Continuity definitions are equivalent
Contents
[hide]Statement
The definitions of continuity for a function f:(X,d)→(Y,d′) from one metric space to another is the same as f:(X,J)→(Y,K) being continuous (where the topologies are those induced by the metric are the same, that is
- ∀a∈X∀ϵ>0∃δ>0:x∈Bδ(a)⟹f(x)∈Bϵ(f(a))
- ∀V∈K:f−1(V)∈J
Proof
⟹
Suppose f:(X,J)→(Y,K) is continuous.
Let V∈K - that is V is open within Y
Let x∈f−1(V) be given.
Then because V is open, ∃ϵ>0 such that Bϵ(f(x))⊂V (note that f(x)∈V by definition of where we choose x from).
But by continuity of f we know that ∃δ>0:a∈Bδ(x)⟹f(a)∈Bϵ(f(x))⊂V
Thus Bδ(x)⊂f−1(V) (as for all a in the ball, the thing f maps it to is in the ball of radius ϵ about f(x)).
Since x was arbitrary we have ∀x∈f−1(V)∃an open ball containing x⊂f−1(V), thus f−1(V) is open.
⟸
Choose any x∈X
Let ϵ>0 be given.
As Bϵ(f(x)) is an open set, the hypothesis implies that f−1(Bϵ(f(x))) is open in X
Since x∈f−1(Bϵ(f(x))) and f−1(Bϵ(f(x))) is open, it is a neighborhood to all of its points, that means
∃δ>0:Bδ(x)⊂f−1(Bϵ(f(x)))
Note: we have now shown that ∀ϵ>0∃δ>0:Bδ(x)⊂f−1(Bϵ(f(x)))
Using the implies and subset relation we see a∈Bδ(x)⟹a∈f−1(Bϵ(f(x))) which then ⟹f(a)∈Bϵ(f(x))
Or just a∈Bδ(x)⟹f(a)∈Bϵ(f(x)))
Thus it is continuous at x, since x was arbitrary, it is continuous.