๐Ÿงฎ Brain Teaser

The Weierstrass M-Test Isn't Needed: Uniform Convergence from Pointwise + Monotone

Let fn:[0,1]โ†’Rf_n : [0,1] \to \mathbb{R} be a sequence of continuous functions converging pointwise to a function f:[0,1]โ†’Rf : [0,1] \to \mathbb{R}, which is also continuous.

Suppose additionally that the convergence is monotone: for each xโˆˆ[0,1]x \in [0,1],

f1(x)โ‰ฅf2(x)โ‰ฅf3(x)โ‰ฅโ‹ฏโ‰ฅf(x).f_1(x) \geq f_2(x) \geq f_3(x) \geq \cdots \geq f(x).

Prove that fnโ†’ff_n \to f uniformly on [0,1][0,1].

This is Dini's theorem โ€” but prove it cleanly using a compactness argument, and appreciate why each hypothesis (continuity of limit, monotonicity, compactness of domain) is genuinely necessary.

uniform convergencecompactnessDini's theoremreal analysismonotone sequences

Answer: Dini's Theorem via Compactness

Key Idea / Intuition

The gap gn(x)=fn(x)โˆ’f(x)g_n(x) = f_n(x) - f(x) is a decreasing sequence of non-negative continuous functions converging pointwise to 00. The key is that on a compact set, you cannot have pointwise-but-not-uniform convergence to 00 for monotone decreasing continuous functions: any hypothetical "escape to non-zero" would force a subsequence of points whose limit contradicts pointwise convergence, via compactness.

The argument is clean: the superlevel sets {x:gn(x)โ‰ฅฮต}\{x : g_n(x) \geq \varepsilon\} are closed (by continuity of gng_n), nested (by monotonicity), and their intersection is empty (by pointwise convergence). Compactness forces one of them to already be empty.


Formal Proof / Solution

Setup. Define gn=fnโˆ’fg_n = f_n - f. Then:

  • Each gng_n is continuous (difference of continuous functions).
  • gn(x)โ‰ฅ0g_n(x) \geq 0 for all xx (since fnโ‰ฅff_n \geq f pointwise).
  • gn(x)โ†˜0g_n(x) \searrow 0 pointwise (monotone decreasing to 00).

We want to show supโกxโˆˆ[0,1]gn(x)โ†’0\sup_{x \in [0,1]} g_n(x) \to 0.

Fix ฮต>0\varepsilon > 0. Define the closed sets

Kn={xโˆˆ[0,1]:gn(x)โ‰ฅฮต}.K_n = \{ x \in [0,1] : g_n(x) \geq \varepsilon \}.

Each KnK_n is closed since gng_n is continuous.

The KnK_n are nested: since gn(x)โ‰ฅgn+1(x)g_n(x) \geq g_{n+1}(x) for all xx, we have Kn+1โІKnK_{n+1} \subseteq K_n.

The intersection is empty: if xโˆˆโ‹‚nKnx \in \bigcap_n K_n, then gn(x)โ‰ฅฮตg_n(x) \geq \varepsilon for all nn, contradicting gn(x)โ†’0g_n(x) \to 0.

So โ‹‚n=1โˆžKn=โˆ…\bigcap_{n=1}^\infty K_n = \emptyset.

Apply compactness: [0,1][0,1] is compact and the KnK_n are closed nested subsets with empty intersection. By the finite intersection property, there must exist some NN such that KN=โˆ…K_N = \emptyset.

This means: for all xโˆˆ[0,1]x \in [0,1], gN(x)<ฮตg_N(x) < \varepsilon.

Since gng_n is decreasing, for all nโ‰ฅNn \geq N and all xโˆˆ[0,1]x \in [0,1]:

0โ‰คgn(x)โ‰คgN(x)<ฮต.0 \leq g_n(x) \leq g_N(x) < \varepsilon.

Hence supโกxgn(x)<ฮต\sup_x g_n(x) < \varepsilon for all nโ‰ฅNn \geq N, i.e., fnโ†’ff_n \to f uniformly. โ– \blacksquare


Why Each Hypothesis Is Necessary

| Hypothesis dropped | Counterexample | |---|---| | Continuity of ff | fn(x)=xnf_n(x) = x^n on [0,1][0,1]: continuous, monotone decreasing, pointwise limit is discontinuous, convergence is not uniform | | Monotonicity | fn(x)=sinโกn(nx)f_n(x) = \sin^n(nx): pointwise โ†’0\to 0 but not uniformly | | Compactness | fn(x)=x/nf_n(x) = x/n on (0,โˆž)(0,\infty): all hypotheses hold but supโกgn=โˆž\sup g_n = \infty |

The compactness argument via nested closed sets is the heart of the proof โ€” it's a perfect illustration of how compactness "forces finiteness" even in a purely analytic statement.