Skip to content
Tutorials Center
My learning

Topics

Lesson 31 of 37 ยท Sequences and Series of Functions

Pointwise and Uniform Convergence

Compare pointwise and uniform convergence of sequences of functions, use the sup norm to test uniformity, and explore classic examples with runnable code.

So far we have studied sequences of numbers. Now each term is a function: fโ‚, fโ‚‚, fโ‚ƒ, ... all defined on the same set D. What should "fโ‚™ converges to f" mean? There are several answers, and the two most important are pointwise and uniform convergence. The difference decides whether properties like continuity survive in the limit.

Pointwise convergence

fโ‚™ โ†’ f pointwise on D if for each fixed x โˆˆ D, the number sequence fโ‚™(x) converges to f(x). In quantifiers:

Text
for every x โˆˆ D and every ฮต > 0 there exists N (which may depend on x) such that
n โ‰ฅ N implies |fโ‚™(x) - f(x)| < ฮต.

Example. fโ‚™(x) = xโฟ on [0, 1]. For 0 โ‰ค x < 1, xโฟ โ†’ 0. At x = 1, 1โฟ = 1. So the pointwise limit is

Text
f(x) = 0 for 0 โ‰ค x < 1,    f(1) = 1.

Each fโ‚™ is continuous, but the limit f is not. Pointwise convergence did not preserve continuity.

Uniform convergence

fโ‚™ โ†’ f uniformly on D if

Text
for every ฮต > 0 there exists N (the same for all x โˆˆ D) such that
n โ‰ฅ N implies |fโ‚™(x) - f(x)| < ฮต for all x โˆˆ D.

Compare with the definition of uniform continuity: again the only difference is the order of "there exists" and "for all x". Define the supremum norm โ€–gโ€–โˆž = sup_{x โˆˆ D} |g(x)|. Then

Text
fโ‚™ โ†’ f uniformly on D   โ‡”   โ€–fโ‚™ - fโ€–โˆž โ†’ 0.

Geometrically: for large n, the entire graph of fโ‚™ lies in the band of width 2ฮต around the graph of f. This makes the sup norm the practical test: compute Mโ‚™ = sup |fโ‚™ - f| and check whether Mโ‚™ โ†’ 0.

Uniform convergence implies pointwise convergence to the same limit, but not conversely.

Worked tests

1. xโฟ on [0, 1). Here f = 0 and sup_{x โˆˆ [0,1)} xโฟ = 1 (approached as x โ†’ 1), so Mโ‚™ = 1 does not tend to 0: not uniform. On [0, r] with r < 1, sup xโฟ = rโฟ โ†’ 0: uniform on that smaller interval.

2. fโ‚™(x) = x/(1 + n xยฒ) on โ„. Pointwise limit 0. For x โ‰ฅ 0, calculus shows the maximum of x/(1 + n xยฒ) is at x = 1/โˆšn with value 1/(2โˆšn). By oddness โ€–fโ‚™โ€–โˆž = 1/(2โˆšn) โ†’ 0: uniform.

3. fโ‚™(x) = sin(nx)/n on โ„. |fโ‚™| โ‰ค 1/n, so โ€–fโ‚™โ€–โˆž โ‰ค 1/n โ†’ 0: uniform limit 0.

Python Estimate the sup norm numerically

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The first block illustrates why the grid cannot decide uniformity by itself: it approaches 1 as the grid gets closer to 1, matching the true supremum.

A picture of the failure

The graphs of xโฟ on [0, 1] hug 0 for most x, then shoot up sharply near 1. No matter how large n, there is always a point where the graph is far above the limit function: the "bump" just moves toward 1.

Python Graphs of x^n and their pointwise limit

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Uniform Cauchy criterion

(fโ‚™) converges uniformly on D iff for every ฮต > 0 there is N such that m, n โ‰ฅ N implies โ€–fโ‚˜ - fโ‚™โ€–โˆž < ฮต. This lets you prove uniform convergence without knowing the limit. It follows from the fact that โ„ is complete: for each x, fโ‚™(x) is a Cauchy sequence hence converges to some f(x); letting m โ†’ โˆž in |fโ‚˜(x) - fโ‚™(x)| < ฮต gives |f(x) - fโ‚™(x)| โ‰ค ฮต for all x.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. What is the pointwise limit of x^n on [0, 1]?

    x^n tends to 0 when 0 <= x < 1 but stays 1 at x = 1.
  2. 2. f_n tends to f uniformly on D exactly when:

    The sup norm of the difference measures the worst-case error over all x.
  3. 3. Does x^n converge uniformly to 0 on [0, 0.9]?

    On [0, 0.9] the largest value of x^n is 0.9^n, which tends to 0.
  4. 4. For f_n(x) = sin(nx)/n on the real line, which is true?

    The bound 1/n does not depend on x, so the sup norm tends to 0.

Summary

  • Pointwise: N may depend on x. Uniform: one N works for all x.
  • fโ‚™ โ†’ f uniformly iff โ€–fโ‚™ - fโ€–โˆž = sup |fโ‚™ - f| โ†’ 0; this is the practical test.
  • xโฟ converges pointwise but not uniformly on [0, 1], and the limit is discontinuous even though each xโฟ is continuous.
  • The uniform Cauchy criterion proves uniform convergence without naming the limit.

Next: Interchanging Limits