = 3|x - 2|.
- 3
We want 3|x-2| < epsilon, i.e., |x-2| < epsilon/3. So choose delta = epsilon/3.
- 4
Proof: given epsilon > 0, let delta = epsilon/3. Assume 0 < |x-2| < delta = epsilon/3. Then |(3x-1)-5| = 3|x-2| < 3*(epsilon/3) = epsilon. Done.
✓ Answer
Choose delta = epsilon/3. Then 0 < |x-2| < delta implies |(3x-1)-5| = 3|x-2| < epsilon.
Practice Problems
Use the epsilon-delta definition to prove lim_{x->4} sqrt(x) = 2.
Common Mistakes
Confusing the limit with the value of the function: thinking lim_{x->a}f(x)=f(a) must be true.
The limit lim_{x->a}f(x) is about the behavior of f near a, NOT at a. The function need not be defined at a, or f(a) may differ from the limit. For example, f(x) = (x^2-1)/(x-1) has a limit of 2 as x->1, even though f(1) is undefined. The condition 0 < |x-a| in the definition explicitly excludes x=a.
Quiz
Historical Background
Newton and Leibniz invented calculus in the 17th century using intuitive notions of infinitesimals. Euler's 18th century calculus was powerful but lacked rigorous foundations. Cauchy (1821) introduced the epsilon-delta language in his Cours d'analyse, though his formulation was still somewhat informal. Weierstrass (1860s) gave the modern precise definition. The need for rigor came from paradoxes: Dirichlet's 1829 function (1 on rationals, 0 on irrationals) required a precise definition to discuss its limits.
- 1821
Cauchy introduces proto-epsilon-delta language in Cours d'analyse
Augustin-Louis Cauchy
- 1860
Weierstrass gives the modern precise epsilon-delta definition
Karl Weierstrass
- 1872
Weierstrass constructs a nowhere-differentiable continuous function, showing rigor is necessary
Karl Weierstrass
Summary
- lim_{x->a}f(x)=L means: for every epsilon>0 there exists delta>0 such that 0<|x-a|<delta implies |f(x)-L|<epsilon.
- Intuition: epsilon-delta is a game -- for any target tolerance epsilon, you can find a neighborhood delta that keeps f(x) within epsilon of L.
- Proof technique: start with |f(x)-L|<epsilon, work backwards to find delta, then write the forward proof.
- Uniqueness: limits are unique (if they exist). The algebra of limits makes polynomials easy: lim p(x) = p(a).
References
- BookSpivak, M. Calculus. 4th ed. Publish or Perish, 2008.
- BookRudin, W. Principles of Mathematical Analysis. McGraw-Hill, 1976.
Mathematics