Of Notes

Characteristic Function

2 min read · revised 15 November 2026 · referred to by 3 notes · ·

For any random variable we define as the characteristic function:

which for a continuous PDF is

Unlike the MGF or CGF, this one guarantees existence and convergence: bounds the integrand by , so for every , with .

Fourier Property

Looks like a Fourier transform ? Because it is: , the transform of the density with the sign of flipped. So we can leverage the derivative identity of the transform,

which with the flipped sign becomes a factor for . That is what the moments below use.

What we do by taking the fourier transform is ask: How do different frequency waves interact with the distribution where the exponential term represents a complex wave.

A very high frequency wave can pass close to a low frequency one at particular points, but that is not enough for them to give the same value: integrates the wave over every weighted by , so what matters is how much each wave cancels against the distribution as a whole.

Left: over \(0 \le x \le 10\pi\) the fast wave \(0.1\sin(100x)+1\) oscillates too quickly to resolve and fills the band \([0.9, 1.1]\), while the slow wave \(\sin(0.1x)\) only enters it near its peak at \(x = 5\pi\). Right: zoomed in there, the two cross again and again: they agree at points, yet as waves they are nothing alike.
Fig. 1. Left: over the fast wave oscillates too quickly to resolve and fills the band , while the slow wave only enters it near its peak at . Right: zoomed in there, the two cross again and again: they agree at points, yet as waves they are nothing alike.

The same thing happens with the magnitude itself: two different frequencies can give . Take with probability each, then

which repeats, so all give magnitude . Equal magnitudes still need not mean equal values, since the phases can differ, and it is the whole complex-valued function across all , not any single magnitude, that pins down the distribution.

Computing Moments

Instead of calculating multiple complicated integrals we use the derivative to generate moments, the same trick as for the MGF. Differentiating under the expectation brings down a factor :

and this can be done times, so generally:

Evaluated at the exponential is and only the moment is left:

/ or Ctrl K
search; empty, it shows notes to revise and lately read
m
mark this note to revise, or unmark it
c
cover maths, proofs and solutions to test yourself
n
uncover the next covered item
y
copy this note’s org-roam link
e
copy this note as markdown, LaTeX kept as TeX
p
the next problem to work; there h hints, s shows the solution, x passes it for today
→ ← j k space
turn the page, when reading in pages
l
on the map, trace the greatest distance, then the longest path
double-click
a formula, to copy its TeX
rest on maths
bind a part of a formula and open the wheel of notes it comes from; scroll or ↑ ↓ widens it, 1–6 goes
?
this card