31 Shuffling cards
Let \(\mathbb {Q} : \mathfrak {S}_n \longrightarrow \mathbb {R}\) be any probability distribution that defines a shuffling process \(\mathbb {Q}^*k\) with a strong uniform stopping rule whose stopping time is \(T\). Then for all \(k \geq 0\),
If \(X\) is a random variable with values in \(\mathfrak {S}_n\), with probability distribution \(\mathbb {Q}\), then we write \(\mathbb {Q}(S)\) for the probability that \(X\) takes a value in \(S \subseteq \mathfrak {S}_n\). Thus \(\mathbb {Q}(S) = \text{Prob}[X \in S]\), and in the case of the uniform distribution \(\mathbb {Q} = \mathbb {U}\) we get
For every subset \(S \subseteq \mathfrak {S}_n\), we get the probability that after \(k\) steps our deck is ordered according to a permutation in \(S\) as
This yields
since
is a difference of two probabilities, so it has absolute value at most 1.
Let \(c \geq 0\) and \(k := \lceil n \log n + cn \rceil \). Then after performing \(k\) top-in-at-random shuffles on a deck of \(n\) cards, the variation distance from the uniform distribution satisfies
TODO
After performing \(k\) riffle shuffles on a deck of \(n\) cards, the variation distance from a uniform distribution satisfies
TODO