Sequences and Series of Functions
pointwise convergence, uniform convergence, Weierstrass M-test, function sequences, power series
1 Role
This page is where real analysis stops studying one function at a time and starts studying whole sequences and series of functions.
That shift matters because many powerful mathematical objects are built exactly this way:
- polynomial approximations
- power series
- iterative approximations
- limit-defined functions
- smoothing and denoising procedures
2 First-Pass Promise
Read this page after Continuity, Compactness, and Completeness.
If you stop here, you should still understand:
- the difference between pointwise and uniform convergence
- why uniform convergence is much stronger and much more useful
- what the Weierstrass M-test is trying to guarantee
- why limit-interchange questions are central in analysis
3 Why It Matters
If a sequence of numbers converges, that is already meaningful.
If a sequence of functions converges, the next question is immediately:
converges in what sense?
That question matters because the wrong notion of convergence can destroy structure.
A pointwise limit of continuous functions need not be continuous. A uniformly convergent sequence of continuous functions, however, does preserve continuity.
So this page is really about a major analysis lesson:
the mode of convergence controls which properties survive in the limit
That lesson later reappears in:
- approximation theory
- Fourier and power series
- optimization and numerical analysis
- probability on function spaces
- diffusion and continuous-model arguments
4 Prerequisite Recall
- continuity preserves limits of points
- compactness and completeness are structural conditions behind existence and stability
- a numerical series is built from partial sums; a functional series works the same way, but each partial sum is itself a function
5 Intuition
Suppose you have functions \(f_n(x)\) defined on a set \(A\).
5.1 Pointwise convergence
For each fixed \(x\), the numbers \(f_n(x)\) converge to some limit \(f(x)\).
This lets the speed of convergence depend on the point \(x\).
5.2 Uniform convergence
The same stage \(N\) works simultaneously for all \(x \in A\).
This is much stronger. It says the entire graph of \(f_n\) gets close to the graph of \(f\), not just one point at a time.
That is why uniform convergence is often the right tool when you want continuity, integration, or approximation theorems to survive the limiting process.
6 Formal Core
Definition 1 (Definition: Pointwise Convergence) Let \(f_n : A \to \mathbb{R}\) and let \(f : A \to \mathbb{R}\).
We say \(f_n \to f\) pointwise on \(A\) if for every \(x \in A\) and every \(\varepsilon > 0\), there exists \(N\) such that for all \(n \ge N\),
\[ |f_n(x) - f(x)| < \varepsilon. \]
The important detail is that \(N\) may depend on both \(\varepsilon\) and \(x\).
Definition 2 (Definition: Uniform Convergence) We say \(f_n \to f\) uniformly on \(A\) if for every \(\varepsilon > 0\), there exists \(N\) such that for all \(n \ge N\) and all \(x \in A\),
\[ |f_n(x) - f(x)| < \varepsilon. \]
Now the same \(N\) must work for every point in the domain.
Theorem 1 (Theorem: Uniform Limits Preserve Continuity) If each \(f_n\) is continuous on \(A\) and \(f_n \to f\) uniformly on \(A\), then \(f\) is continuous on \(A\).
This theorem is one of the main reasons uniform convergence matters so much.
Theorem 2 (Theorem: Weierstrass M-Test) Suppose \(f_n : A \to \mathbb{R}\) satisfy
\[ |f_n(x)| \le M_n \qquad \text{for all } x \in A \]
and the numerical series
\[ \sum_{n=1}^\infty M_n \]
converges.
Then the function series
\[ \sum_{n=1}^\infty f_n(x) \]
converges uniformly on \(A\).
This is the cleanest first-pass tool for proving uniform convergence of a series of functions.
7 Worked Example
Consider the sequence of functions on \([0,1]\):
\[ f_n(x) = x^n. \]
For each fixed \(x \in [0,1)\), we have
\[ x^n \to 0. \]
At the point \(x=1\), we have
\[ f_n(1)=1. \]
So the pointwise limit is
\[ f(x)= \begin{cases} 0, & 0 \le x < 1, \\ 1, & x=1. \end{cases} \]
This limit function is not continuous at \(x=1\).
But every \(f_n(x)=x^n\) is continuous on \([0,1]\).
Therefore the convergence cannot be uniform.
This example is one of the most important first-pass warnings in analysis:
pointwise convergence is often too weak to preserve continuity.
Now compare that with a simple uniform-convergence example:
\[ g_n(x) = \frac{x}{n} \qquad \text{on } [0,1]. \]
Then for every \(x \in [0,1]\),
\[ \left|g_n(x)-0\right| = \frac{x}{n} \le \frac{1}{n}. \]
Given \(\varepsilon > 0\), choose \(N\) so that
\[ \frac{1}{N} < \varepsilon. \]
Then for all \(n \ge N\) and all \(x \in [0,1]\),
\[ \left|g_n(x)\right| \le \frac{1}{n} \le \frac{1}{N} < \varepsilon. \]
So \(g_n \to 0\) uniformly on \([0,1]\).
Here the same threshold works for the whole interval, and the limit function remains continuous.
8 Computation Lens
In numerical work and approximation, the practical question is often:
Do I have one error tolerance that works on the whole domain, or only point by point?
That is the pointwise/uniform distinction in disguise.
Uniform convergence is the more robust notion when you want:
- approximation quality across an interval
- stable interchange of limit and continuity
- certified truncation error bounds for series expansions
The M-test is especially useful because it reduces a functional convergence question to an ordinary numerical series.
9 Application Lens
9.1 Approximation And Power Series
Taylor and power series are only useful because convergence questions can be asked in a structured way. Uniform convergence helps explain when finite truncations are trustworthy on a whole region.
9.2 Optimization And Continuous Models
When objective functions, gradients, kernels, or value functions are approximated by sequences, the mode of convergence controls what properties survive in the limit.
9.3 Probability And Generative Modeling
Later probability and diffusion-style arguments often depend on careful interchange of limits, expectations, integrals, or continuous transformations. This page is the first clean warning that such interchanges are never automatic.
10 Stop Here For First Pass
If you can now explain:
- why pointwise convergence allows the threshold to depend on the point
- why uniform convergence uses one threshold for the whole domain
- why continuity can fail under pointwise limits
- why the M-test is a bridge from numerical series to function-series control
then this page has done its first-pass job.
11 Go Deeper
The next natural module step is:
For now, the best live neighboring pages are:
12 Sources and Further Reading
- MIT 18.100A Introduction to Analysis -
First pass- official analysis course page whose later lectures cover pointwise and uniform convergence and their consequences. Checked2026-04-25. - MIT 18.100A Fall 2020 Full Lecture Notes -
First pass- official notes with a clean arc through pointwise convergence, uniform convergence, the M-test, and interchange results. Checked2026-04-25. - Basic Analysis I by Jiří Lebl -
Second pass- open textbook with a careful treatment of sequences of functions and power-series examples. Checked2026-04-25.