-
Random Forests and Networks Analysis
Authors:
L. Avena,
F. Castell,
A. Gaudilliere,
C. Melot
Abstract:
D. Wilson~\cite{[Wi]} in the 1990's described a simple and efficient algorithm based on loop-erased random walks to sample uniform spanning trees and more generally weighted trees or forests spanning a given graph. This algorithm provides a powerful tool in analyzing structures on networks and along this line of thinking, in recent works~\cite{AG1,AG2,ACGM1,ACGM2} we focused on applications of spa…
▽ More
D. Wilson~\cite{[Wi]} in the 1990's described a simple and efficient algorithm based on loop-erased random walks to sample uniform spanning trees and more generally weighted trees or forests spanning a given graph. This algorithm provides a powerful tool in analyzing structures on networks and along this line of thinking, in recent works~\cite{AG1,AG2,ACGM1,ACGM2} we focused on applications of spanning rooted forests on finite graphs. The resulting main conclusions are reviewed in this paper by collecting related theorems, algorithms, heuristics and numerical experiments. A first foundational part on determinantal structures and efficient sampling procedures is followed by four main applications: 1) a random-walk-based notion of well-distributed points in a graph 2) how to describe metastable dynamics in finite settings by means of Markov intertwining dualities 3) coarse graining schemes for networks and associated processes 4) wavelets-like pyramidal algorithms for graph signals.
△ Less
Submitted 5 November, 2017;
originally announced November 2017.
-
Intertwining wavelets or Multiresolution analysis on graphs through random forests
Authors:
Luca Avena,
Fabienne Castell,
Alexandre Gaudillière,
Clothilde Mélot
Abstract:
We propose a new method for performing multiscale analysis of functions defined on the vertices of a finite connected weighted graph. Our approach relies on a random spanning forest to downsample the set of vertices, and on approximate solutions of Markov intertwining relation to provide a subgraph structure and a filter bank leading to a wavelet basis of the set of functions. Our construction inv…
▽ More
We propose a new method for performing multiscale analysis of functions defined on the vertices of a finite connected weighted graph. Our approach relies on a random spanning forest to downsample the set of vertices, and on approximate solutions of Markov intertwining relation to provide a subgraph structure and a filter bank leading to a wavelet basis of the set of functions. Our construction involves two parameters q and q'. The first one controls the mean number of kept vertices in the downsampling, while the second one is a tuning parameter between space localization and frequency localization. We provide an explicit reconstruction formula, bounds on the reconstruction operator norm and on the error in the intertwining relation, and a Jackson-like inequality. These bounds lead to recommend a way to choose the parameters q and q'. We illustrate the method by numerical experiments.
△ Less
Submitted 1 May, 2018; v1 submitted 14 July, 2017;
originally announced July 2017.
-
Approximate and exact solutions of intertwining equations through random spanning forests
Authors:
Luca Avena,
Fabienne Castell,
Alexandre Gaudillière,
Clothilde Melot
Abstract:
For different reversible Markov kernels on finite state spaces, we look for families of probability measures for which the time evolution almost remains in their convex hull. Motivated by signal processing problems and metastability studies we are interested in the case when the size of such families is smaller than the size of the state space, and we want such distributions to be with small overl…
▽ More
For different reversible Markov kernels on finite state spaces, we look for families of probability measures for which the time evolution almost remains in their convex hull. Motivated by signal processing problems and metastability studies we are interested in the case when the size of such families is smaller than the size of the state space, and we want such distributions to be with small overlap among them. To this aim we introduce a squeezing function to measure the common overlap of such families, and we use random forests to build random approximate solutions of the associated intertwining equations for which we can bound from above the expected values of both squeezing and total variation errors. We also explain how to modify some of these approximate solutions into exact solutions by using those eigenvalues of the associated Laplacian with the largest absolute values.
△ Less
Submitted 20 February, 2017;
originally announced February 2017.
-
Persistence exponent for random processes in Brownian scenery
Authors:
Fabienne Castell,
Nadine Guillotin-Plantard,
Frederique Watbled
Abstract:
In this paper we consider the persistence properties of random processes in Brownian scenery, which are examples of non-Markovian and
non-Gaussian processes. More precisely we study the asymptotic behaviour for large $T$, of the probability $P[ \sup\_{t\in[0,T]} Δ\_t \leq 1] $
where $Δ\_t = \int\_{\mathbb{R}} L\_t(x) \, dW(x).$
Here $W={W(x); x\in\mathbb{R}}$ is a two-sided standard real Browni…
▽ More
In this paper we consider the persistence properties of random processes in Brownian scenery, which are examples of non-Markovian and
non-Gaussian processes. More precisely we study the asymptotic behaviour for large $T$, of the probability $P[ \sup\_{t\in[0,T]} Δ\_t \leq 1] $
where $Δ\_t = \int\_{\mathbb{R}} L\_t(x) \, dW(x).$
Here $W={W(x); x\in\mathbb{R}}$ is a two-sided standard real Brownian motion and ${L\_t(x); x\in\mathbb{R},t\geq 0}$ is
the local time of some self-similar random process $Y$, independent from the process $W$. We thus generalize the results of \cite{BFFN} where the increments of $Y$ were assumed to be independent.
△ Less
Submitted 24 February, 2015; v1 submitted 1 July, 2014;
originally announced July 2014.
-
Exponential moments of self-intersection local times of stable random walks in subcritical dimensions
Authors:
Fabienne Castell,
Clément Laurent,
Clothilde Mélot
Abstract:
Let $(X_t, t \geq 0)$ be an $α$-stable random walk with values in $\Z^d$. Let $l_t(x) = \int_0^t δ_x(X_s) ds$ be its local time. For $p>1$, not necessarily integer, $I_t = \sum_x l_t^p(x)$ is the so-called $p$-fold self- intersection local time of the random walk. When $p(d -α) < d$, we derive precise logarithmic asymptotics of the probability $P(I_t \geq r_t)$ for all scales $r_t \gg \E(I_t)$. Ou…
▽ More
Let $(X_t, t \geq 0)$ be an $α$-stable random walk with values in $\Z^d$. Let $l_t(x) = \int_0^t δ_x(X_s) ds$ be its local time. For $p>1$, not necessarily integer, $I_t = \sum_x l_t^p(x)$ is the so-called $p$-fold self- intersection local time of the random walk. When $p(d -α) < d$, we derive precise logarithmic asymptotics of the probability $P(I_t \geq r_t)$ for all scales $r_t \gg \E(I_t)$. Our result extends previous works by Chen, Li and Rosen 2005, Becker and König 2010, and Laurent 2012.
△ Less
Submitted 22 May, 2012;
originally announced May 2012.
-
On the local time of random processes in random scenery
Authors:
Fabienne Castell,
Nadine Guillotin--Plantard,
Françoise Pène,
Bruno Schapira
Abstract:
Random walks in random scenery are processes defined by $Z_n:=\sum_{k=1}^nξ_{X_1+...+X_k}$, where basically $(X_k,k\ge 1)$ and $(ξ_y,y\in\mathbb Z)$ are two independent sequences of i.i.d. random variables. We assume here that $X_1$ is $\ZZ$-valued, centered and with finite moments of all orders. We also assume that $ξ_0$ is $\ZZ$-valued, centered and square integrable. In this case H. Kesten and…
▽ More
Random walks in random scenery are processes defined by $Z_n:=\sum_{k=1}^nξ_{X_1+...+X_k}$, where basically $(X_k,k\ge 1)$ and $(ξ_y,y\in\mathbb Z)$ are two independent sequences of i.i.d. random variables. We assume here that $X_1$ is $\ZZ$-valued, centered and with finite moments of all orders. We also assume that $ξ_0$ is $\ZZ$-valued, centered and square integrable. In this case H. Kesten and F. Spitzer proved that $(n^{-3/4}Z_{[nt]},t\ge 0)$ converges in distribution as $n\to \infty$ toward some self-similar process $(Δ_t,t\ge 0)$ called Brownian motion in random scenery. In a previous paper, we established that ${\mathbb P}(Z_n=0)$ behaves asymptotically like a constant times $n^{-3/4}$, as $n\to \infty$. We extend here this local limit theorem: we give a precise asymptotic result for the probability for $Z$ to return to zero simultaneously at several times. As a byproduct of our computations, we show that $Δ$ admits a bi-continuous version of its local time process which is locally Hölder continuous of order $1/4-δ$ and $1/6-δ$, respectively in the time and space variables, for any $δ>0$. In particular, this gives a new proof of the fact, previously obtained by Khoshnevisan, that the level sets of $Δ$ have Hausdorff dimension a.s. equal to 1/4. We also get the convergence of every moment of the normalized local time of $Z$ toward its continuous counterpart.
△ Less
Submitted 15 February, 2012;
originally announced February 2012.
-
Limit theorems for one and two-dimensional random walks in random scenery
Authors:
Fabienne Castell,
Nadine Guillotin--Plantard,
Françoise Pène
Abstract:
Random walks in random scenery are processes defined by $Z_n:=\sum_{k=1}^nξ_{X_1+...+X_k}$, where $(X_k,k\ge 1)$ and $(ξ_y,y\in{\mathbb Z}^d)$ are two independent sequences of i.i.d. random variables with values in ${\mathbb Z}^d$ and $\mathbb R$ respectively. We suppose that the distributions of $X_1$ and $ξ_0$ belong to the normal basin of attraction of stable distribution of index $α\in(0,2]$ a…
▽ More
Random walks in random scenery are processes defined by $Z_n:=\sum_{k=1}^nξ_{X_1+...+X_k}$, where $(X_k,k\ge 1)$ and $(ξ_y,y\in{\mathbb Z}^d)$ are two independent sequences of i.i.d. random variables with values in ${\mathbb Z}^d$ and $\mathbb R$ respectively. We suppose that the distributions of $X_1$ and $ξ_0$ belong to the normal basin of attraction of stable distribution of index $α\in(0,2]$ and $β\in(0,2]$. When $d=1$ and $α\ne 1$, a functional limit theorem has been established in \cite{KestenSpitzer} and a local limit theorem in \cite{BFFN}. In this paper, we establish the convergence of the finite-dimensional distributions and a local limit theorem when $α=d$ (i.e. $α= d=1$ or $α=d=2$) and $β\in (0,2]$. Let us mention that functional limit theorems have been established in \cite{bolthausen} and recently in \cite{DU} in the particular case where $β=2$ (respectively for $α=d=2$ and $α=d=1$).
△ Less
Submitted 23 March, 2011;
originally announced March 2011.
-
Parabolic Anderson model with a finite number of moving catalysts
Authors:
Fabienne Castell,
Onur Gün,
Grégory Maillard
Abstract:
We consider the parabolic Anderson model (PAM) which is given by the equation $\partial u/\partial t = κΔu + ξu$ with $u\colon\, \Z^d\times [0,\infty)\to \R$, where $κ\in [0,\infty)$ is the diffusion constant, $Δ$ is the discrete Laplacian, and $ξ\colon\,\Z^d\times [0,\infty)\to\R$ is a space-time random environment that drives the equation. The solution of this equation describes the evolution of…
▽ More
We consider the parabolic Anderson model (PAM) which is given by the equation $\partial u/\partial t = κΔu + ξu$ with $u\colon\, \Z^d\times [0,\infty)\to \R$, where $κ\in [0,\infty)$ is the diffusion constant, $Δ$ is the discrete Laplacian, and $ξ\colon\,\Z^d\times [0,\infty)\to\R$ is a space-time random environment that drives the equation. The solution of this equation describes the evolution of a "reactant" $u$ under the influence of a "catalyst" $ξ$. In the present paper we focus on the case where $ξ$ is a system of $n$ independent simple random walks each with step rate $2dρ$ and starting from the origin. We study the \emph{annealed} Lyapunov exponents, i.e., the exponential growth rates of the successive moments of $u$ w.r.t.\ $ξ$ and show that these exponents, as a function of the diffusion constant $κ$ and the rate constant $ρ$, behave differently depending on the dimension $d$. In particular, we give a description of the intermittent behavior of the system in terms of the annealed Lyapunov exponents, depicting how the total mass of $u$ concentrates as $t\to\infty$. Our results are both a generalization and an extension of the work of Gärtner and Heydenreich 2006, where only the case $n=1$ was investigated.
△ Less
Submitted 23 March, 2011; v1 submitted 23 October, 2010;
originally announced October 2010.
-
A local limit theorem for random walks in random scenery and on randomly oriented lattices
Authors:
Fabienne Castell,
Nadine Guillotin-Plantard,
Françoise Pène,
Bruno Schapira
Abstract:
Random walks in random scenery are processes defined by $Z_n:=\sum_{k=1}^nξ_{X_1+...+X_k}$, where $(X_k,k\ge 1)$ and $(ξ_y,y\in\mathbb Z)$ are two independent sequences of i.i.d. random variables. We assume here that their distributions belong to the normal domain of attraction of stable laws with index $α\in (0,2]$ and $β\in (0,2]$ respectively. These processes were first studied by H. Kesten a…
▽ More
Random walks in random scenery are processes defined by $Z_n:=\sum_{k=1}^nξ_{X_1+...+X_k}$, where $(X_k,k\ge 1)$ and $(ξ_y,y\in\mathbb Z)$ are two independent sequences of i.i.d. random variables. We assume here that their distributions belong to the normal domain of attraction of stable laws with index $α\in (0,2]$ and $β\in (0,2]$ respectively. These processes were first studied by H. Kesten and F. Spitzer, who proved the convergence in distribution when $α\neq 1$ and as $n\to \infty$, of $n^{-δ}Z_n$, for some suitable $δ>0$ depending on $α$ and $β$. Here we are interested in the convergence, as $n\to \infty$, of $n^δ{\mathbb P}(Z_n=\lfloor n^δ x\rfloor)$, when $x\in \RR$ is fixed. We also consider the case of random walks on randomly oriented lattices for which we obtain similar results.
△ Less
Submitted 9 February, 2010;
originally announced February 2010.
-
Large deviations for intersection local times in critical dimension
Authors:
Fabienne Castell
Abstract:
Let $(X_t,t\geq0)$ be a continuous time simple random walk on $\mathbb{Z}^d$ ($d\geq3$), and let $l_T(x)$ be the time spent by $(X_t,t\geq0)$ on the site $x$ up to time $T$. We prove a large deviations principle for the $q$-fold self-intersection local time $I_T=\sum_{x\in\mathbb{Z}^d}l_T(x)^q$ in the critical case $q=\frac{d}{d-2}$. When $q$ is integer, we obtain similar results for the intersect…
▽ More
Let $(X_t,t\geq0)$ be a continuous time simple random walk on $\mathbb{Z}^d$ ($d\geq3$), and let $l_T(x)$ be the time spent by $(X_t,t\geq0)$ on the site $x$ up to time $T$. We prove a large deviations principle for the $q$-fold self-intersection local time $I_T=\sum_{x\in\mathbb{Z}^d}l_T(x)^q$ in the critical case $q=\frac{d}{d-2}$. When $q$ is integer, we obtain similar results for the intersection local times of $q$ independent simple random walks.
△ Less
Submitted 2 October, 2010; v1 submitted 9 December, 2008;
originally announced December 2008.
-
Self-Intersection Times for Random Walk, and Random Walk in Random Scenery in dimensions d>4
Authors:
Amine Asselah,
Fabienne Castell
Abstract:
We consider Random Walk in Random Scenery, denoted $X_n$, where the random walk is symmetric on $Z^d$, with $d>4$, and the random field is made up of i.i.d random variables with a stretched exponential tail decay, with exponent $α$ with $1<α$. We present asymptotics for the probability, over both randomness, that $\{X_n>n^β\}$ for $1/2<β<1$. To obtain such asymptotics, we establish large deviati…
▽ More
We consider Random Walk in Random Scenery, denoted $X_n$, where the random walk is symmetric on $Z^d$, with $d>4$, and the random field is made up of i.i.d random variables with a stretched exponential tail decay, with exponent $α$ with $1<α$. We present asymptotics for the probability, over both randomness, that $\{X_n>n^β\}$ for $1/2<β<1$. To obtain such asymptotics, we establish large deviations estimates for the the self-intersection local times process.
△ Less
Submitted 19 October, 2005; v1 submitted 30 September, 2005;
originally announced September 2005.
-
A note on random walk in random scenery
Authors:
Amine Asselah,
Fabienne Castell
Abstract:
We consider a d-dimensional random walk in random scenery X(n), where the scenery consists of i.i.d. with exponential moments but a tail decay of the form exp(-c t^a) with a<d/2. We study the probability, when averaged over both randomness, that {X(n)>ny}. We show that this probability is of order exp(-(ny)^b) with b=a/(a+1).
We consider a d-dimensional random walk in random scenery X(n), where the scenery consists of i.i.d. with exponential moments but a tail decay of the form exp(-c t^a) with a<d/2. We study the probability, when averaged over both randomness, that {X(n)>ny}. We show that this probability is of order exp(-(ny)^b) with b=a/(a+1).
△ Less
Submitted 5 January, 2005;
originally announced January 2005.
-
Large deviations for Brownian motion in a random scenery
Authors:
A. Asselah,
F. Castell
Abstract:
We prove large deviations principles in large time, for the Brownian occupation time in random scenery. The random scenery is constant on unit cubes, and consist of i.i.d. bounded variables, independent of the Brownian motion. This model is a time-continuous version of Kesten and Spitzer's random walk in random scenery. We prove large deviations principles in ``quenched'' and ``annealed'' settin…
▽ More
We prove large deviations principles in large time, for the Brownian occupation time in random scenery. The random scenery is constant on unit cubes, and consist of i.i.d. bounded variables, independent of the Brownian motion. This model is a time-continuous version of Kesten and Spitzer's random walk in random scenery. We prove large deviations principles in ``quenched'' and ``annealed'' settings.
△ Less
Submitted 13 September, 2002;
originally announced September 2002.
-
Existence of quasi-stationary measures for asymmetric attractive particle systems on $\ZZ^d$
Authors:
A. Asselah,
F. Castell
Abstract:
We show the existence of non-trivial quasi-stationary measures for conservative attractive particle systems on $\ZZ^d$ conditioned on avoiding an increasing local set $\A$. Moreover, we exhibit a sequence of measures $\{ν_n\}$, whose $ω$-limit set consists of quasi-stationary measures. For zero range processes, with stationary measure $\nur$, we prove the existence of an $L^2(\nur)$ nonnegative…
▽ More
We show the existence of non-trivial quasi-stationary measures for conservative attractive particle systems on $\ZZ^d$ conditioned on avoiding an increasing local set $\A$. Moreover, we exhibit a sequence of measures $\{ν_n\}$, whose $ω$-limit set consists of quasi-stationary measures. For zero range processes, with stationary measure $\nur$, we prove the existence of an $L^2(\nur)$ nonnegative eigenvector for the generator with Dirichlet boundary on $\A$, after establishing a priori bounds on the $\{ν_n\}$.
△ Less
Submitted 28 February, 2002;
originally announced February 2002.
-
Quenched large deviations for diffusions in a random Gaussian shear flow drift
Authors:
A. Asselah,
F. Castell
Abstract:
We prove a full large deviations principle in large time, for a diffusion process with random drift V, which is a centered Gaussian shear flow random field. The large deviations principle is established in a ``quenched'' setting, i.e. is valid almost surely in the randomness of V.
We prove a full large deviations principle in large time, for a diffusion process with random drift V, which is a centered Gaussian shear flow random field. The large deviations principle is established in a ``quenched'' setting, i.e. is valid almost surely in the randomness of V.
△ Less
Submitted 27 February, 2002;
originally announced February 2002.
