Skip to content
Tutorials Center
My learning

Topics

Lesson 33 of 37 · Sequences and Series of Functions

Power Series and the Weierstrass M-Test

Prove the Weierstrass M-test for uniform convergence of series, compute radii of convergence with the Cauchy-Hadamard formula, and differentiate power series term by term.

A series of functions Σ gₖ(x) converges (pointwise or uniformly) when its sequence of partial sums Sₙ = g₁ + ... + gₙ does. Checking the sup norm of the partial sums is awkward; the Weierstrass M-test replaces it with a comparison to a series of numbers. Power series are the most important application, because inside their interval of convergence they behave almost like polynomials.

The Weierstrass M-test

Theorem. Let gₖ : D → ℝ and suppose there are constants Mₖ ≥ 0 with |gₖ(x)| ≤ Mₖ for all x ∈ D and Σ Mₖ converges. Then Σ gₖ converges uniformly (and absolutely) on D.

Proof. For each x, Σ |gₖ(x)| converges by comparison, so the series converges pointwise to some S(x). For the tail,

Text
| S(x) - Sₙ(x) | = | Σ_{k > n} gₖ(x) | ≤ Σ_{k > n} Mₖ,

and the right side does not depend on x and tends to 0 since Σ Mₖ converges. Hence ‖S - Sₙ‖∞ → 0. ∎

Example. Σ sin(k² x)/k² on all of ℝ: |sin(k² x)/k²| ≤ 1/k² and Σ 1/k² converges, so the series converges uniformly. Each partial sum is continuous, hence by the uniform limit theorem the sum is continuous on ℝ.

Python The M-test tail bound in action

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

Program input · 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 reference uses 20000 terms, so its own error is below 5·10⁻⁵; the comparison is meaningful for the larger errors in the table.

Radius of convergence

A power series is Σ aₖ (x - c)ᵏ.

Theorem (Cauchy-Hadamard). Let R = 1 / limsup |aₖ|^{1/k} (with R = ∞ if the limsup is 0, and R = 0 if it is ∞). Then the series converges absolutely for |x - c| < R and diverges for |x - c| > R. At |x - c| = R anything can happen.

When the limit exists, R = lim |aₖ/aₖ₊₁| (ratio form). Examples (centered at 0):

Series R Behavior at x = ±R
Σ xᵏ 1 diverges at both
Σ xᵏ/k 1 converges at -1, diverges at 1
Σ xᵏ/k² 1 converges at both
Σ xᵏ/k! ∞ always converges
Σ k! xᵏ 0 converges only at 0

Uniform convergence inside. If 0 < r < R, the series converges uniformly on [c - r, c + r]: pick r < ρ < R; then |aₖ| ρᵏ is bounded, so |aₖ| rᵏ ≤ C (r/ρ)ᵏ, a convergent geometric bound, and the M-test applies. Consequently the sum f(x) is continuous on (c - R, c + R).

Python Estimate the radius with the root and ratio formulas

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

Program input · 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.

For Σ xᵏ/k! both estimates grow with k, consistent with R = ∞.

Term-by-term differentiation and integration

Theorem. If f(x) = Σ aₖ (x - c)ᵏ has radius R > 0, then f is differentiable on (c - R, c + R) with

Text
f'(x) = Σ_{k ≥ 1} k aₖ (x - c)^{k-1},

and the differentiated series has the same radius R (since k^{1/k} → 1). Likewise ∫_c^x f = Σ aₖ (x - c)^{k+1}/(k+1). The proof combines uniform convergence on [c - r, c + r] with the derivative theorem from the previous lesson. In particular, f has derivatives of all orders and aₖ = f^{(k)}(c)/k!: the series is its own Taylor series.

Example. From Σ xᵏ = 1/(1 - x) for |x| < 1, integrating termwise gives -ln(1 - x) = Σ_{k≥1} xᵏ/k, and differentiating gives 1/(1-x)² = Σ k x^{k-1}.

Python Check termwise operations on the geometric series

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

Program input · 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.

Check your understanding

Quick check · 4 questions

of 4 answered You scored / 4 · perfect!

  1. 1. The Weierstrass M-test needs which condition?

    A bound independent of x whose series converges gives uniform convergence by comparison.
  2. 2. The radius of convergence of the sum of x^k / k! is:

    |a_k / a_(k+1)| = k + 1 grows without bound.
  3. 3. A power series with radius R converges uniformly on:

    Uniform convergence holds on compact sub-intervals inside the interval of convergence.
  4. 4. Differentiating a power series with radius R term by term gives a series with radius:

    The factor k does not change the limsup of |a_k|^(1/k), since k^(1/k) tends to 1.

Summary

  • Weierstrass M-test: if |gₖ| ≤ Mₖ on D and Σ Mₖ < ∞, then Σ gₖ converges uniformly on D.
  • Cauchy-Hadamard: R = 1/limsup |aₖ|^{1/k}; absolute convergence inside, divergence outside, endpoints must be checked separately.
  • Inside the radius, power series converge uniformly on compact subintervals, are continuous, and can be differentiated and integrated term by term.
  • Combined with Taylor's theorem, this explains when a function equals its Taylor series.

Next: Practice: Continuity and Differentiation