Quantcast
Viewing latest article 8
Browse Latest Browse All 10

Short cycles in random permutations

Previously we showed that the distribution of fixed points of a random permutation of Image may be NSFW.
Clik here to view.
n
elements behaves asymptotically (in the limit as Image may be NSFW.
Clik here to view.
n \to \infty
) like a Poisson random variable with parameter Image may be NSFW.
Clik here to view.
\lambda = 1
. As it turns out, this generalizes to the following.

Theorem: As Image may be NSFW.
Clik here to view.
n \to \infty
, the number of cycles of length Image may be NSFW.
Clik here to view.
1, 2, ... k
of a random permutation of Image may be NSFW.
Clik here to view.
n
elements are asymptotically independent Poisson with parameters Image may be NSFW.
Clik here to view.
1, \frac{1}{2}, ... \frac{1}{k}
.

This is a fairly strong statement which essentially settles the asymptotic description of short cycles in random permutations.

Proof

We will prove pointwise convergence of moment generating functions. First, the Poisson random variable Image may be NSFW.
Clik here to view.
X_{\lambda}
with parameter Image may be NSFW.
Clik here to view.
\lambda
is the random variable which takes on non-negative integer values with probabilities

Image may be NSFW.
Clik here to view.
\displaystyle \mathbb{P}(X_{\lambda} = k) = e^{-\lambda} \frac{\lambda^k}{k!}
.

Image may be NSFW.
Clik here to view.
X_{\lambda}
therefore has moment generating function

Image may be NSFW.
Clik here to view.
\displaystyle \mathbb{E}(e^{t X_{\lambda}}) = e^{-\lambda} \sum_{k \ge 0} \frac{e^{tk} \lambda^k}{k!} = e^{\lambda (e^t - 1)}

which is the exponential generating function of the Touchard polynomials

Image may be NSFW.
Clik here to view.
\displaystyle T_n(\lambda) = \sum_{k=0}^n S(n, k) \lambda^k

where Image may be NSFW.
Clik here to view.
S(n, k)
are the Stirling numbers of the second kind. These specialize to the Bell numbers when Image may be NSFW.
Clik here to view.
\lambda = 1
as expected.

Because we are discussing Image may be NSFW.
Clik here to view.
k
random variables, we should compute a joint moment generating function. The joint moment generating function of Image may be NSFW.
Clik here to view.
k
independent Poisson random variables with parameters Image may be NSFW.
Clik here to view.
\lambda_1, ... \lambda_k
is

Image may be NSFW.
Clik here to view.
\displaystyle \mathbb{E}(\exp \left( t_1 X_{\lambda_1} + ... + t_k X_{\lambda_k} \right) ) = \exp \left( \lambda_1 (e^{t_1} - 1) + ... + \lambda_k (e^{t_k} - 1) \right)
.

Back to permutations. By the exponential formula, letting Image may be NSFW.
Clik here to view.
c_k(\sigma)
denote the number of cycles of length Image may be NSFW.
Clik here to view.
k
in a permutation Image may be NSFW.
Clik here to view.
\sigma
, we have

Image may be NSFW.
Clik here to view.
\displaystyle \sum_{n \ge 0} \frac{t^n}{n!} \sum_{\sigma \in S_n} z_1^{c_1(\sigma)} ... z_k^{c_k(\sigma)} = \exp \left( z_1 t + ... + z_k \frac{t^k}{k} + \frac{t^{k+1}}{k+1} + ... \right)

which simplifies to

Image may be NSFW.
Clik here to view.
\displaystyle \frac{1}{1 - t} \exp \left( (z_1 - 1) t + ... + (z_k - 1) \frac{t^k}{k} \right)
.

It is a general and straightforward observation that if Image may be NSFW.
Clik here to view.
f(t)
is a power series with radius of convergence greater than Image may be NSFW.
Clik here to view.
1
, then Image may be NSFW.
Clik here to view.
\frac{f(t)}{1 - t}
has a power series expansion whose coefficients asymptotically approach Image may be NSFW.
Clik here to view.
f(1)
. Substituting Image may be NSFW.
Clik here to view.
z_i = e^{t_i}
, we conclude that the coefficients of

Image may be NSFW.
Clik here to view.
\displaystyle \sum_{n \ge 0} \frac{t^n}{n!} \sum_{\sigma \in S_n} e^{t_1 c_1(\sigma)} ... e^{t_k c_k(\sigma)}

asymptotically approach

Image may be NSFW.
Clik here to view.
\displaystyle \exp \left( (e^{t_1} - 1) + ... + (e^{t_k} - 1) \frac{1}{k} \right)
.

But the former is precisely the joint moment generating function of Image may be NSFW.
Clik here to view.
c_1, ... c_k
, and the latter is precisely the joint moment generating function of independent Poisson random variables with parameters Image may be NSFW.
Clik here to view.
1, ... \frac{1}{k}
. The conclusion follows.

Mean and variance

An exact result which can be deduced using the above methods is that the expected number of Image may be NSFW.
Clik here to view.
k
-cycles of a random permutation of Image may be NSFW.
Clik here to view.
n
elements is Image may be NSFW.
Clik here to view.
\frac{1}{k}
if Image may be NSFW.
Clik here to view.
k \le n
and Image may be NSFW.
Clik here to view.
0
otherwise. It follows that the total expected number of cycles is the harmonic number

Image may be NSFW.
Clik here to view.
\displaystyle H_n = 1 + \frac{1}{2} + ... + \frac{1}{n} \sim \log n
.

Since Poisson random variables have the same mean and variance, and by the asymptotic independence statement above, we expect the variance of the total number of cycles to also be asymptotic to Image may be NSFW.
Clik here to view.
\log n
. This is in fact true, and can be shown using the exponential formula as above.

In The Anatomy of Integers and Permutations, Granville suggested that the decomposition of a random permutation into cycles should be thought of as analogous to the decomposition of a random integer into prime factors. In light of this analogy, the above computation should be thought of as roughly analogous to the Hardy-Ramanujan theorem.


Viewing latest article 8
Browse Latest Browse All 10

Trending Articles