Real Analysis · Undergraduate

Limits of Sequences: The ε-N Definition

Quick answer

A sequence a₁, a₂, a₃, … converges to L when its terms eventually stay as close to L as anyone demands. Precisely: for every ε > 0 there is an N such that |aₙ − L| < ε for all n > N. A proof finds N in terms of ε, often by working backward from the inequality. To show that a sequence diverges, find one ε that no N can meet, as for (−1)ⁿ. From the definition, limits are unique and convergent sequences are bounded.

What you'll learn

  • State the ε-N definition of a limit
  • Prove limits like 1/n → 0 and (2n + 1)/(n + 3) → 2
  • Prove that a sequence diverges
  • Prove that limits are unique and convergent sequences are bounded

Eventually as close as anyone wants

A sequence (an)(a_n) converges to LL, written an→La_n \to L, if

∀ε>0 ∃N ∀n>N: ∣an−L∣<ε\forall\varepsilon > 0\ \exists N\ \forall n > N:\ |a_n - L| < \varepsilon

Read it as a game. A challenger names a tolerance ε\varepsilon. You answer with an index NN such that every term after the NN-th lies inside the band (L−ε,L+ε)(L - \varepsilon, L + \varepsilon). If you can always answer, the limit is LL. It mirrors the ε-δ definition for functions, with “far enough along the sequence” in place of “close enough to aa”.

Why the order of the quantifiers matters

NN is chosen after ε\varepsilon, so it may depend on ε\varepsilon: smaller tolerances usually need larger NN. But NN is chosen before nn, so one NN must work for every later term at once. The definition does not ask that terms get closer at every step, only that from some point on they never leave the band. A sequence can wander early and still converge; what happens in the first million terms never affects the limit.

Worked examples

Common mistakes

Practice problems

  1. Prove that 3n2→0\tfrac{3}{n^2} \to 0.

    Answer

    Choose N>3εN > \sqrt{\tfrac{3}{\varepsilon}}.

    Full solution

    For n>Nn > N, 3n2<3N2<ε\tfrac{3}{n^2} < \tfrac{3}{N^2} < \varepsilon.

  2. Prove that nn+1→1\tfrac{n}{n + 1} \to 1.

    Answer

    ∣nn+1−1∣=1n+1\left|\tfrac{n}{n + 1} - 1\right| = \tfrac{1}{n + 1}; choose N>1εN > \tfrac{1}{\varepsilon}.

    Full solution

    For n>Nn > N, 1n+1<1N<ε\tfrac{1}{n + 1} < \tfrac{1}{N} < \varepsilon.

  3. For an=2n+1n+3a_n = \tfrac{2n + 1}{n + 3} and ε=0.1\varepsilon = 0.1, find the smallest whole number NN such that every n>Nn > N has ∣an−2∣<0.1|a_n - 2| < 0.1.

    Answer

    4747

    Full solution

    Need 5n+3<0.1\tfrac{5}{n + 3} < 0.1, that is n+3>50n + 3 > 50, so n>47n > 47.

  4. Prove that an=na_n = n diverges.

    Answer

    It is unbounded, and convergent sequences are bounded.

    Full solution

    For any LL and ε=1\varepsilon = 1, terms with n>∣L∣+1n > |L| + 1 are more than 11 away from LL.

  5. Prove that if an→La_n \to L, then ∣an∣→∣L∣|a_n| \to |L|.

    Answer

    ∣∣an∣−∣L∣∣≤∣an−L∣\big||a_n| - |L|\big| \le |a_n - L|.

    Full solution

    The same NN that works for an→La_n \to L works for ∣an∣→∣L∣|a_n| \to |L|.

  6. Does ∣an∣→1|a_n| \to 1 imply that ana_n converges? Answer yes or no.

    Answer

    No

    Full solution

    an=(−1)na_n = (-1)^n has ∣an∣=1|a_n| = 1 for all nn, yet it diverges.

  7. Write the negation of an→La_n \to L in symbols.

    Answer

    ∃ε>0 ∀N ∃n>N: ∣an−L∣≥ε\exists\varepsilon > 0\ \forall N\ \exists n > N:\ |a_n - L| \ge \varepsilon

    Full solution

    Flip each quantifier and negate the inequality.

  8. A student proves 1n→0\tfrac{1}{n} \to 0 by saying ”1n\tfrac{1}{n} gets smaller and smaller.” Why is that not a proof?

    Answer

    Decreasing does not imply a limit of 00: 1+1n1 + \tfrac{1}{n} also decreases but tends to 11.

    Full solution

    A proof must show the terms eventually stay within every ε\varepsilon of 00, with an explicit NN.

Frequently asked questions

What does aₙ → L mean?

For every ε > 0 there is a natural number N such that |aₙ − L| < ε whenever n > N.

How do you find N in a proof?

Work backward: simplify |aₙ − L|, find which n make it smaller than ε, and choose N accordingly. Then write the proof forward.

How do you prove a sequence diverges?

Show that for every candidate L, some ε > 0 has terms at distance at least ε from L beyond every N.

Can a sequence have two limits?

No. If aₙ → L and aₙ → M, the terms would eventually be within ε of both, which forces |L − M| < 2ε for every ε, so L = M.

Is every convergent sequence bounded?

Yes. Beyond some N all terms lie within 1 of L, and the finitely many earlier terms have a largest size.

What to learn next