Namespaces
Variants
Actions

Difference between revisions of "Wiener sausage"

From Encyclopedia of Mathematics
Jump to: navigation, search
(Importing text file)
 
m (AUTOMATIC EDIT (latexlist): Replaced 47 formulas out of 50 by TEX code with an average confidence of 2.0 and a minimal confidence of 2.0.)
Line 1: Line 1:
Let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w1301001.png" />, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w1301002.png" />, be the standard [[Brownian motion|Brownian motion]] in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w1301003.png" /> (i.e. the [[Markov process|Markov process]] with generator <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w1301004.png" />) starting at <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w1301005.png" />. Let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w1301006.png" />, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w1301007.png" /> denote its probability law and expectation on path space. The Wiener sausage with radius <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w1301008.png" /> is the process defined by
+
<!--This article has been texified automatically. Since there was no Nroff source code for this article,  
 +
the semi-automatic procedure described at https://encyclopediaofmath.org/wiki/User:Maximilian_Janisch/latexlist
 +
was used.
 +
If the TeX and formula formatting is correct, please remove this message and the {{TEX|semi-auto}} category.
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w1301009.png" /></td> </tr></table>
+
Out of 50 formulas, 47 were replaced by TEX code.-->
  
where <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010010.png" /> is the open ball with radius <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010011.png" /> around <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010012.png" />.
+
{{TEX|semi-auto}}{{TEX|partial}}
 +
Let $\beta ( t )$, $t \geq 0$, be the standard [[Brownian motion|Brownian motion]] in $\mathbf{R} ^ { d }$ (i.e. the [[Markov process|Markov process]] with generator $\Delta / 2$) starting at $0$. Let $\textsf{P}$, $\mathsf{E}$ denote its probability law and expectation on path space. The Wiener sausage with radius $a &gt; 0$ is the process defined by
 +
 
 +
\begin{equation*} W ^ { a } ( t ) = \bigcup _ { 0 \leq s \leq t } B _ { a } ( \beta ( s ) ) , \quad t \geq 0, \end{equation*}
 +
 
 +
where $B _ { a } ( x )$ is the open ball with radius $a$ around $x \in \mathbf{R} ^ { d }$.
  
 
The Wiener sausage is an important mathematical object, because it is one of the simplest examples of a non-Markovian functional of Brownian motion. It plays a key role in the study of various stochastic phenomena, such as heat conduction and trapping in random media, as well as in the analysis of spectral properties of random Schrödinger operators (cf. also [[Schrödinger equation|Schrödinger equation]]).
 
The Wiener sausage is an important mathematical object, because it is one of the simplest examples of a non-Markovian functional of Brownian motion. It plays a key role in the study of various stochastic phenomena, such as heat conduction and trapping in random media, as well as in the analysis of spectral properties of random Schrödinger operators (cf. also [[Schrödinger equation|Schrödinger equation]]).
  
A lot is known about the behaviour of the volume of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010013.png" /> as <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010014.png" />. For instance,
+
A lot is known about the behaviour of the volume of $W ^ { a } ( t )$ as $t \rightarrow \infty$. For instance,
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010015.png" /></td> </tr></table>
+
\begin{equation*} \mathsf{E} | W ^ { a } ( t ) | \sim \left\{ \begin{array} { l l } { \sqrt { \frac { 8 t } { \pi } } , } &amp; { d = 1, } \\ { \frac { 2 \pi t } { \operatorname { log } t } , } &amp; { d = 2, } \\ { \kappa _ { a } t , } &amp; { d \geq 3, } \end{array} \right. \end{equation*}
  
with <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010016.png" /> the Newtonian capacity of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010017.png" /> associated with the Green's function of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010018.png" /> (cf. also [[Green function|Green function]]; [[Capacity|Capacity]]), and
+
with $\kappa _ { a } = a ^ { d - 2 } 2 \pi ^ { d / 2 } / \Gamma ( ( d - 2 ) / 2 )$ the Newtonian capacity of $B _ { a } ( 0 )$ associated with the Green's function of $( - \Delta / 2 ) ^ { - 1 }$ (cf. also [[Green function|Green function]]; [[Capacity|Capacity]]), and
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010019.png" /></td> </tr></table>
+
\begin{equation*} \operatorname { Var } | W ^ { a } ( t ) | \asymp \left\{ \begin{array} { l l } { t , } &amp; { d = 1, } \\ { \frac { t ^ { 2 } } { \operatorname { log } ^ { 4 } t } , } &amp; { d = 2, } \\ { \operatorname { tlog } t , } &amp; { d = 3, } \\ { t , } &amp; { d \geq 4 } \end{array} \right. \end{equation*}
  
([[#References|[a8]]], [[#References|[a6]]]). Moreover, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010020.png" /> satisfies the [[Strong law of large numbers|strong law of large numbers]] and the [[Central limit theorem|central limit theorem]] for <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010021.png" />; the limit law is Gaussian for <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010022.png" /> and non-Gaussian for <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010023.png" /> ([[#References|[a7]]]). Note that for <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010024.png" /> the Wiener sausage is a sparse object: since the Brownian motion typically travels a distance <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010025.png" /> in each direction, the last two displays show that most of the space in the convex hull of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010026.png" /> is not covered.
+
([[#References|[a8]]], [[#References|[a6]]]). Moreover, $| W^ { a } ( t ) |$ satisfies the [[Strong law of large numbers|strong law of large numbers]] and the [[Central limit theorem|central limit theorem]] for $d \geq 2$; the limit law is Gaussian for $d \geq 3$ and non-Gaussian for $d = 2$ ([[#References|[a7]]]). Note that for $d \geq 2$ the Wiener sausage is a sparse object: since the Brownian motion typically travels a distance $\sqrt { t }$ in each direction, the last two displays show that most of the space in the convex hull of $W ^ { a } ( t )$ is not covered.
  
The large deviation behaviour of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010027.png" /> in the downward direction has been studied in [[#References|[a5]]], [[#References|[a4]]] and [[#References|[a9]]]. For <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010028.png" /> the outcome, proved in successive stages of refinement, reads as follows:
+
The large deviation behaviour of $| W^ { a } ( t ) |$ in the downward direction has been studied in [[#References|[a5]]], [[#References|[a4]]] and [[#References|[a9]]]. For $d \geq 2$ the outcome, proved in successive stages of refinement, reads as follows:
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010029.png" /></td> </tr></table>
+
\begin{equation*} \operatorname { lim } _ { t \rightarrow \infty } \frac { f ( t ) ^ { 2 / d } } { t } \operatorname { log } \mathsf{P} ( | W ^ { a } ( t ) | \leq f ( t ) ) = - \frac { 1 } { 2 } \lambda _ { d } \end{equation*}
  
for any <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010030.png" /> satisfying <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010031.png" /> and
+
for any $f : \mathbf{R} _ { + } \rightarrow \mathbf{R} _ { + }$ satisfying $\operatorname { lim } _ { t \rightarrow \infty } f ( t ) = \infty$ and
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010032.png" /></td> </tr></table>
+
\begin{equation*} f ( t ) = \left\{ \begin{array} { l l } { o \left( \frac { t } { \operatorname { log } t } \right) , } &amp; { d = 2, } \\ { o ( t ) , } &amp; { d \geq 3, } \end{array} \right. \end{equation*}
  
where <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010033.png" /> is the smallest Dirichlet eigenvalue of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010034.png" /> on the ball with unit volume. The optimal strategy for the Brownian motion to realize the large deviation is to stay inside a ball with volume <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010035.png" /> until time <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010036.png" />, i.e., the Wiener sausage covers this ball entirely and nothing outside. This comes from the Faber–Krahn isoperimetric inequality (cf. also [[Rayleigh–Faber–Krahn inequality|Rayleigh–Faber–Krahn inequality]]), and the cost of staying inside the ball is
+
where $\lambda _ { d } &gt; 0$ is the smallest Dirichlet eigenvalue of $- \Delta$ on the ball with unit volume. The optimal strategy for the Brownian motion to realize the large deviation is to stay inside a ball with volume $f ( t )$ until time $t$, i.e., the Wiener sausage covers this ball entirely and nothing outside. This comes from the Faber–Krahn isoperimetric inequality (cf. also [[Rayleigh–Faber–Krahn inequality|Rayleigh–Faber–Krahn inequality]]), and the cost of staying inside the ball is
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010037.png" /></td> </tr></table>
+
\begin{equation*} \operatorname { exp } \left[ - \frac { 1 } { 2 } \lambda _ { d } \frac { t } { f ( t ) ^ { 2 / d  } } \right] \end{equation*}
  
 
to leading order. Note that, apparently, a large deviation below the scale of the mean  "squeezes all the empty space out of the Wiener sausage" .
 
to leading order. Note that, apparently, a large deviation below the scale of the mean  "squeezes all the empty space out of the Wiener sausage" .
  
The above analysis of the large deviation behaviour has recently been extended to cover the moderate deviation behaviour. It is proved in [[#References|[a2]]] that for <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010038.png" />,
+
The above analysis of the large deviation behaviour has recently been extended to cover the moderate deviation behaviour. It is proved in [[#References|[a2]]] that for $d \geq 3$,
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010039.png" /></td> <td valign="top" style="width:5%;text-align:right;">(a1)</td></tr></table>
+
<table class="eq" style="width:100%;"> <tr><td style="width:94%;text-align:center;" valign="top"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010039.png"/></td> <td style="width:5%;text-align:right;" valign="top">(a1)</td></tr></table>
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010040.png" /></td> </tr></table>
+
\begin{equation*} = - I ^ { \kappa_a }  ( b ) \in ( - \infty , 0 ) , \text { for all } 0 &lt; b &lt; \kappa _ { a }, \end{equation*}
  
and a variational representation is derived for the rate function <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010041.png" />. The optimal strategy for the Brownian motion to realize the moderate deviation is such that the Wiener sausage  "looks like a Swiss cheese" : <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010042.png" /> has random holes whose sizes are of order <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010043.png" /> and whose density varies on scale <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010044.png" />. This is markedly different from the optimal strategy behind the large deviation. Note that, apparently, a moderate deviation on the scale of the mean  "does not squeeze all the empty space out of the Wiener sausage" . (a1) has also been extended to <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010045.png" />.
+
and a variational representation is derived for the rate function <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010041.png"/>. The optimal strategy for the Brownian motion to realize the moderate deviation is such that the Wiener sausage  "looks like a Swiss cheese" : $W ^ { a } ( t )$ has random holes whose sizes are of order $1$ and whose density varies on scale $t ^ { 1 / d }$. This is markedly different from the optimal strategy behind the large deviation. Note that, apparently, a moderate deviation on the scale of the mean  "does not squeeze all the empty space out of the Wiener sausage" . (a1) has also been extended to $d = 2$.
  
It turns out that the rate function <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010046.png" /> exhibits rich behaviour as a function of the dimension. In particular, for <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010047.png" /> it is non-analytic at a certain critical value inside <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010048.png" />, which is associated with a collapse transition in the optimal strategy.
+
It turns out that the rate function $b \mapsto I ^ { \kappa_a } ( b )$ exhibits rich behaviour as a function of the dimension. In particular, for $d \geq 5$ it is non-analytic at a certain critical value inside $( 0 , \kappa _ { a} )$, which is associated with a collapse transition in the optimal strategy.
  
Finally, the moderate and large deviations of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/w/w130/w130100/w13010049.png" /> in the upward direction are a complicated issue. Here the optimal strategy is entirely different from the previous ones, because the Wiener sausage tries to expand rather than to contract. Partial results have been obtained in [[#References|[a3]]] [[#References|[a1]]], and [[#References|[a11]]].
+
Finally, the moderate and large deviations of $| W^ { a } ( t ) |$ in the upward direction are a complicated issue. Here the optimal strategy is entirely different from the previous ones, because the Wiener sausage tries to expand rather than to contract. Partial results have been obtained in [[#References|[a3]]] [[#References|[a1]]], and [[#References|[a11]]].
  
 
More background can be found in [[#References|[a10]]].
 
More background can be found in [[#References|[a10]]].
  
 
====References====
 
====References====
<table><TR><TD valign="top">[a1]</TD> <TD valign="top">  M. van den Berg,  E. Bolthausen,  "Asymptotics of the generating function for the volume of the Wiener sausage"  ''Probab. Th. Rel. Fields'' , '''99'''  (1994)  pp. 389–397</TD></TR><TR><TD valign="top">[a2]</TD> <TD valign="top">  M. van den Berg,  E. Bolthausen,  F. den Hollander,  "Moderate deviations for the volume of the Wiener sausage"  ''Ann. of Math.''  (to appear in 2001)</TD></TR><TR><TD valign="top">[a3]</TD> <TD valign="top">  M. van den Berg,  B. Tóth,  "Exponential estimates for the Wiener sausage"  ''Probab. Th. Rel. Fields'' , '''88'''  (1991)  pp. 249–259</TD></TR><TR><TD valign="top">[a4]</TD> <TD valign="top">  E. Bolthausen,  "On the volume of the Wiener sausage"  ''Ann. Probab.'' , '''18'''  (1990)  pp. 1576–1582</TD></TR><TR><TD valign="top">[a5]</TD> <TD valign="top">  M.D. Donsker,  S.R.S. Varadhan,  "Asymptotics for the Wiener sausage"  ''Commun. Pure Appl. Math.'' , '''28'''  (1975)  pp. 525–565</TD></TR><TR><TD valign="top">[a6]</TD> <TD valign="top">  J.-F. Le Gall,  "Sur une conjecture de M. Kac"  ''Probab. Th. Rel. Fields'' , '''78'''  (1988)  pp. 389–402</TD></TR><TR><TD valign="top">[a7]</TD> <TD valign="top">  J.-F. Le Gall,  "Fluctuation results for the Wiener sausage"  ''Ann. Probab.'' , '''16'''  (1988)  pp. 991–1018</TD></TR><TR><TD valign="top">[a8]</TD> <TD valign="top">  F. Spitzer,  "Electrostatic capacity, heat flow and Brownian motion"  ''Z. Wahrsch. Verw. Gebiete'' , '''3'''  (1964)  pp. 110–121</TD></TR><TR><TD valign="top">[a9]</TD> <TD valign="top">  A.-S. Sznitman,  "Long time asymptotics for the shrinking Wiener sausage"  ''Commun. Pure Appl. Math.'' , '''43'''  (1990)  pp. 809–820</TD></TR><TR><TD valign="top">[a10]</TD> <TD valign="top">  A.-S. Sznitman,  "Brownian motion, obstacles and random media" , Springer  (1998)</TD></TR><TR><TD valign="top">[a11]</TD> <TD valign="top">  Y. Hamana,  H. Kesten,  "A large deviation result for the range of random walk and for the Wiener sausage"  ''preprint March''  (2000)</TD></TR></table>
+
<table><tr><td valign="top">[a1]</td> <td valign="top">  M. van den Berg,  E. Bolthausen,  "Asymptotics of the generating function for the volume of the Wiener sausage"  ''Probab. Th. Rel. Fields'' , '''99'''  (1994)  pp. 389–397</td></tr><tr><td valign="top">[a2]</td> <td valign="top">  M. van den Berg,  E. Bolthausen,  F. den Hollander,  "Moderate deviations for the volume of the Wiener sausage"  ''Ann. of Math.''  (to appear in 2001)</td></tr><tr><td valign="top">[a3]</td> <td valign="top">  M. van den Berg,  B. Tóth,  "Exponential estimates for the Wiener sausage"  ''Probab. Th. Rel. Fields'' , '''88'''  (1991)  pp. 249–259</td></tr><tr><td valign="top">[a4]</td> <td valign="top">  E. Bolthausen,  "On the volume of the Wiener sausage"  ''Ann. Probab.'' , '''18'''  (1990)  pp. 1576–1582</td></tr><tr><td valign="top">[a5]</td> <td valign="top">  M.D. Donsker,  S.R.S. Varadhan,  "Asymptotics for the Wiener sausage"  ''Commun. Pure Appl. Math.'' , '''28'''  (1975)  pp. 525–565</td></tr><tr><td valign="top">[a6]</td> <td valign="top">  J.-F. Le Gall,  "Sur une conjecture de M. Kac"  ''Probab. Th. Rel. Fields'' , '''78'''  (1988)  pp. 389–402</td></tr><tr><td valign="top">[a7]</td> <td valign="top">  J.-F. Le Gall,  "Fluctuation results for the Wiener sausage"  ''Ann. Probab.'' , '''16'''  (1988)  pp. 991–1018</td></tr><tr><td valign="top">[a8]</td> <td valign="top">  F. Spitzer,  "Electrostatic capacity, heat flow and Brownian motion"  ''Z. Wahrsch. Verw. Gebiete'' , '''3'''  (1964)  pp. 110–121</td></tr><tr><td valign="top">[a9]</td> <td valign="top">  A.-S. Sznitman,  "Long time asymptotics for the shrinking Wiener sausage"  ''Commun. Pure Appl. Math.'' , '''43'''  (1990)  pp. 809–820</td></tr><tr><td valign="top">[a10]</td> <td valign="top">  A.-S. Sznitman,  "Brownian motion, obstacles and random media" , Springer  (1998)</td></tr><tr><td valign="top">[a11]</td> <td valign="top">  Y. Hamana,  H. Kesten,  "A large deviation result for the range of random walk and for the Wiener sausage"  ''preprint March''  (2000)</td></tr></table>

Revision as of 16:55, 1 July 2020

Let $\beta ( t )$, $t \geq 0$, be the standard Brownian motion in $\mathbf{R} ^ { d }$ (i.e. the Markov process with generator $\Delta / 2$) starting at $0$. Let $\textsf{P}$, $\mathsf{E}$ denote its probability law and expectation on path space. The Wiener sausage with radius $a > 0$ is the process defined by

\begin{equation*} W ^ { a } ( t ) = \bigcup _ { 0 \leq s \leq t } B _ { a } ( \beta ( s ) ) , \quad t \geq 0, \end{equation*}

where $B _ { a } ( x )$ is the open ball with radius $a$ around $x \in \mathbf{R} ^ { d }$.

The Wiener sausage is an important mathematical object, because it is one of the simplest examples of a non-Markovian functional of Brownian motion. It plays a key role in the study of various stochastic phenomena, such as heat conduction and trapping in random media, as well as in the analysis of spectral properties of random Schrödinger operators (cf. also Schrödinger equation).

A lot is known about the behaviour of the volume of $W ^ { a } ( t )$ as $t \rightarrow \infty$. For instance,

\begin{equation*} \mathsf{E} | W ^ { a } ( t ) | \sim \left\{ \begin{array} { l l } { \sqrt { \frac { 8 t } { \pi } } , } & { d = 1, } \\ { \frac { 2 \pi t } { \operatorname { log } t } , } & { d = 2, } \\ { \kappa _ { a } t , } & { d \geq 3, } \end{array} \right. \end{equation*}

with $\kappa _ { a } = a ^ { d - 2 } 2 \pi ^ { d / 2 } / \Gamma ( ( d - 2 ) / 2 )$ the Newtonian capacity of $B _ { a } ( 0 )$ associated with the Green's function of $( - \Delta / 2 ) ^ { - 1 }$ (cf. also Green function; Capacity), and

\begin{equation*} \operatorname { Var } | W ^ { a } ( t ) | \asymp \left\{ \begin{array} { l l } { t , } & { d = 1, } \\ { \frac { t ^ { 2 } } { \operatorname { log } ^ { 4 } t } , } & { d = 2, } \\ { \operatorname { tlog } t , } & { d = 3, } \\ { t , } & { d \geq 4 } \end{array} \right. \end{equation*}

([a8], [a6]). Moreover, $| W^ { a } ( t ) |$ satisfies the strong law of large numbers and the central limit theorem for $d \geq 2$; the limit law is Gaussian for $d \geq 3$ and non-Gaussian for $d = 2$ ([a7]). Note that for $d \geq 2$ the Wiener sausage is a sparse object: since the Brownian motion typically travels a distance $\sqrt { t }$ in each direction, the last two displays show that most of the space in the convex hull of $W ^ { a } ( t )$ is not covered.

The large deviation behaviour of $| W^ { a } ( t ) |$ in the downward direction has been studied in [a5], [a4] and [a9]. For $d \geq 2$ the outcome, proved in successive stages of refinement, reads as follows:

\begin{equation*} \operatorname { lim } _ { t \rightarrow \infty } \frac { f ( t ) ^ { 2 / d } } { t } \operatorname { log } \mathsf{P} ( | W ^ { a } ( t ) | \leq f ( t ) ) = - \frac { 1 } { 2 } \lambda _ { d } \end{equation*}

for any $f : \mathbf{R} _ { + } \rightarrow \mathbf{R} _ { + }$ satisfying $\operatorname { lim } _ { t \rightarrow \infty } f ( t ) = \infty$ and

\begin{equation*} f ( t ) = \left\{ \begin{array} { l l } { o \left( \frac { t } { \operatorname { log } t } \right) , } & { d = 2, } \\ { o ( t ) , } & { d \geq 3, } \end{array} \right. \end{equation*}

where $\lambda _ { d } > 0$ is the smallest Dirichlet eigenvalue of $- \Delta$ on the ball with unit volume. The optimal strategy for the Brownian motion to realize the large deviation is to stay inside a ball with volume $f ( t )$ until time $t$, i.e., the Wiener sausage covers this ball entirely and nothing outside. This comes from the Faber–Krahn isoperimetric inequality (cf. also Rayleigh–Faber–Krahn inequality), and the cost of staying inside the ball is

\begin{equation*} \operatorname { exp } \left[ - \frac { 1 } { 2 } \lambda _ { d } \frac { t } { f ( t ) ^ { 2 / d } } \right] \end{equation*}

to leading order. Note that, apparently, a large deviation below the scale of the mean "squeezes all the empty space out of the Wiener sausage" .

The above analysis of the large deviation behaviour has recently been extended to cover the moderate deviation behaviour. It is proved in [a2] that for $d \geq 3$,

(a1)

\begin{equation*} = - I ^ { \kappa_a } ( b ) \in ( - \infty , 0 ) , \text { for all } 0 < b < \kappa _ { a }, \end{equation*}

and a variational representation is derived for the rate function . The optimal strategy for the Brownian motion to realize the moderate deviation is such that the Wiener sausage "looks like a Swiss cheese" : $W ^ { a } ( t )$ has random holes whose sizes are of order $1$ and whose density varies on scale $t ^ { 1 / d }$. This is markedly different from the optimal strategy behind the large deviation. Note that, apparently, a moderate deviation on the scale of the mean "does not squeeze all the empty space out of the Wiener sausage" . (a1) has also been extended to $d = 2$.

It turns out that the rate function $b \mapsto I ^ { \kappa_a } ( b )$ exhibits rich behaviour as a function of the dimension. In particular, for $d \geq 5$ it is non-analytic at a certain critical value inside $( 0 , \kappa _ { a} )$, which is associated with a collapse transition in the optimal strategy.

Finally, the moderate and large deviations of $| W^ { a } ( t ) |$ in the upward direction are a complicated issue. Here the optimal strategy is entirely different from the previous ones, because the Wiener sausage tries to expand rather than to contract. Partial results have been obtained in [a3] [a1], and [a11].

More background can be found in [a10].

References

[a1] M. van den Berg, E. Bolthausen, "Asymptotics of the generating function for the volume of the Wiener sausage" Probab. Th. Rel. Fields , 99 (1994) pp. 389–397
[a2] M. van den Berg, E. Bolthausen, F. den Hollander, "Moderate deviations for the volume of the Wiener sausage" Ann. of Math. (to appear in 2001)
[a3] M. van den Berg, B. Tóth, "Exponential estimates for the Wiener sausage" Probab. Th. Rel. Fields , 88 (1991) pp. 249–259
[a4] E. Bolthausen, "On the volume of the Wiener sausage" Ann. Probab. , 18 (1990) pp. 1576–1582
[a5] M.D. Donsker, S.R.S. Varadhan, "Asymptotics for the Wiener sausage" Commun. Pure Appl. Math. , 28 (1975) pp. 525–565
[a6] J.-F. Le Gall, "Sur une conjecture de M. Kac" Probab. Th. Rel. Fields , 78 (1988) pp. 389–402
[a7] J.-F. Le Gall, "Fluctuation results for the Wiener sausage" Ann. Probab. , 16 (1988) pp. 991–1018
[a8] F. Spitzer, "Electrostatic capacity, heat flow and Brownian motion" Z. Wahrsch. Verw. Gebiete , 3 (1964) pp. 110–121
[a9] A.-S. Sznitman, "Long time asymptotics for the shrinking Wiener sausage" Commun. Pure Appl. Math. , 43 (1990) pp. 809–820
[a10] A.-S. Sznitman, "Brownian motion, obstacles and random media" , Springer (1998)
[a11] Y. Hamana, H. Kesten, "A large deviation result for the range of random walk and for the Wiener sausage" preprint March (2000)
How to Cite This Entry:
Wiener sausage. Encyclopedia of Mathematics. URL: http://encyclopediaofmath.org/index.php?title=Wiener_sausage&oldid=11710
This article was adapted from an original article by F. den Hollander (originator), which appeared in Encyclopedia of Mathematics - ISBN 1402006098. See original article