3 %% A {\it coupling} with transition matrix $P$ is a process $(X_t,Y_t)_{t\geq 0}$
4 %% such that both $(X_t)$ and $(Y_t)$ are markov chains of matric $P$; moreover
5 %% it is required that if $X_s=Y_s$, then for any $t\geq s$, $X_t=Y_t$.
6 %% A results provides that if $(X_t,Y_t)_{t\geq 0}$ is a coupling, then
7 %% $$d(t)\leq \max_{x,y} P_{x,y}(\{\tau_{\rm couple} \geq t\}),$$
8 %% with $\tau_{\rm couple}=\min_t\{X_t=Y_t\}$.
11 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
12 \section{Random walk on the modified Hypercube}
15 Let $\Omega=\{0,1\}^N$ be the set of words of length $N$. Let $E=\{(x,y)\mid
16 x\in \Omega, y\in \Omega,\ x=y \text{ or } x\oplus y \in 0^*10^*\}$. Let $h$
17 be a function from $\Omega$ into $\{1,\ldots,N\}$.
19 We denote by $E_h$ the set $E\setminus\{(x,y)\mid x\oplus y =
20 0^{N-h(x)}10^{h(x)-1}\}$. We define the matrix $P_h$ has follows:
23 P_h(x,y)=0 & \text{ if } (x,y)\notin E_h\\
24 P_h(x,x)=\frac{1}{2}+\frac{1}{2N} & \\
25 P_h(x,x)=\frac{1}{2N} & \text{otherwise}\\
31 We denote by $\ov{h}$ the function from $\Omega$ into $\omega$ defined
32 by $x\oplus\ov{h}(x)=0^{N-h(x)}10^{h(x)-1}.$
33 The function $\ov{h}$ is said {\it square-free} if for every $x\in E$,
34 $\ov{h}(\ov{h}(x))\neq x$.
36 \begin{Lemma}\label{lm:h}
37 If $\ov{h}$ is bijective and square-free, then $h(\ov{h}^{-1}(x))\neq h(x)$.
44 Let $Z$ be a random variable over
45 $\{1,\ldots,N\}\times\{0,1\}$ uniformaly distributed. For $X\in \Omega$, we
46 define, with $Z=(i,x)$,
50 f(X,Z)=X\oplus (0^{N-i}10^{i-1}) & \text{if } x=1 \text{ and } i\neq h(X),\\
51 f(X,Z)=X& \text{otherwise.}
55 The pair $f,Z$ is a random mapping representation of $P_h$.
60 %%%%%%%%%%%%%%%%%%%%%%%%%%%ù
61 \section{Stopping time}
63 An integer $\ell\in\{1,\ldots,N\}$ is said {\it fair} at time $t$ if there
64 exists $0\leq j <t$ such that $Z_j=(\ell,\cdot)$ and $h(X_j)\neq \ell$.
67 Let $\ts$ be the first time all the elements of $\{1,\ldots,N\}$
68 are fair. The integer $\ts$ is a randomized stopping time for
69 the markov chain $(X_t)$.
73 The integer $\ts$ is a strong stationnary time.
77 Let $\tau_\ell$ be the first time that $\ell$ is fair. The random variable
78 $Z_{\tau_\ell-1}$ is of the form $(\ell,\delta)$ with $\delta\in\{0,1\}$ and
79 $\delta=1$ with probability $\frac{1}{2}$ and $\delta=0$ with probability
80 $\frac{1}{2}$. Since $h(X_{\tau_\ell-1})\neq\ell$ the value of the $\ell$-th
81 bit of $X_{\tau_\ell}$ is $\delta$. By symetry, for $t\geq \tau_\ell$, the
82 $\ell$-th bit of $X_t$ is $0$ or $1$ with the same probability, proving the
86 \begin{Theo} \label{prop:stop}
87 If $\ov{h}$ is bijective and square-free, then
88 $E[\ts]\leq 8N^2+ N\ln (N+1)$.
91 For each $x\in \Omega$ and $\ell\in\{1,\ldots,N\}$, let $S_{x,\ell}$ be the
92 random variable counting the number of steps done until reaching from $x$ a state where
93 $\ell$ is fair. More formaly
94 $$S_{x,\ell}=\min \{m \geq 1\mid h(X_m)\neq \ell\text{ and }Z_m=\ell\text{ and } X_0=x\}.$$
97 $$\lambda_h=\max_{x,\ell} S_{x,\ell}.$$
100 \begin{Lemma}\label{prop:lambda}
101 If $\ov{h}$ is a square-free bijective function, then one has $E[\lambda_h]\leq 8N^2.$
105 For evey $x$, every $\ell$, one has $\P(S_{x,\ell})\leq 2)\geq
106 \frac{1}{4N^2}$. Indeed, if $h(x)\neq \ell$, then
107 $\P(S_{x,\ell}=1)=\frac{1}{2N}\geq \frac{1}{4N^2}$. If $h(x)=\ell$, then
108 $\P(S_{x,\ell}=1)=0$. Let $X_0=x$. Since $\ov{h}$ is square-free,
109 $\ov{h}(\ov{h}^{-1}(x))\neq x$. It follows that $(x,\ov{h}^{-1}(x))\in E_h$.
110 Thefore $P(X_1=\ov{h}^{-1}(x))=\frac{1}{2N}$. Now,
111 by Lemma~\ref{lm:h}, $h(\ov{h}^{-1}(x))\neq h(x)$. Therefore
112 $\P(S_{x,\ell}=2\mid X_1=\ov{h}^{-1}(x))=\frac{1}{2N}$, proving that
113 $\P(S_{x,\ell}\leq 2)\geq \frac{1}{4N^2}$.
115 Therefore, $\P(S_{x,\ell}\geq 2)\leq 1-\frac{1}{4N^2}$. By induction, one
116 has, for every $i$, $\P(S_{x,\ell}\geq 2i)\leq
117 \left(1-\frac{1}{4N^2}\right)^i$.
119 since $S_{x,\ell}$ is positive, it is known~\cite[lemma 2.9]{}, that
120 $$E[S_{x,\ell}]=\sum_{i=1}^{+\infty}\P(S_{x,\ell}\geq i).$$
121 Since $\P(S_{x,\ell}\geq i)\geq \P(S_{x,\ell}\geq i+1)$, one has
122 $$E[S_{x,\ell}]=\sum_{i=1}^{+\infty}\P(S_{x,\ell}\geq i)\leq
123 \P(S_{x,\ell}\geq 1)+2 \sum_{i=1}^{+\infty}\P(S_{x,\ell}\geq 2i).$$
125 $$E[S_{x,\ell}]\leq 1+2
126 \sum_{i=1}^{+\infty}\left(1-\frac{1}{4N^2}\right)^i=1+2(4N^2-1)=8N^2-2,$$
127 which concludes the proof.
130 Let $\ts^\prime$ be the first time that there are exactly $N-1$ fair
133 \begin{Lemma}\label{lm:stopprime}
134 One has $E[\ts^\prime]\leq N \ln (N+1).$
138 This is a classical Coupon Collector's like problem. Let $W_i$ be the
139 random variable counting the number of moves done in the markov chain while
140 we had exactly $i-1$ fair bits. One has $\ts^\prime=\sum_{i=1}^{N-1}W_i$.
141 But when we are at position $x$ with $i-1$ fair bits, the probability of
142 obtaining a new fair bit is either $1-\frac{i-1}{N}$ if $h(x)$ is fair,
143 or $1-\frac{i-2}{N}$ if $h(x)$ is not fair. It follows that
144 $E[W_i]\leq \frac{N}{N-i+2}$. Therefore
145 $$E[\ts^\prime]=\sum_{i=1}^{N-1}E[W_i]\leq N\sum_{i=1}^{N-1}
146 \frac{1}{N-i+2}=N\sum_{i=3}^{N+1}\frac{1}{i}.$$
148 But $\sum_{i=1}^{N+1}\frac{1}{i}\leq 1+\ln(N+1)$. It follows that
149 $1+\frac{1}{2}+\sum_{i=3}^{N+1}\frac{1}{i}\leq 1+\ln(N+1).$
151 $E[\ts^\prime]\leq N (-\frac{1}{2}+\ln(N+1))\leq N\ln(N+1)$.
154 One can now prove Theo~\ref{prop:stop}.
157 One has $\ts\leq \ts^\prime+\lambda_h$. Therefore,
158 Theorem~\ref{prop:stop} is a direct application of
159 lemma~\ref{prop:lambda} and~\ref{lm:stopprime}.