Polymer dynamics in the depinned phase: metastability with logarithmic barriers (Q714951)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Polymer dynamics in the depinned phase: metastability with logarithmic barriers
scientific article

    Statements

    Polymer dynamics in the depinned phase: metastability with logarithmic barriers (English)
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    12 October 2012
    0 references
    Consider the Markov process \(\{\eta^L_t, t\geq 0\}\) with space state \[ \Omega_L=\{\eta=(\eta_{-L},\dotsc, \eta_{L})\in\mathbb{Z}^{2L+1} : \eta_{\pm L}=0, \;|\eta_{x+1}-\eta_x|=1,\;x\neq \pm L\},\quad L\in \mathbb{N}, \] and a spin-flip dynamics such that there is repulsion near \(\eta=0\), i.e., if, at the update instant for the site \(x\), \(\eta_{x-1}=\eta_{x+1}=h\in \{-1,1\}\), then \(\eta_x\) becomes \(0\) with probability \(\frac{\lambda}{1+\lambda}\) and \(2h\) with probability \(\frac{1}{1+\lambda}\), \(\lambda <1\), while, if \(h\notin \{-1,1\}\), then the two possible update values are equiprobable (the independent Poisson clocks on sites \(x\neq \pm L\) have rate 1). Denote by \(T_{m,L}\) its mixing time, \(T_{r,L}\) the inverse of its spectral gap, \(\operatorname{P}_{t,L}(\eta,\cdot)\) its distribution at time \(t\) and \(\operatorname{P}^\eta_L\) its law with initial state \(\eta\), \(\pi_L\) its stationary measure, \(\Omega^+_L=-\Omega^{-}_L\) the subset of \(\Omega_L\) such that \(\eta_x>0\) for all \(x\) except in a suitable neighbourhood of \(\pm L\). The authors study, using in many estimates the coupling method, the \(L\gg 1\) behaviour of the process, showing that \[ \limsup_{L\to\infty}\frac{\log T_{m,L} }{\log L}\leq \frac 52 +2, \] and that, in the considered asymptotics, (i) there exist two sets \(S^\pm\), \(S^+=-S^{-} \subset \Omega^+_{L}\), such that \(\pi_L(S^\pm)=\frac 12 +o(1)\) and, for \(\eta\in S^\pm\), \[ \operatorname{P}^\eta_L (\inf\{\tau\geq 0 :\eta^L_\tau\in S^{\mp}\}>t) = e^{-\frac t{2T_{r,L}}} + o(1); \] (ii) both \(\operatorname{P}_{t,L}(\eta_M, \cdot)\) and \(\sum_\eta \pi_L(\eta|\Omega^{+}_L) \operatorname{P}_{t,L}(\eta,\cdot)\) approach in total variation distance \[ \frac{1+e^{-\frac t{T_{r,L}}}}2 \pi_L(\cdot|\Omega^{+}_L) + \frac{1-e^{-\frac t{T_{r,L}}}}2 \pi_L(\cdot|\Omega^{-}_L), \] where \(\eta_M=(0,1,2,\dotsc, L-1,L,L-1,\dotsc,2,1,0)\).
    0 references
    0 references
    0 references
    reversible Markov processes
    0 references
    polymer pinning model
    0 references
    metastability
    0 references
    quasi-stationary distribution
    0 references
    0 references
    0 references