For all, there exists
The [[universal-quantifier|universal quantifier]] ∀ means “for every,” and the [[existential-quantifier|existential quantifier]] ∃ means “there exists at least one.” “∀ε > 0, ε > 0” is true; “∃x, x² = 2” is true over the reals, false over the rationals. To prove ∀x P(x) you must handle an arbitrary x — you may not pick a convenient one. To prove ∃x P(x) you only need to exhibit one witness.
Order matters enormously. “∀x ∃y (y > x)” says every number has something bigger — true. Swap them: “∃y ∀x (y > x)” says one fixed y beats every x — false, there is no largest real. When you read a definition in analysis, track the order of quantifiers like you track parentheses: a later quantifier may depend on the variables chosen before it.
The negation rule
Here is the one rule that does all the work. To form a [[negation-of-a-quantifier|negation]], push the ¬ inward: ∀ becomes ∃, ∃ becomes ∀, and the inner statement gets negated. ¬(∀x P(x)) is ∃x ¬P(x): “not everyone has property P” means “someone lacks it.” ¬(∃x P(x)) is ∀x ¬P(x): “no one has P” means “everyone lacks it.” Apply it left to right, flipping each quantifier as you pass it.
Negating the definition of "a_n -> L" mechanically. Converges: for all eps > 0, there exists N, for all n >= N, |a_n - L| < eps (∀ eps) (∃ N) (∀ n) ( ... ) Flip every quantifier and negate the inside. The inside |a_n - L| < eps negates to |a_n - L| >= eps. Does NOT converge to L: there exists eps > 0, for all N, there exists n >= N, |a_n - L| >= eps (∃ eps) (∀ N) (∃ n) ( ... ) In words: some fixed gap eps that the terms keep exceeding, no matter how far out (n >= N) you look. That is exactly what it means for the sequence to stay away from L. Check on a_n = (-1)^n with L = 1: take eps = 1. For any N, the next odd n >= N gives a_n = -1, so |a_n - 1| = 2 >= 1. The negation holds, so (-1)^n does not converge to 1.
Notice what the negation gives you: a single bad ε that the sequence keeps violating. This is why [[divergence|divergence]] proofs and [[counterexample|counterexamples]] always begin “take ε = …”. The negation of a ∀-statement is an ∃-statement, and an ∃-statement is proved by exhibiting one explicit witness.
Quantifiers inside implications
One more piece. To negate an implication, recall that P ⇒ Q is the same as ¬P ∨ Q, so ¬(P ⇒ Q) is P ∧ ¬Q — the hypothesis holds but the conclusion fails. Combined with the quantifier rule, you can now negate any statement in analysis. The phrase [[arbitrarily-small|arbitrarily small]] is really “∀ε > 0,” so its negation is “∃ε > 0 below which we cannot go” — a positive gap.