Tutorial 8
Tutorial 8
Tutorial 8
Sequence of Functions
Definition (c.f. Definition 8.1.1). Let (fn ) be a sequence of functions defined on A ⊆ R and
let f be a function defined on A. (fn ) is said to converge (pointwisely) to f on A if (fn (x))
converges to f (x) for each x ∈ A. In this case, we denote
f (x) = lim fn (x) or f = lim fn .
n→∞ n→∞
Now we show that (fn ) converges uniformly to the zero function. We need to find the
maximum value of fn (x) = xe−nx on [0, ∞) for each n. Differentiating fn gives
fn′ (x) = (1 − nx)e−nx .
Hence it has only one critical point at x = 1/n. Now at the endpoints and critical point,
1
fn (0) = 0, fn (1/n) = and lim fn (x) = 0.
ne x→∞
The following theorem is an analogue of the Cauchy Criterion for sequence of functions.
Example 2 (c.f. Section 8.1, Ex.4 & Ex.14). Let fn : [0, ∞) → R be defined by
xn
fn (x) = .
1 + xn
Let 0 < b < 1. Show that (fn ) converges uniformly on [0, b] but not uniformly on [0, 1].
Let’s show that (fn ) converges uniformly to the zero function on [0, b]. Note that
xn xn bn
− 0 = ≤ = bn , ∀n ∈ N, ∀x ∈ [0, b].
1 + xn 1 + xn 1 + 0n
Let ε > 0. Since 0 < b < 1, bn → 0 as n → ∞. Hence we can choose N ∈ N such that bn < ε
whenever n ≥ N . It follows that
xn
− 0 ≤ bn < ε, ∀n ≥ N, ∀x ∈ [0, b].
1 + xn
To see that (fn ) does not converge uniformly on [0, 1], we need to show that there exist ε > 0
such that whenever N ∈ N, there exists n ≥ N and x ∈ [0, 1] such that
xn 1/2 1
|fn (x) − f (x)| = n
−0 = = .
1+x 1 + 1/2 3
Remark. Due to the proposition below, it suffices to show that (fn ) does not converge
uniform to its pointwise limit f .
Observe that in Example 2, the argument for non-uniform convergence is similar to the
lemma below.
Lemma (c.f. Lemma 8.1.5). Let (fn ) be a sequence of real-valued functions defined on A ⊆ R
and let f : A → R be a function. Then (fn ) does not converge uniformly to f on A if and
only if there exists ε > 0, a subsequence (fnk ) of (fn ) and a sequence (xk ) in A such that
Proof. (⇒) Suppose fn does not converge uniformly to f on A. By definition, there exists
ε > 0 such that whenever N ∈ N, there exists n ≥ N and x ∈ A such that
• Similarly for k = 3, 4, ..., consider N = nk−1 + 1. Then there exists nk > nk−1 and
xk ∈ A such that
|fnk (xk ) − f (xk )| ≥ ε.
The implication is then clear from the construction. (⇐) We need to show that there exists
ε > 0 such that whenever N ∈ N, there exists n ≥ N and x ∈ A such that
Obviously, we need to take ε > 0 as in the assumption. For each N ∈ N, we can choose
k ∈ N such that nk ≥ N . Taking n = nk and x = xk , we have n ≥ N , x ∈ A and
Example 3 (c.f. Section 8.1, Ex.24). Let (fn ) be a sequence of functions that converges
uniformly to f on A and that satisfies |fn (x)| ≤ M for all n ∈ N and all x ∈ A. If g
is a continuous function defined on the interval [−M, M ], show that the sequence (g ◦ fn )
converges uniformly to a well-defined function g ◦ f on A.
Solution. To see that the function g ◦ f is well-defined, i.e., f (x) ∈ [−M, M ] for all x ∈ A,
we can simply take limits on both sides of the inequality. We then need to show that for all
ε > 0, there exists N ∈ N such that
Hence whenever n ≥ N and x ∈ A, take u = fn (x) and v = f (x). Then we have |u − v| < δ
and u, v ∈ [−M, M ]. Therefore