Ageing, dynamical scaling and conformal invariance
Extracted source text ยท arXiv cond-mat/0302483
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% Latex2e file from Malte Henkel %%
%% %%
%% Version 1.0 -3 24.02.03 %%
%% %%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\documentclass[12pt]{article}
\usepackage{graphicx}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\input epsf.sty
% quelques abbreviations pratiques
\newcommand{\BEQ}{\begin{equation}} % Gleichungen Anfang ..
\newcommand{\BEA}{\begin{eqnarray}}
\newcommand{\EEQ}{\end{equation}} % .. und Ende
\newcommand{\EEA}{\end{eqnarray}}
\newcommand{\eps}{\varepsilon} % epsilon
\newcommand{\vph}{\varphi} % rundes phi
\newcommand{\vth}{\vartheta} % Deutsch-Delta
\newcommand{\D}{{\rm d}} % gerades d fuer Ableitungen
\newcommand{\II}{{\rm i}} % gerades i fuer komplexe Einheit
\renewcommand{\Re}{{\rm Re\ }} % Realteil
\renewcommand{\Im}{{\rm Im\ }} % Imaginaerteil
\newcommand{\arcosh}{{\rm arcosh\,}} % arcosh-Funktion
\newcommand{\erf}{{\rm erf\,}} % erf-Funktion
\newcommand{\wit}[1]{\widetilde{#1}} % weite Schlange
\newcommand{\rar}{\rightarrow} % Pfeil nach rechts
\renewcommand{\vec}[1]{\boldsymbol{#1}} % Vektoren fettgedruckt
\newcommand{\zeile}[1]{\vskip #1 \baselineskip} % N Zeilen ueberschlagen
% mit \zeile{N}
\newcommand{\vekz}[2]
{\mbox{${\begin{array}{c} #1 \\ #2 \end{array}}$}}
% \vekz{a}{b} erzeugt einen zweikomponentigen
% Vektor mit den Elementen a,b.
\begin{document}
\begin{titlepage}
%%{\hfill \tt \today; pr\'eliminaire !}
\vskip 1.5 cm
\begin{center}
{\Large \bf Ageing, dynamical scaling and conformal
invariance}\footnote{presented at the International Conference on Theoretical
Physics TH2002, Paris 22--27~july~2002}
\end{center}
\vskip 2.0 cm
\centerline{ {\bf Malte Henkel} }
\vskip 0.5 cm
\centerline {Laboratoire de Physique des
Mat\'eriaux\footnote{Laboratoire associ\'e au CNRS UMR 7556} et}
\centerline{
Laboratoire Europ\'een de Recherche Universitaire Sarre-Lorraine,}
\centerline{
Universit\'e Henri Poincar\'e Nancy I, B.P. 239,}
\centerline{
F -- 54506 Vand{\oe}uvre l\`es Nancy Cedex, France}
\begin{abstract}
\noindent
Building on an analogy with conformal invariance, local scale transformations
consistent with dynamical scaling are constructed.
Two types of local scale invariance are found which act as dynamical
space-time symmetries of certain non-local
free field theories. The scaling form of two-point functions is completely
fixed by the requirement of local scale invariance. These predictions
are confirmed through tests in the $3D$ ANNNI model at its Lifshitz point
and in ageing phenomena of simple ferromagnets, here studied through the
kinetic Ising model with Glauber dynamics.
\end{abstract}
\end{titlepage}
Scale invariance is a central notion of modern theories of critical and
collective phenomena. We are interested in systems with strongly anisotropic
or dynamical criticality. In these systems, two-point functions satisfy the
scaling form
\begin{equation} \label{gl:skala}
G(t,\vec{r}) = b^{2x} G(b^{\theta}t, b\vec{r}) = t^{-2x/\theta} \Phi\left(
r t^{-1/\theta}\right) = r^{-2x} \Omega\left( t r^{-\theta}\right)
\end{equation}
where $t$ stands for `temporal' and $\vec{r}$ for `spatial' coordinates,
$x$ is a scaling dimension, $\theta$ the anisotropy exponent (when $t$
corresponds to physical time, $\theta=z$ is called the dynamical exponent) and
$\Phi,\Omega$ are scaling functions. Physical realizations of this are numerous,
see \cite{Card96} and references therein. For isotropic critical systems,
$\theta=1$ and the `temporal' variable $t$ becomes just another coordinate.
It is well-known that in this case, scale invariance (\ref{gl:skala}) with
a constant rescaling factor $b$ can be replaced by the larger group
of conformal
transformations $b=b(t,\vec{r})$ such that angles are preserved. It turns
out that in the case of one space and one time dimensions, conformal invariance
becomes an important dynamical symmetry from which many physically relevant
conclusions can be drawn \cite{Bela84}.
Given the remarkable success of conformal invariance descriptions of
equilibrium phase transitions, one may wonder whether similar extensions of
scale invariance also exist when $\theta\neq 1$. Indeed, for $\theta=2$ the
analogue of the conformal group is known to be the Schr\"odinger group
\cite{Nied72,Hage72} (and apparently already known to Lie). While applications
of the Schr\"odinger group as dynamical space-time symmetry are known
\cite{Henk94}, we are interested here in the more general case when
$\theta\ne 1,2$. We shall first describe the construction of these
{\em local scale transformations}, show that they act as a dynamical
symmetry, then derive the functions $\Phi,\Omega$ and finally comment upon
some physical applications. For the sake of concisenes, we shall present
the main results formally as propositions. For details
we refer the reader to \cite{Henk02}.
The defining axioms of our notion of {\em local scale invariance} from which
our results will be derived, are as follows (for simplicity, in $d=1$ space
dimensions).
\begin{enumerate}
\item We seek space-time transformations with infinitesimal generators $X_n$,
such that time undergoes a M\"obius transformation
\BEQ \label{3:Moeb}
t\to t' = \frac{\alpha t + \beta}{\gamma t +\delta} \;\; ; \;\;
\alpha\delta - \beta\gamma =1
\EEQ
and we require that even after the action on the space coordinates is
included, the commutation relations
$\left[ X_n , X_m \right] = (n-m) X_{n+m}$
remain valid. This is motivated from the fact that this condition is
satisfied for both
conformal and Schr\"odinger invariance.
\item The generator $X_0$ of scale transformations is
$X_0 = - t \partial_t - {\theta}^{-1} r \partial_r - {x}/{\theta}$
with a scaling dimension $x$. Similarly, the generator of time translations
is $X_{-1}=-\partial_t$.
\item Spatial translation invariance is required.
\item Since the Schr\"odinger group acts on wave functions through a projective
representation, generalizations thereof should be expected to occur in the
general case. Such extra terms will be called {\em mass terms}. Similarly,
extra terms coming from the scaling dimensions should be present.
\item The generators when applied to a two-point function should yield
a finite number of independent conditions, i.e. of the form $X_n G=0$.
\end{enumerate}
\noindent
{\bf Proposition 1}: {\it Consider the generators}
\BEQ \label{gl:X}
X_n = - t^{n+1}\partial_t
- \sum_{k=0}^{n} \left(\vekz{n+1}{k+1}\right)
A_{k0}r^{\theta k +1} t^{n-k} \partial_r
- \sum_{k=0}^{n} \left(\vekz{n+1}{k+1}\right) B_{k0}r^{\theta k} t^{n-k}~~~
\EEQ
{\it where the coefficients $A_{k0}$ and $B_{k0}$ are given by the
recurrences $A_{n+1,0} = \theta A_{n0} A_{10}$,
$B_{n+1,0} = \frac{\theta}{n-1}\left( n B_{n0}A_{10} - A_{n0} B_{10}\right)$
for $n\geq 2$ where $A_{00}=1/\theta$, $B_{00}=x/\theta$ and in addition one of
the following conditions holds: (a) $A_{20}= \theta A_{10}^2$ (b)
$A_{10}=A_{20}=0$ (c) $A_{20}=B_{20}=0$ (d) $A_{10}=B_{10}=0$.
These are the most general linear first-order operators in
$\partial_t$ and $\partial_r$ consistent with the above axioms 1. and 2.
and which satisfy the commutation relations
$[X_{n},X_{n'}]=(n-n')X_{n+n'}$ for all $n,n'\in\mathbb{Z}$.}
\noindent
Closed but lengthy expressions of the $X_n$ for all $n\in\mathbb{Z}$ are
known \cite{Henk02}. In order to include space translations, we set
$\theta=2/N$ and use the short-hand
$X_n = -t^{n+1}\partial_t - a_n \partial_r - b_n$. We then define
\BEQ \label{gl:Y}
Y_m = Y_{k-N/2} = - \frac{2}{N(k+1)}\left(
\frac{\partial a_k(t,r)}{\partial r}\partial_r
+\frac{\partial b_k(t,r)}{\partial r} \right)
\EEQ
where $m=-\frac{N}{2} + k$ and $k$ is an integer.
Clearly, $Y_{-N/2}=-\partial_r$ generates space translations.
\noindent {\bf Proposition 2:}
{\it The generators $X_n$ and $Y_m$ defined in eqs.~(\ref{gl:X},\ref{gl:Y})
satisfy the commutation relations}
\BEQ \label{XYComm_II}
\left[ X_n , X_{n'} \right] = (n-n') X_{n+n'} \;\; , \;\;
\left[ X_n , Y_m \right] = \left( n \frac{N}{2} - m \right) Y_{n+m}
\EEQ
{\it in one of the following three cases:
(i) $B_{10}$ arbitrary, $A_{10}=A_{20}=B_{20}=0$ and $N$ arbitrary.
(ii) $B_{10}$ and $B_{20}$ arbitrary, $A_{10}=A_{20}=0$ and $N=1$.
(iii) $A_{10}$ and $B_{10}$ arbitrary, $A_{20}=A_{10}^2$,
$B_{20}=\frac{3}{2}A_{10} B_{10}$ and $N=2$.}
\noindent
In each case, the generators depend on two free parameters. The physical
interpretation of the free constants $A_{10},A_{20},B_{10},B_{20}$ is still
open. In the cases (ii) and (iii), the generators close into
a Lie algebra, see \cite{Henk02} for details. For case (i), a closed Lie
algebra exists if $B_{10}=0$.
Turning to the mass terms, we now restrict to the projective transformations
in time, because we shall only need those in the applications later.
It is enough to give merely the `special' generator $X_1$ which reads
for $B_{10}=0$ as follows \cite{Henk02}
\BEQ \label{gl:X1}
X_1 = -t^2\partial_t - N t r \partial_r - N x t -\alpha r^{2} \partial_t^{N-1}
- \beta r^{2} \partial_r^{2(N-1)/N} - \gamma \partial_r^{2(N-1)/N} r^{2}~~~
\EEQ
where $\alpha,\beta,\gamma$ are free parameters (the cases (ii,iii) of Prop. 2
do not give anything new). Furthermore, it turns out
that the relation $[X_1,Y_{N/2}]=0$ for $N$ integer is only satisfied in one
of the two cases (I) $\beta=\gamma=0$ which we call {\em Type~I}
and (II) $\alpha=0$ which we call {\em Type~II}.
In both cases, all generators can be obtained
by repeated commutators of $X_{-1}=-\partial_t$, $Y_{-N/2}=-\partial_r$ and
$X_1$, using (\ref{XYComm_II}). Commutators between two generators $Y_m$ are
non-trivial and in general only close on certain `physical' states. One might
call such a structure a {\em weak Lie algebra}.
These results depend on the construction \cite{Henk02} of {\em commuting}
fractional derivatives satisfying the rules
$\partial_r^{a+b} =\partial_r^a \partial_r^b$
and $[\partial_r^a,r]=a\partial_r^{a-1}$ (the standard Riemann-Liouville
fractional derivative is not commutative, see e.g. \cite{Hilf00}).
For $N=1$, the generators of both Type I and Type II reduce to those of the
Schr\"odinger group. For $N=2$, Type I reproduces the well-known generators of
$2D$ conformal invariance (without central charge) and Type II gives another
infinite-dimensional group whose Lie algebra is isomorphic to the one of
$2D$ conformal invariance \cite{Henk02}.
Dynamical symmetries can now be discussed as follows, by calculating the
commutator of the `Schr\"odinger-operator' $\cal S$ with $X_1$.
We take $d=1$ and $B_{10}=0$ for simplicity.
\noindent
{\bf Proposition 3:} {\it The realization of Type I sends any solution
$\psi(t,{r})$ with scaling dimension $x=1/2-(N-1)/N$ of the
differential equation}
\BEQ \label{WelleI}
\mathcal{S} \psi(t,\vec{r}) =
\left( -\alpha \partial_t^N + \left(\frac{N}{2}\right)^2
{\partial}_{r}^2 \right) \psi(t,{r}) = 0
\EEQ
{\it into another solution of the same equation.}
\noindent {\bf Proposition 4:} {\it The realization
of Type II sends any solution $\psi(t,r)$ with scaling dimension
$x=(\theta-1)/2+(2-\theta)\gamma/(\beta+\gamma)$ of the differential equation}
\BEQ \label{WelleII}
\mathcal{S} \psi(t,r) =
\left( -(\beta+\gamma) \partial_t + \frac{1}{\theta^2} \partial_r^{\theta}
\right) \psi(t,r) = 0
\EEQ
{\it into another solution of the same equation.}
\noindent
In both cases, $\cal S$ is a Casimir operator of the `Galilei'-subalgebra
generated from $X_{-1}, Y_{-N/2}$ and the generalized Galilei-transformation
$Y_{-N/2+1}$. The equations (\ref{WelleI},\ref{WelleII}) can be seen as
equations of motion of certain free field theories, where $x$ is the scaling
dimension of that free field $\psi$. These free field theories are
non-local, unless $N$ or $\theta$ are integers, respectively.
{}From a physical point of view, these wave equations suggest that the
applications of Types I and II are very different. Indeed, eq.~(\ref{WelleI})
is typical for equilibrium systems with a scaling anisotropy introduced through
competing uniaxial interactions. Paradigmatic cases of this are so-called
Lifshitz points which occur for example in magnetic systems when an
ordered ferromagnetic, a disordered paramagnetic and an incommensurate phase
meet (see \cite{Dieh02} for a recent review). On the other hand,
eq.~(\ref{WelleII}) is reminiscent of a Langevin equation which may describe
the temporal evolution of a physical system, with a dynamical exponent
$z=\theta$. In any case, causality requirements
can only be met by an evolution equation of first order in $\partial_t$.
Next, we find the scaling functions $\Phi,\Omega$ in eq.~(\ref{gl:skala}) from
the assumption that $G$ transforms covariantly under local scale
transformations.
\noindent
{\bf Proposition 5:} {\it Local scale invariance implies that for Type I,
the function $\Omega(v)$ must satisfy}
\BEQ \label{gl:Omega}
\left( \alpha \partial_v^{N-1} - v^2 \partial_v - N x \right) \Omega(v) = 0
\EEQ
{\it together with the boundary conditions $\Omega(0)=\Omega_0$ and
$\Omega(v)\sim\Omega_{\infty} v^{-Nx}$ for $v\to\infty$. For Type II, we have}
\BEQ \label{gl:Phi}
\left( \partial_u +\theta(\beta+\gamma)u\partial_u^{2-z}
+2\theta(2-z)\gamma\partial_u^{1-z}\right)\Phi(u)=0
\EEQ
{\it with the boundary conditions $\Phi(0)=\Phi_0$ and
$\Phi(u)\sim\Phi_{\infty}u^{-2x}$ for $u\to\infty$.}
\noindent
Here $\Omega_{0,\infty}$ and $\Phi_{0,\infty}$ are constants. The ratio
$\beta/\gamma$ turns out to be universal and related to $x$.
From the linear differential equations (\ref{gl:Omega},\ref{gl:Phi})
the scaling functions $\Omega(v)$ and $\Phi(u)$ can be found
explicitly using standard methods \cite{Henk02}.
%%----------------------------------------------------------------------------%%
\begin{figure}[h]
\centerline{\epsfxsize=3.65in\epsfbox
{th2002_fig1.eps}
}
\caption{Scaling function $\Psi(v)$ of the spin-spin correlator at the
Lifshitz point in the $3D$ ANNNI model. The main plot the scaling of selected
data for several values of $r_{\perp}$ is shown. In the inset, the full set of
numerical data (gray points) is compared to the solution of
eq.~(\ref{gl:Omega}). The data are from {\protect \cite{Plei01}}.
\label{Bild1}}
\end{figure}
%%----------------------------------------------------------------------------%%
Given these explicit results, the idea of local scale invariance can be
tested in specific models. Indeed, the predictions for $\Omega(v)$ coming from
Type I with $N=4$ nicely agree with cluster Monte Carlo data \cite{Plei01}
for the spin-spin and energy-energy correlators of the $3D$ ANNNI model at its
Lifshitz point. Extensive simulations led to a considerable
improvement in the precision of the estimation of the location of the
Lifshitz point and the Lifshitz point exponents
$\alpha_L=0.18(2), \beta_L=0.238(5),\gamma_L=1.36(3)$, respectively,
of the specific heat, the magnetization and the susceptibility were
obtained \cite{Plei01}.
These values are in good agreement with the results of a careful two-loop
study of the ANNNI model within renormalized field theory \cite{Shpo01}.
In figure~\ref{Bild1}, we show the scaling function
$\Psi(v) := v^{\zeta} \Omega(v)$ of the spin-spin correlator. Here
$\zeta=(2/\theta)(d_{\perp}+\theta)/(2+\gamma_L/\beta_L)$, where $d_{\perp}=2$
is the number of transverse dimensions without
competing terms and $\theta\simeq 1/2$ within the numerical accuray of the
data (see \cite{Plei01,Shpo01,Henk02} for a fuller discussion of this point).
Clearly, the data agree with the prediction of local scale invariance.
On the other hand, the predictions of
Type II have been tested extensively in the context of ageing ferromagnetic
spin systems to which we turn now.
Consider a ferromagnetic spin system (e.g. an Ising model) prepared in a
high-temperature initial state and then quenched to some temperature $T$ at or
below the critical temperature $T_c$. Then the system is left to evolve freely
(for recent reviews, see \cite{Cate00,Godr02}).
It turns out that clusters of a typical time-dependent size $L(t)\sim t^{1/z}$
form and grow, where $z$ is the dynamical exponent. Furthermore, two-time
observables such as the response function $R(t,s;\vec{r}-\vec{r}')=
\delta\langle\sigma_{\vec{r}}(t)\rangle/\delta h_{\vec{r}'}(s)$ depend on
{\em both} $t$ and $s$, where
$\sigma_{\vec{r}}$ is a spin variable and $h_{\vec{r}'}$ the conjugate magnetic
field. This breaking of time-translation
invariance is called {\em ageing}. We are mainly interested in the
autoresponse function $R(t,s)=R(t,s;\vec{0})$. One finds a dynamic scaling
behaviour $R(t,s)\sim s^{-1-a} f_R(t/s)$ with $f_R(x)\sim x^{-\lambda_R/z}$
for $x\gg 1$ and where $\lambda_R$ and $a$ are exponents to be determined.
In order to apply local scale invariance
to this problem, we must take into account that time translation invariance
does {\em not} hold. The simplest way to do this is to remark that the
Type~II-subalgebra spanned by $X_0, X_1$ and the $Y_m$ leaves the initial
line $t=0$ invariant, see (\ref{gl:X1}). Therefore the autoresponse function
$R(t,s)$ is fixed by the two covariance conditions $X_0 R = X_1 R=0$.
%%This leads to
\noindent
{\bf Proposition 6:}
{\it For a statistical non-equilibrium system which
satisfies local scale invariance the autoresponse function $R(t,s)$ takes the
form}
\BEQ \label{gl:R}
R(t,s) = r_0 \left( t/s \right)^{1+a-\lambda_R/z} \left( t-s\right)^{-1-a}
\;\; , \;\; t > s~~~~
\EEQ
{\it where $a$ and $\lambda_R/z$ are non-equilibrium exponents and $r_0$ is a
normalization constant.}
\noindent The functional form of $R$
is completely fixed once the exponents
$a$ and $\lambda_R/z$ are known. Similarly,
$R(t,s;\vec{r})=R(t,s) \Phi\left(r (t-s)^{-1/z}\right)$ gives the
spatio-temporal response, with the
scaling function $\Phi(u)$ determined by (\ref{gl:Phi}) \cite{Henk02}.
The reader might wonder whether invariance under the special transformation
generated by $X_1$ actually holds in spite of the absence of time translation
invariance. An answer to this question can be given \cite{Henk02c} for ageing
systems {\em below} criticality, where $z=2$ for a non-conserved order
parameter. In this case, the fluctuations in the system merely come from the
disordered initial conditions and the presence of these must be taken into
account when considering the transformation of the effective action $S$
(in the context of Martin-Siggia-Rose theory, see e.g. \cite{Card96} and refs
therein).
\noindent
{\bf Proposition 7:} \cite{Henk02c} {\it Consider an ageing system described
by a local action $S$ and which is invariant under spatial translations,
phase shifts, Galilei transformations and dilatations with $z=2$.
Then $\delta S=0$ under the action of the special Schr\"odinger
transformation $X_1$.}
A quantitative test of the prediction (\ref{gl:R}) is best carried out through
the consideration of an integrated response function, such as the
thermoremanent magnetization $M_{\rm TRM}(t,s)=h\int_{0}^{s} \!\D u\,R(t,u)$,
where the system is quenched in a small magnetic field $h$ which is turned off
after the waiting time $s$ has elapsed. The magnetization is then measured at
a later time $t$. One has the following scaling form for $s$
large \cite{Henk02a}
\BEQ \label{gl:MTRM}
M_{\rm TRM}(t,s)/h = r_0\, s^{-a} f_M(t/s) + r_1\, s^{-\lambda_R/z} g_M(t/s)
\EEQ
Local scale invariance furnishes, via (\ref{gl:R}),
explicit predictions for the
scaling functions $f_M(x)$ and $g_M(x)$ \cite{Henk02a} where $r_{0,1}$ are
normalization constants. In concrete systems
such as the $2D$ Glauber-Ising model where exponents $a=1/z=1/2$ and
$\lambda_R/z\sim 0.6-0.8$, the two terms are often of the same order
of magnitude and one might expect
to find a strong cross-over behaviour in the scaling of $M_{\rm TRM}$.
%%----------------------------------------------------------------------------%%
\begin{figure}[ht]
\centerline{\epsfxsize=1.75in\ \epsfbox{
th2002_fig3a.eps} ~~~~~~
\epsfxsize=1.75in\ \epsfbox{
th2002_fig3b.eps}
\epsfxsize=1.55in\ \epsfbox{
th2002_fig3c.eps}
}
\caption{Scaling of the thermoremanent magnetization $M_{\rm TRM}(t,s)$ as
a function of $s$. Data for the $3D$ spherical model with initial
average magnetization $M_0=0.5$ (left) and $M_0=0$ (middle) are shown, with
$x=t/s=5$ and the exponents $a$ and $\lambda_R/z$ taken from \cite{Pico02}.
At the right, data from the $2D$ Glauber-Ising model at $T=0$
and with the exponents $a=1/2$ and $\lambda_R=1.54$ are displayed.
The values of $x$ are 3 (circles), 5 (squares), 7 (diamonds).
The data are from \cite{Henk02a}.
\label{Bild3}}
\end{figure}
%%----------------------------------------------------------------------------%%
We now illustrate this crossover explictly. In the left and middle panel of
figure~\ref{Bild3} \cite{Henk02a}, we display $M_{\rm TRM}(t,s)$ as a function
of the waiting time $s$ for the the kinetic $3D$ spherical model with
a non-conserved order parameter, for two values of the initial
magnetization $M_0$. Here, the full curve gives the exact solution of
the Langevin equation, the dash-dotted line gives the asymptotic scaling
according to (\ref{gl:MTRM}) and the dotted and dashed lines give the
individual contributions. The cross-over between a contribution
$\sim s^{-\lambda_R/z}$
for $s$ small and a contribution $\sim s^{-a}$ for $s$ large is clearly seen.
In the right panel of figure~\ref{Bild3},
analogous data for the $2D$ Glauber-Ising model are
shown. The full curves are the predictions of local scale invariance where
the constants $r_{0,1}$ were fixed using the data for $x=3$. Then, for any
different value of $x$, a definite prediction without any free parameter at all
is obtained. We compare this with data for $x=5$ and $x=7$ and find a nice
agreement. We point out that our data disagree with
a recent suggestion \cite{Corb02} that the exponent $a=1/4$ in the
$2D$ Glauber-Ising model at $T<T_c$.
%%----------------------------------------------------------------------------%%
\begin{figure}[h]
\centerline{\epsfxsize=3.65in\epsfbox
{th2002_fig2.eps}
}
\caption{Scaling of the thermoremanent magnetization
$M_{\rm TRM}(t,s)/h=\rho(t,s)=s^{-a}\rho(t/s)$ as
a function of $x=t/s$ in the $3D$ Glauber-Ising model, at (left) $T=3$ and
(right) $T=T_c$. The full curve is the prediction obtained by integrating
{\protect eq.~(\ref{gl:R})}. The data are from \cite{Henk01}.
\label{Bild2}}
\end{figure}
%%----------------------------------------------------------------------------%%
The prediction (\ref{gl:R}) for the autoresponse has been confirmed
in several physically distinct systems undergoing ageing,
see \cite{Godr02,Henk01,Cann01,Cala02,Pico02} and
references therein. As an example, we show data from the $3D$ Glauber-Ising
model, both below and at criticality, in figure~\ref{Bild2}. The
expected scaling behaviour $M_{\rm TRM}(t,s)/h = s^{-a} \rho(t/s)$ is well
realized and the form of $\rho(x)$ agrees nicely with local scale
invariance, where the values of $a$ and of $\lambda_R/z$ were taken from the
literature, see \cite{Henk01,Henk02} and references therein.
These confirmations (which do go beyond free field theory)
suggest that (\ref{gl:R}) should hold
independently of (i) the value of the dynamical exponent $z$ (ii) the spatial
dimensionality $d>1$ (iii) the numbers of components of the order parameter and
the global symmetry group (iv) the spatial range of the interactions (v) the
presence of spatially long-range initial correlations (vi) the value of the
temperature $T$ (vii) the presence of weak disorder. Very recently,
the space-time response functions $R(t,s;\vec{r})$ were calculated in the
$2D$ and $3D$ Glauber-Ising model after a quench to below the critical point.
Our results agree perfectly with the prediction of local scale invariance with
$z=2$ \cite{Henk02b}.\footnote{A recent two-loop calculation \cite{Cala03} in
the kinetic O($N$)-model {\em at\/} criticality recovers (\ref{gl:R}), up to
a small correction. A detailed discussion will be given elsewhere.}
Summarizing, we have shown that local scale transformations exist for any
value of $z$ or $\theta$, act as dynamical symmetries of certain non-local
free field theories and appear to be realized as space-time symmetries in
some strongly anisotropic critical systems of physical interest. We also showed
how the theory of local scale invariance may be applied to resolve
standing questions in ageing phenomena.
It is a pleasure to thank M. Pleimling for a yearlong fruitful collaboration,
C. Godr\`eche, J.-M. Luck, M. Pae{\ss}ens, A. Picone and J. Unterberger for
their collaboration on the work described here and P. Calabrese, H.W. Diehl,
A. Gambassi and M. Shpot for useful discussions or correspondence.
This work was supported by CINES Montpellier (projet pmn2095) and the
Bayerisch-Franz\"osisches Hochschulzentrum (BFHZ).
{\small
\begin{thebibliography}{999}
\bibitem{Card96} J.L. Cardy, {\it Scaling and Renormalization in
Statistical Mechanics}, Cambridge University Press (1996)
\bibitem{Bela84} A.A. Belavin, A.M. Polyakov and
A.B. Zamolodchikov, Nucl. Phys. {\bf B241}, 333 (1984)
\bibitem{Nied72} U. Niederer, Helv. Phys. Acta {\bf 45}, 802 (1972)
\bibitem{Hage72} C.R. Hagen, Phys. Rev. {\bf D5}, 377 (1972)
\bibitem{Henk94} M. Henkel, J. Stat. Phys. {\bf 75}, 1023 (1994)
\bibitem{Henk02} M. Henkel, Nucl. Phys. {\bf B641}, 405 (2002)
\bibitem{Hilf00} R. Hilfer (ed), {\it Applications of Fractional Calculus in
Physics}, World Scientific (Singapore 2000)
\bibitem{Dieh02} H.W. Diehl, Acta physica slovaka {\bf 52}, 271 (2002)
\bibitem{Plei01} M. Pleimling and M. Henkel, Phys. Rev. Lett. {\bf 87}, 125702
(2001)
\bibitem{Shpo01} M. Shpot and H.W. Diehl, Nucl. Phys. {\bf B612}, 340 (2001)
\bibitem{Cate00} M.E. Cates and M.R. Evans (eds), {\it Soft and Fragile
Matter}, IOP (Bristol 2000).
\bibitem{Godr02} C. Godr\`eche and J.-M. Luck, J. Phys. Cond. Matt. {\bf 14},
1589 (2002)
\bibitem{Henk02c} M. Henkel and J. Unterberger, {\tt hep-th/0302187}
\bibitem{Henk02a} M. Henkel, M. Paessens and M. Pleimling,
{\tt cond-mat/0211583}
\bibitem{Pico02} A. Picone and M. Henkel, J. Phys. {\bf A35}, 5575 (2002)
\bibitem{Corb02} F. Corberi, E. Lippiello and M. Zannetti, Eur. Phys. J.
{\bf B24}, 359 (2001); Phys. Rev. {\bf E65}, 046136 (2002);
{\tt cond-mat/0211609}
\bibitem{Henk01} M. Henkel, M. Pleimling, C. Godr\`eche and J.-M. Luck,
Phys. Rev. Lett. {\bf 87}, 265701 (2001)
\bibitem{Cann01} S.A. Cannas, D.A. Stariolo and F.A. Tamarit,
Physica {\bf A294}, 362 (2001)
\bibitem{Cala02} P. Calabrese and A. Gambassi, Phys. Rev. {\bf E65},
066120 (2002); {\bf B66}, 212407 (2002)
\bibitem{Henk02b} M. Henkel and M. Pleimling, {\tt cond-mat/0302482}
\bibitem{Cala03} P. Calabrese and A. Gambassi, Phys. Rev. {\bf E66},
066101 (2002); {\tt cond-mat/0211062}
\end{thebibliography}
}
\end{document}
%% End of file th2002.tex