Series of functions

A series of functions on a set D is an expression

n=1fn(x),xD,

understood via its partial sums

SN(x)=n=1Nfn(x).

The series converges (to S) in some mode if SNS in that mode.

Modes of convergence

Pointwise convergence

SNS pointwise on D if for every xD we have SN(x)S(x).

Uniform convergence

SNS uniformly on D means the convergence is uniform in x (so the same N0 works for all xD).

This is the main mode of convergence used to justify swapping limits with operations.
See: uniform convergence (definition, Cauchy criterion, M-test, and standard consequences).

Canonical examples

Power series

A power series

n=0an(zz0)n

converges uniformly on compact subsets of its disk of convergence D(z0,R). This is the key analytic reason term-by-term differentiation/integration works inside the radius.

Radius of convergence

The radius of convergence R[0,] characterizes the disk |zz0|<R where the series converges absolutely. Within this disk, the convergence is uniform on compact sets. Outside, for |zz0|>R, the series diverges.

The radius R is determined by the following criteria:

  1. Cauchy-Hadamard Formula: This is the general definition:1R=lim supn|an|n
  2. Ratio Test (D'Alembert criterion): If the following limit exists (or is infinite):R=limn|anan+1|

On the boundary |zz0|=R, the series may converge or diverge; this behavior must be studied case-by-case.

See: analytic function and Taylor's theorem

Laurent series

A Laurent series in an annulus r1<|zz0|<r2 converges absolutely in the annulus and uniformly on compact subannuli.
See: Laurent series.

Fourier series

Fourier series are often discussed in norms like L2:

SNfL20,

which is weaker than uniform convergence. As a result, pointwise/uniform statements and term-by-term manipulations generally need extra hypotheses.
See: Fourier series expansion and isomorphism between l_2 and L_2.

See also