Professional Documents
Culture Documents
Textbook Reflected Brownian Motions in The KPZ Universality Class 1St Edition Thomas Weiss Ebook All Chapter PDF
Textbook Reflected Brownian Motions in The KPZ Universality Class 1St Edition Thomas Weiss Ebook All Chapter PDF
Textbook Reflected Brownian Motions in The KPZ Universality Class 1St Edition Thomas Weiss Ebook All Chapter PDF
https://textbookfull.com/product/the-oxford-handbook-on-the-
united-nations-thomas-g-weiss/
https://textbookfull.com/product/motions-of-ice-hydrometeors-in-
the-atmosphere-numerical-studies-and-implications-pao-k-wang/
https://textbookfull.com/product/the-united-nations-and-changing-
world-politics-revised-and-updated-with-a-new-introduction-8th-
edition-thomas-g-weiss/
https://textbookfull.com/product/the-new-politics-of-class-in-
britain-the-political-exclusion-of-the-working-class-1st-edition-
evans/
America Inc Innovation and Enterprise in the National
Security State 1st Edition Linda Weiss
https://textbookfull.com/product/america-inc-innovation-and-
enterprise-in-the-national-security-state-1st-edition-linda-
weiss/
Sorting Out the New South City Race Class and Urban
Development in Charlotte 1875 1975 2nd Edition Thomas
W. Hanchett
https://textbookfull.com/product/sorting-out-the-new-south-city-
race-class-and-urban-development-in-charlotte-1875-1975-2nd-
edition-thomas-w-hanchett/
https://textbookfull.com/product/brownian-motion-martingales-and-
stochastic-calculus-1st-edition-jean-francois-le-gall/
https://textbookfull.com/product/brownian-motion-martingales-and-
stochastic-calculus-1st-edition-jean-francois-le-gall-2/
https://textbookfull.com/product/class-on-screen-the-global-
working-class-in-contemporary-cinema-sarah-attfield/
SPRINGER BRIEFS IN MATHEMATIC AL PHYSICS 18
Thomas Weiss
Patrik Ferrari
Herbert Spohn
Reflected
Brownian
Motions in the
KPZ Universality
Class
123
SpringerBriefs in Mathematical Physics
Volume 18
Series editors
Nathanaël Berestycki, Cambridge, UK
Mihalis Dafermos, Cambridge, UK
Tohru Eguchi, Tokyo, Japan
Atsuo Kuniba, Tokyo, Japan
Matilde Marcolli, Pasadena, USA
Bruno Nachtergaele, Davis, USA
SpringerBriefs are characterized in general by their size (50-125 pages) and fast
production time (2-3 months compared to 6 months for a monograph).
Briefs are available in print but are intended as a primarily electronic publication to
be included in Springer’s e-book package.
Herbert Spohn
123
Thomas Weiss Herbert Spohn
Zentrum Mathematik Zentrum Mathematik, M5
Technische Universität München Technische Universität München
Garching Munich
Germany Germany
Patrik Ferrari
Institut für Angewandte Mathematik
Universität Bonn
Bonn
Germany
v
Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2 One-Sided Reflected Brownian Motions and Related Models . . . . . . . 9
3 Determinantal Point Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4 Airy Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
5 Packed and Periodic Initial Conditions . . . . . . . . . . . . . . . . . . . . . . . . . 45
6 Stationary Initial Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
7 More General Initial Conditions and Their Asymptotics . . . . . . . . . . . 97
vii
Chapter 1
Introduction
with W (x, t) normalized space-time white noise. We use here units of height, space,
and time such that all coupling parameters take a definite value. For a solution
of (1.1), the function x → h(x, t) is locally like a Brownian motion, which is
too singular for the nonlinearity to be well-defined as written. This difficulty was
© The Author(s) 2017 1
T. Weiss et al., Reflected Brownian Motions in the KPZ Universality Class,
SpringerBriefs in Mathematical Physics, DOI 10.1007/978-3-319-49499-9_1
2 1 Introduction
resolved through the regularity structures of Hairer (2013), see also Gubinelli and
Perkowski (2015) for a somewhat different approach. Kardar (1987) noted one link to
quantum integrability. He considered the moments of eh and established that they are
related to the δ-Bose gas with attractive interactions, which is an integrable quantum
many-body system (Lieb and Liniger 1963). More precisely, one defines
n
E eh(yα ,t) = f t (y ) (1.2)
α=1
n
n
Hn = − 21 ∂ y2α − 1
2
δ(yα − yα ). (1.4)
α=1 α=α =1
Almost thirty years later the generator of the asymmetric simple exclusion process
(ASEP) was diagonalized through the Bethe ansatz. In case of N sites, the ASEP
configuration space is {0, 1} N signalling a similarity with quantum spin chains. In
fact, the ASEP generator can be viewed as the Heisenberg chain with an imagi-
nary XY-coupling. For the totally asymmetric limit (TASEP) and half filled lattice,
Gwa and Spohn (1992) established that the spectral gap of the generator is of order
N −3/2 . The same order is argued for the KPZ equation. This led to the strong belief
that, despite their very different set-up, both models have the same statistical prop-
erties on large space-time scales. In the usual jargon of statistical mechanics, both
models are expected to belong to the same universality class, baptized KPZ univer-
sality class according to its most prominent representative.
The KPZ equation is solved with particular initial conditions. Of interest are
(i) sharp wedge, h(x, 0) = −c0 |x| in the limit c0 → ∞, (ii) flat, h(x, 0) = 0,
and (iii) stationary, h(x, 0) = B(x) with B(x) two sided Brownian motion. The
quantity of prime interest is the distribution of h(0, t) for large t. More ambitiously,
but still feasible in some models, is the large time limit of the joint distribution of
{h(xα , t), α = 1, ..., n}.
In our notes we consider an integrable system of interacting diffusions, which is
governed by the coupled stochastic differential equations
where {B j (t)} a collection of independent standard Brownian motions. For the para-
meter β > 0 we will eventually consider only the limit β → ∞. But for the purpose
of our discussion we keep β finite for a while. In fact there is no choice, no other
system of this structure is known to be integrable. The index set depends on the prob-
lem, mostly we choose j ∈ Z. Note that x j interacts only with its left index neighbor
1 Introduction 3
x j−1 . The drift depends on the slope, as it should be for a proper height function.
But the exponential dependence on x j − x j−1 is very special, however familiar from
other integrable systems. The famous Toda chain (Toda 1967) is a classical integrable
system with exponential nearest neighbour interaction. Its quantized version is also
integrable (Sutherland 1978).
Interaction with only the left neighbor corresponds to the total asymmetric version.
Partial asymmetry would read
dx j (t) = pβe−β(x j −x j−1 ) − (1 − p)βe−β(x j+1 −x j ) dt + dB j (t) (1.6)
d n
n
β −2 f t (m)
= ∂α f t (m)
+ 1
2
δ(m α − m α ) f t (m)
− n f t (m),
(1.8)
dt α=1 α,α =1
where δ is the Kronecker delta and ∂α f (m) = f (..., m α , ...) − f (..., m α − 1, ...).
When comparing with (1.4), instead of ∂α one could have guessed the discrete Lapla-
cian α = −∂αT ∂α with T denoting the transpose. To obtain such a result, the drift
in (1.5) would have to be replaced by βe−β(x j −x j−1 ) + βe−β(x j −x j+1 ) . But the linear
equations for the exponential moments are no longer Bethe integrable. Note however
that the semigroups exp[∂α t] and exp[α t] differ on a large scale only by a uniform
translation proportional to t. With such a close correspondence one would expect
system (1.5) to be in the KPZ universality class for any β > 0, as has been verified to
some extent (Amir et al. 2011; Sasamoto and Spohn 2010; Borodin et al. 2015). Also
partial asymmetry, 21 < p < 1, should be in the KPZ universality class. In fact, the
true claim is by many orders more sweeping: the exponential in (1.6), p = 21 , can be
replaced by “any” function of x j − x j−1 , except for the linear one, and the system is
still in the KPZ universality class. There does not seem to be a promising idea around
of how to prove such a property. Current techniques heavily rely on integrability.
4 1 Introduction
where
( j) 1 /2+ξw
w− j
2
Φt (ξ, t) = dw etw (1.10)
2πi iR+δ
with δ > 0 as first established by Sasamoto and Wadati (1998). There is a similar
formula for the TASEP (Schütz 1997). Such formuli nourish the hope to uncover
interesting features of the model.
1 Introduction 5
The three initial conditions of particular interest, wedge, flat, and stationary, are
easily transcribed to system (1.5) and become (i) packed, half-infinite system with
x j (0) = 0 for j = 1, 2, ..., (ii) periodic, x j (0) = j for j ∈ Z, (iii) Poisson,
{x j (0), j ∈ Z} is a Poisson process with constant density. According to our discus-
sion, in the latter case one might think that the quantity of prime interest is x1 (t). But
the reflection induces a propagation of statistical fluctuations, as can also be seen
from (1.7) together with (1.8). Their propagation speed is 1 and the correct quantity
is x t (t) with t denoting integer part. Along other space-time observation rays a
central limit type behavior would be observed.
As for other models in the KPZ universality class, our asymptotic analysis is lim-
ited to a single time and arbitrary number of spatial, resp. index points. Only recently
Johansson (2016) posted a result on the joint distribution of (x t (t), x αt (αt)),
α > 0, and identified its universal limit. Possibly such progress will lead eventually
to a complete understanding of the Airy sheet and the KPZ fixed point (Corwin et al.
2015).
Following the format of Springerbriefs, there is a short summary for each chapter.
But let us still explain of how our material is organized as a whole. The following
three chapters provide background material. In Chap. 2 we properly define the infinite
system of reflected Brownian motions as the solution of a martingale problem and
provide a variational formulation of this solution. Also the uniform Poisson process is
identified as stationary measure. In Chap. 3 we introduce the theory of determinantal
point processes and some related material on Fredholm determinants. At first sight
this looks unconnected. But to study the quantities of prime interest one first identifies
a “hidden” signed determinantal process which leads to an analytically more tractable
representation. In the long time limit we will arrive at a stochastic process which
describes the limiting spatial statistics. Such a process has been baptised Airy process,
in analogy to the Airy kernel and Airy operator which are one of the defining elements.
In fact there are several Airy processes depending on the initial conditions and on
the window of observation. The literature on Airy processes is somewhat dispersed.
Chapter 4 provides a streamlined account.
In Chap. 5 we investigate the two deterministic initial data, packed and periodic,
while in Chap. 6 we study random initial data as defined through a Poisson process.
These results will be used to discuss more general initial data, which should be
viewed as an open ended enterprise. One natural choice is to have in the left half
lattice either packed, periodic, or Poisson join up with either one of them in the right
half lattice. An example would be to have periodic to the left and Poisson to the
right. This then leads to distinct cross over processes. The mixed cases are studied
in Chap. 7. Slow decorrelation, referring to space-like paths more general than fixed
time, is a further topic. Each core chapter builds on a suitable asymptotic analysis.
In the early days the required techniques were developed ad hoc for the particular
model. Over the years a common strategy based on contour integrations has been
established, which will be also used here. Thus on the basis of a specific example
one can learn a technique applicable also to other models.
6 1 Introduction
References
H. Bethe. Zur Theorie der Metalle. I. Eigenwerte und Eigenfunktionen der linearen Atomkette.
Zeitschrift für Physik 71, 205–226 (1931)
E.H. Lieb, W. Liniger, Exact analysis of an interacting Bose gas. I. the general solution and the
ground state. Phys. Rev. 130, 1605–1616 (1963)
C.N. Yang, Some exact results for the many-body problem in one dimension with repulsive delta-
function interaction. Phys. Rev. Lett. 19, 1312–1315 (1967)
B. Sutherland, Beautiful Models. 70 Years of Exactly Solved Quantum Many-Body Problems (World
Scientific Publishing, New Jersey, 2004)
M.T. Batchelor, The Bethe ansatz after 75 years. Phys. Today 60, 36 (2007)
J. Simon, W.S. Bakr, R. Ma, M.E. Tai, P.M. Preiss, M. Greiner, Quantum simulation of antiferro-
magnetic spin chains in an optical lattice. Nat. 472, 307–312 (2011)
M. Kardar, G. Parisi, Y.Z. Zhang, Dynamic scaling of growing interfaces. Phys. Rev. Lett. 56,
889–892 (1986)
A.L. Barabasi, H.E. Stanley, Fractal Concepts in Surface Growth (Cambridge University Press,
Cambridge, 1995)
P. Meakin, Fractals Scaling and Growth Far from Equilibrium (Cambridge University Press, Cam-
bridge, 1998)
K. Johansson, Random matrices and determinantal processes. Math. Stat. Phys. Sess. LXXXIII:
Lect. Notes Les Houches Summer Sch. 2005, 1–56 (2006)
H. Spohn, Exact solutions for KPZ-type growth processes, random matrices, and equilibrium shapes
of crystals. Phys. A 369, 71–99 (2006)
J. Quastel, Introduction to KPZ. Curr. Dev. Math. 2011, 125–194 (2011)
A. Borodin and V. Gorin. Lectures on integrable probability, (2012). arXiv:1212.3351
A. Borodin, L. Petrov, Integrable probability: from representation theory to Macdonald processes.
Probab. Surv. 11, 1–58 (2014)
H. Spohn. The Kardar-Parisi-Zhang equation - a statistical physics perspective. in Stochastic
Processes and Random Matrices, École d’Été Physique, Les Houches (Oxford University Press,
2015). arXiv:1601.00499
T. Halpin-Healy, Y.-C. Zhang, Kinetic roughening phenomena, stochastic growth, directed polymers
and all that. Phys. Rep. 254, 215–414 (1995)
J. Krug, Origins of scale invariance in growth processes. Adv. Phys. 46, 139–282 (1997)
T. Sasamoto and H. Spohn, The 1+1-dimensional Kardar-Parisi-Zhang equation and its universality
class. J. Stat. Mech. P01031 (2011)
P.L. Ferrari and H. Spohn. Random growth models. in The Oxford Handbook of Random Matrix
Theory, ed. By J. Baik, G. Akemann, P. Di Francesco (2011)
I. Corwin, The Kardar-Parisi-Zhang equation and universality class. Random Matrices Theory Appl.
1, 1130001 (2012)
K. Takeuchi, Experimental approaches to universal out-of-equilibrium scaling laws: turbulent liquid
crystal and other developments. J. Stat. Mech. P01006 (2014)
J. Quastel, H. Spohn, The one-dimensional KPZ equation and its universality class. J. Stat. Phys.
160, 965–984 (2015)
T. Halpin-Healy, K. Takeuchi, A KPZ cocktail-shaken, not stirred: Toasting 30 years of kinetically
roughened surfaces. J. Stat. Phys. 160, 794–814 (2015)
M. Hairer, Solving the KPZ equation. Ann. Math. 178, 559–664 (2013)
M. Gubinelli and N. Perkowski. KPZ reloaded, (2015). arXiv:1508.03877
M. Kardar, Replica Bethe ansatz studies of two-dimensional interfaces with quenched random
impurities. Nucl. Phys. B 290, 582–602 (1987)
L.-H. Gwa, H. Spohn, Bethe solution for the dynamical-scaling exponent of the noisy Burgers
equation. Phys. Rev. A 46, 844–854 (1992)
M. Toda, Vibration of a chain with nonlinear interaction. J. Phys. Soc. Jpn. 22, 431 (1967)
References 7
B. Sutherland, A brief history of the quantum soliton with new results on the quantization of the
toda lattice. Rocky Mt. J. Math. 8, 413–430 (1978)
M.Z. Guo, G.C. Papanicolaou, S.R.S. Varadhan, Nonlinear diffusion limit for a system with nearest
neighbor interactions. Commun. Math. Phys. 118, 31–59 (1988)
C.-C. Chang, H.-T. Yau, Fluctuations of one dimensional Ginzburg-Landau models in nonequilib-
rium. Commun. Math. Phys. 145, 209–234 (1992)
G. Amir, I. Corwin, J. Quastel, Probability distribution of the free energy of the continuum directed
random polymer in 1+1 dimensions. Commun. Pure Appl. Math. 64, 466–537 (2011)
T. Sasamoto, H. Spohn, Exact height distributions for the KPZ equation with narrow wedge initial
condition. Nucl. Phys. B 834, 523–542 (2010)
A. Borodin, I. Corwin, P.L. Ferrari, B. Vető, Height fluctuations for the stationary KPZ equation.
Math. Phys. Anal. Geom. 18, 1–95 (2015)
P. Calabrese, P. Le Doussal, A. Rosso, Free-energy distribution of the directed polymer at high
temperature. EPL 90, 20002 (2010)
T. Imamura, T. Sasamoto, Stationary correlations for the 1D KPZ equation. J. Stat. Phys. 150,
908–939 (2013)
V. Dotsenko, Replica Bethe ansatz derivation of the Tracy-Widom distribution of the free energy
fluctuations in one-dimensional directed polymers. J. Stat. Mech. P07010, (2010)
A. Borodin and I. Corwin, Macdonald processes. Probab. Theory Relat. Fields (online first), (2013)
T. Sasamoto, H. Spohn, Point-interacting Brownian motions in the KPZ universality class. Electron.
J. Probab. 20(87), 28 (2015)
I. Karatzas, S. Pal, M. Shkolnikov, Systems of Brownian particles with asymmetric collisions. Ann.
Inst. H. Poincaré Probab. Stat. 52, 323–354 (2016)
T.E. Harris, Diffusion with collisions between particles. J. Appl. Probab. 2, 323–338 (1965)
T. Sasamoto, M. Wadati, Determinantal form solution for the derivative nonlinear Schrödinger type
model. J. Phys. Soc. Jpn. 67, 784–790 (1998)
G.M. Schütz, Exact solution of the master equation for the asymmetric exclusion process. J. Stat.
Phys. 88, 427–445 (1997)
K. Johansson. Two time distribution in Brownian directed percolation. Comm. Math. Phys., online
first:1–52, (2016)
I. Corwin, J. Quastel, D. Remenik, Renormalization fixed point of the KPZ universality class. J.
Stat. Phys. 160, 815–834 (2015)
Chapter 2
One-Sided Reflected Brownian Motions
and Related Models
For reflected Brownian motions with index set Z+ we properly define the dynamics
using an iterated Skorokhod construction. The full index set Z requires a proof
of well-posedness. In addition, we establish that the uniform Poisson process is
stationary under the dynamics and discuss previous results for the model.
n
Yk,n (t) = sup (Bi (si ) − Bi (si−1 )) (2.1.3)
0≤sk ≤...≤sn−1 ≤t i=k
for k ≤ n, with the convention sk−1 = 0 and sn = t, allows for an equivalent explicit
expression:
xn (t) = max {Yk,n (t) + ζk }. (2.1.4)
k∈[1,n]
A canonical and in fact the most studied initial condition for the system {xn (t), n ≥ 1}
is the one where all particles start at zero. We call this packed initial condition:
n
Yk−1,n (t) = sup (Bi (si ) − Bi (si−1 ))
0≤sk−1 ≤sk ≤...≤sn−1 ≤t i=k−1
(2.2.2)
n
≥ sup (Bi (si ) − Bi (si−1 )) = Yk,n (t),
0=sk−1 ≤sk ≤...≤sn−1 ≤t i=k
n
xn (t) = Y1,n (t) = sup (Bi (si ) − Bi (si−1 )), (2.2.3)
0≤s1 ≤...≤sn−1 ≤t i=1
There are other interpretations for the quantity xn (t) than the system of reflected
Brownian motions focused on in this work, and each interpretation has inspired
different results over the last decades. One of these is seeing it as a sequence of
Brownian queues in series. A famous theorem of queueing theory, Burke’s theorem,
which states that the output of a stable, stationary M/M/1 queue is Poisson, can be
adapted to the Brownian setting (O’Connell and Yor 2001). This will be employed
in Sect. 2.3.4.
Furthermore, xn (t) can be viewed as a model of directed last-passage percolation
through a random medium, or equivalently a zero-temperature directed polymer in
a random environment. This model is constructed as follows:
Consider the space R+ × Z and assign white noise dBn as random background
weight on each line R+ × {n} for n ≥ 1. An up-right path is characterized by its
jumping points si and consists of line segments [sn−1 , sn ] × {n}, see Fig. 2.1. The set
of up-right paths going from (t1 , n 1 ) to (t2 , n 2 ) can then be parameterized by
Π (t1 , n 1 ; t2 , n 2 ) = s ∈ Rn 2 −n 1 +2 |t1 = sn 1 −1 ≤ sn 1 ≤ · · · ≤ sn 2 = t2 . (2.2.4)
The percolation time or weight of a path π ∈ Π is the integral over the background
weights along the path. Explicitly, we have:
n2
w(
π) = (Bi (si ) − Bi (si−1 )) . (2.2.5)
i=n 1
The last passage percolation time is given by the supremum over all such paths:
1
s1 s2 s3 s4 t
12 2 One-Sided Reflected Brownian …
The supremum is almost surely attained by a unique path π ∗ , called the maximizer. It
exists because the supremum can be rewritten as a composition of a finite maximum
and a supremum of a continuous function over a compact set. Uniqueness follows
from elementary properties of the Brownian measure. Most importantly, from the
definition, we have
xn (t) = L (0,1)→(t,n) . (2.2.7)
Although apparently more difficult to handle, attention has been turning to these
positive temperature models recently, because, among other things, they allow for a
scaling limit to the KPZ equation by tuning the parameter β in the right way.
The behaviour of xn (t) under the packed initial condition is quite well under-
stood by now. Notice that by Brownian scaling we have the distributional identity
d √
{xn (t), n ≥√1} = { t xn (1), n ≥ 1}. The first result has been a law of large numbers,
i.e. xn (1)/ n converges to a constant (Glynn and Whitt 1991), therein it was already
conjectured it equals 2, which has been proven subsequently (Seppäläinen 1997).
Much of the results that followed exploited connections to a random matrix model,
so let us introduce it in full generality right away:
Let bi,i (t) for 1 ≤ i ≤ N and bi, j (t), bi, j (t) for 1 ≤ i < j ≤ N , be independent
Brownian motions. Define a stochastic process H (t), t ≥ 0, on the space of N × N
Hermitian matrices by
2.2 Packed Initial Conditions 13
Denote by λn1 (t) ≤ λn2 (t) ≤ · · · ≤ λnn (t) the ordered eigenvalues of the n × n
principal minor of the matrix H (t). The process {λnk (t), 1 ≤ k ≤ n ≤ N } is called
the Dyson Brownian minor process. It is a classical result, that the eigenvalues of
consecutive minors are interlaced, λkn+1 (t) ≤ λnk (t) ≤ λk+1
n+1
(t), so this process lives
in the Gelfand–Tsetlin cone:
GT N = xkn ∈ R, 1 ≤ k ≤ n ≤ N , xkn+1 ≤ xkn ≤ xk+1
n+1
. (2.2.11)
Its fixed time distribution is the eigenvalue distribution of the Gaussian unitary
ensemble (GUE):
1 − yk2
N
P λnk (t) ∈ dyk , 1 ≤ k ≤ n = e 2t (yi − yk )2 d y. (2.2.13)
Z n (t) k=1 k<i
motions, which are reflected downwards resp. upwards from B11 (t). Iteratively con-
struct Bkn (t) as a Brownian motion reflected downwards from Bkn−1 (t) and upwards
n−1
from Bk−1 (t), with the peripheral processes Bkn (t) for k = 1 or k = n being
reflected from one process only. By construction, we have xn (t) = Bnn (t). The
process {Bkn (t), 1 ≤ k ≤ n ≤ N } is called Warren’s process, and has been intro-
duced and studied in Warren (2007). Restricted to one layer n, it is distributed as a
Dyson’s Brownian motion. Warren’s process shares the fixed time distribution with
the GUE minor process,
d
{λnk (1), 1 ≤ k ≤ n ≤ N } = {Bkn (1), 1 ≤ k ≤ n ≤ N }. (2.2.14)
There are also formulas for the transition density of the system along the edge, {Bnn (t),
1 ≤ n ≤ N } as well as for the system of two consecutive layers
{Bnk (t), 1 ≤ k ≤ n, N − 1 ≤ n ≤ N }.
The connection between Warren’s process and the Dyson Brownian minor process
does not, however, hold in full generality. The common dynamics of any amount
of consecutive layers of Warrens process, i.e. the process {Bnk (t), 1 ≤ k ≤ n,
N1 ≤ n ≤ N2 }, is simply given by Dyson’s SDE (2.2.12) for the layer n = N1
and the reflection SDE’s for the higher order layers. In the Dyson Brownian minor
process, on the other hand, the common dynamics of two consecutive layers is given
by a more complicated SDE (see (2.30) in Adler et al. (2014)) and the evolution of
three or more consecutive layers is not even a Markov process anymore. Interestingly,
both processes still show the same distribution along so-called space-like paths, i.e.
sequences of points (n i , ti ) satisfying ti ≤ ti+1 and n i ≥ n i+1 , in which case also
determinantal formulas exist (Adler et al. 2014; Ferrari and Frings 2010).
The determinantal formulas coming from the random matrix model are suitable
for asymptotic analysis to show multi-point scaling limits, where the Airy2 process
arises. It appears both for correlations of xn (t) along the n direction and the t direction,
as well as along general space-like paths. Reference (Johansson 2003) gives a sketch
of the proof for both directions, (Adler and Moerbeke 2005) prove the scaling limit
with correlations along t rigorously. A complete proof for correlations along n is
given in Sect. 5.1.
The main focus of this work is showing determinantal formulas and scaling limits
for other initial conditions. Unfortunately the connection to random matrices breaks
down in this case. In fact, neither the Airy1 process nor the Airystat process have
ever been found in a scaling limit of a random matrix model. This has been rather
surprising, as the one-point distribution of the Airy1 process, the GOE Tracy–Widom
distribution, does arise in such a model, namely as the limiting distribution of the
largest eigenvalue of a Gaussian real symmetric matrix, the GOE ensemble (Borne-
mann et al. 2008).
2.3 Infinite Particle Systems 15
2.3.1 Definition
At first sight it might seem trivial to extend the definition of the half-infinite system
of one-sided reflected Brownian motions to an infinite number of particles {xn (t),
n ∈ Z}, by letting the index k run over (−∞, n] in (2.1.4). However, it has to be
shown that this maximum is finite, which is only the case for initial conditions which
are not too √closely spaced together. Roughly said, the growth rate of −ζ−k has to be
faster than k for large k. We call those initial conditions admissible.
Knowing the law of large numbers under packed initial conditions, it√is reasonable
to expect such a behaviour. It implies that Yk,n (t) grows roughly as 2 (n − k)t, so
in order for the maximum
√ in (2.3.2) being attained by a finite k for we need −ζ−k to
grow faster than k.
is finite.
By Proposition 2.4 below, which is proven in Sect. 2.3.2, this maximum exists and
is finite. More specifically, we will show that for ζ being any admissible initial con-
dition, the infinite system {xn (t), n ∈ Z} is the limit of certain half-infinite systems
{xn(M) (t), n ≥ −M} as M → ∞, where
Thus as desired xn(M) (t) is a Brownian motion starting from ζn and reflected off by
(M)
xn−1 for n > −M.
16 2 One-Sided Reflected Brownian …
Proposition 2.4 For any t > 0, n ∈ Z there exists almost surely a k ≤ n maximizing
Yk,n (t) + ζk , i.e. the maximum in (2.3.2) exists. Furthermore, for any T > 0,
as well as
lim sup |xn(M) (t) − xn (t)| = 0, a.s.. (2.3.7)
M→∞ t∈[0,T ]
The convergence result allows for taking the limit in (2.3.4), implying that the system
{xn (t), n ∈ Z} satisfies the Skorokhod equation, too.
The initial conditions we are actually interested in are corresponding to the two
remaining fundamental geometries in the KPZ universality class. The first one is the
flat surface, which translates into periodic initial conditions x(0) = ζflat , defined by
ζ0stat = 0,
λ−1 Expn , for n > 0, (2.3.9)
ζnstat − ζn−1
stat
=
ρ−1 Expn , for n ≤ 0.
2.3.2 Well-Definedness
For the proof of Proposition 2.4 we first need the following concentration inequality:
2.3 Infinite Particle Systems 17
Proposition 2.5 For each T > 0 there exists a constant C > 0 such that for all
k < m, δ > 0,
Yk,m (T )
≥ 2 + δ ≤ const · e−(m−k+1) δ .
2/3
P √ (2.3.10)
(m − k + 1)T
This proposition is proven in Sect. 5.1.2. Another necessary lemma, that will be
proven an intuitive way in Sect. 2.3.3, is:
Lemma 2.6 Consider 0 ≤ t1 ≤ t2 and m, Mt1 , Mt2 such that
(Mti ) (Mti )
xm (ti ) = xm (ti ) = xm (ti ), for i = 1, 2. (2.3.11)
Then
(Mt2 ) (Mt2 )
xm (t1 ) = xm (t1 ) = xm (t1 ). (2.3.12)
and
(M)
xn(M) (t) = max ζn + Bn (t), sup (xn−1 (s) + Bn (t) − Bn (s)) (2.3.14)
0≤s≤t
for n > −M. This system differs from xn(M) (t) just in the drift of the leftmost particle,
which of course influences all other particles as well (the choice of the extra drift
is because the system with infinite many particles in R− generates a drift ρ). This
system of particles satisfies
xn(M) (t) = max Y−M,n (t) + ζ−M , max {Yk,n (t) + ζk } , (2.3.15)
k∈[−M+1,n]
with
n
Yk,n (t) = sup ρsk+1 + (Bi (si+1 ) − Bi (si )) . (2.3.16)
0≤sk+1 ≤...≤sm ≤t i=k
It is now straightforward to show ∞ M=0 P(A M ) < ∞. In fact, summability of the
probabilities of the first set in (2.3.18) is a consequence of Proposition 2.5, applied
with δ = 1, while the second set is covered by the definition of an admissible initial
condition. For the third set, notice that the left hand side is a Gaussian distribution
independent of M, while the right hand side is dominated by the M χ term for large
M. Finiteness of the sum allows applying Borel–Cantelli, i.e. A M occurs only finitely
many times almost surely. This means, that a.s. there exists a MT , such that for all
M ≥ MT the following three inequalities hold:
Y−M,n (T ) < 3 (M + n + 1)T
M χ < ζn − ζ−M (2.3.19)
− Yn,n (T ) < −ρT − 3 (M + n + 1)T + M χ
resulting in
xn(MT ) (T ) = xn(MT ) (T ). (2.3.22)
Repeating the same argument, we see that for every t ∈ [0, T ] there exists Mt
such that xn (t) = xn(Mt ) (t) = xn(Mt ) (t). Applying Lemma 2.6 then gives xn (t) =
xn(MT ) (t) = xn(MT ) (t) for every t ∈ [0, T ]. This settles the convergence and the
existence of a finite maximizing k in (2.3.2).
To see (2.3.6), which is equivalent to supt∈[0,T ] |xm(MT ) (t)| < ∞, we apply the
bound
n
|Yk,n (t)| ≤ sup Bi (s) − inf Bi (s) < ∞. (2.3.23)
0≤s≤t 0≤s≤t
i=k
n2
w(
π) = Bi (si ) − Bi (si−1 ) + (ζi − ζi−1 )1si−1 =0 , (2.3.24)
i=n 1
and the percolation time L (0,n 1 )→(t,n 2 ) again as the supremum over the weight of all
paths. As t → 0 it is clear that any contribution from the Brownian background
weight will converge to 0, so the path tries to accumulate as much of the Dirac
weights as possible, i.e. we have the initial condition
n2
lim L (0,n 1 )→(t,n 2 ) = (ζi − ζi−1 ) = ζn 2 − ζn 1 −1 . (2.3.25)
t→0
i=n 1
L (0,n 1 )→(t,n 2 ) = L (0,n 1 )→(t,n 2 ) + ζn 1 −1 , (2.3.26)
Proof (Proof of Lemma 2.6) For t1 = t2 there is nothing to prove, so let t1 < t2 . For
(Mt )
each i, the equation xm (ti ) = xm i (ti ) implies that the maximizing paths of the LHS
and the RHS are equal on the restriction to (si , i ≥ Mi ), i.e. they have the same exit
point ei that satisfies ei ≥ Mti . Now if e1 ≥ e2 , then also e1 ≥ Mt2 , which means
that the path maximizing xm (t1 ) is contained in the set Π (0, −Mt2 ; t1 , m), resulting
(Mt )
in xm (t1 ) = xm 2 (t1 ).
If, however, e1 < e2 , then the maximizing path segments (0, e1 ) → (t1 , m)
and (0, e2 ) → (t2 , m) would need to have an intersection point (t ∗ , m ∗ ). We can
then construct a new maximizing path for xm (t1 ) by stringing together the seg-
ments (0, e1 ) → (0, e2 ) → (t ∗ , m ∗ ) → (t1 , m), where the middle segment is part
of the xm (t2 )-maximizing path and the last segment is part of the original xm (t1 )-
maximizing path. This contradicts the uniqueness of the maximizing path.
(Mt )
The equality xm (t1 ) = xm 2 (t1 ) is shown in the same way.
20 2 One-Sided Reflected Brownian …
2.3.4 Stationarity
We establish a useful property which will allow us to study our system of interacting
Brownian motions through a system with a left-most Brownian particle.
Proposition 2.7 Under the initial condition x(0) = ζ = ζstat (λ, ρ) defined in
(2.3.9), for each n ≤ 0 the process
xn (t) − ζn − ρt (2.3.29)
Remark 1 Proposition 2.7 allows us to restrict our attention to the half-infinite sys-
tem. In fact, conditioned on the path of x0 , the systems of particles {xn (t), n < 0}
and {xn (t), n > 0} are independent, as it is clear by the definition of the system. Then
(2.3.29) implies that the law of {xn (t), n > 0} is the same as the one obtained replac-
ing the infinitely many particles {xm (t), m ≤ 0} with a single Brownian motion
x0 (t) which has a drift ρ. This property will be used to derive our starting result,
Proposition 6.2.
Remark 2 From a stochastic analysis point of view, we find that the system
{xn (t), n ≥ 0} satisfies
(M)
is a Brownian motion. Now assume xn−1 (t) − ζn−1 − ρt is a Brownian motion. By
definition,
Then
D(t) − ζ − ρt (2.3.35)
and
d(t) = B(t) + q(0) − q(t), (2.3.37)
we can apply Theorem 2 (O’Connell and Yor 2001), i.e., d(t) is a Brownian motion.
Now,
d √ d ρ
q(0) = sup{−B(s) − C(s) + ρs} = sup{ 2B(s) − ρs} = sup B(s) − s , (2.3.38)
s≤0 s≥0 s≥0 2
Proof This formula can be found in Borodin and Salminen (2012), Part II, Sect. 2,
Eq. (1.1.4). We provide here two different proofs of the claim.
(a) The random variable
sup (B(s) − ρs) (2.3.41)
0≤s≤t
d
B̃(t) = sup (B(t) − B(s) − ρ(t − s))
0≤s≤t
(2.3.42)
d
= sup (B(t) − B(t − s) − ρs) = sup (B(s) − ρs),
0≤s≤t 0≤s≤t
where we changed s into t − s and used the fact that B(t) − B(t − s) has the same
distribution of B(s). As t → ∞, this converges to the stationary distribution of this
process, which is the exponential distribution with parameter 2ρ.
(b) For x ≥ 0, define the stopping time τx = inf{s ≥ 0 | B(s) − sρ/2 ≥ x}. Then,
1
The process s → Ms = eρB(s)− 2 ρ s is a martingale (the geometric Brownian motion),
2
References
A.V. Skorokhod, Stochastic equations for diffusions in a bounded region. Theory Probab. Appl. 6,
264–274 (1961)
R.F. Anderson, S. Orey, Small random perturbation of dynamical systems with reflecting boundary.
Nagoya Math. J. 60, 189–216 (1976)
N. O’Connell, M. Yor, Brownian analogues of Burke’s theorem. Stoch. Proc. Appl. 96, 285–304
(2001)
K. Johansson, Shape fluctuations and random matrices. Comm. Math. Phys. 209, 437–476 (2000)
K. Johansson, Discrete polynuclear growth and determinantal processes. Comm. Math. Phys. 242,
277–329 (2003)
T. Seppäläinen, B. Valkó, Bounds for scaling exponents for a 1+1 dimensional directed polymer in
a brownian environment. ALEA 7, 451–476 (2010)
References 23
A. Borodin, I. Corwin, P.L. Ferrari, B. Vető, Height fluctuations for the stationary KPZ equation.
Math. Phys. Anal. Geom. 18, 1–95 (2015)
P.W. Glynn, W. Whitt, Departures from many queues in series. Ann. Appl. Probab. 1, 546–572
(1991)
T. Seppäläinen, A scaling limit for queues in series. Ann. Appl. Probab. 7, 855–872 (1997)
F.J. Dyson, A Brownian-motion model for the eigenvalues of a random matrix. J. Math. Phys. 3,
1191–1198 (1962)
C.A. Tracy, H. Widom, Level-spacing distributions and the Airy kernel. Comm. Math. Phys. 159,
151–174 (1994)
J. Gravner, C.A. Tracy, H. Widom, Limit theorems for height fluctuations in a class of discrete space
and time growth models. J. Stat. Phys. 102, 1085–1132 (2001)
Y. Baryshnikov, GUEs and queues. Probab. Theory Relat. Fields 119, 256–274 (2001)
K. Johansson, E. Nordenstam, Eigenvalues of GUE minors. Electron. J. Probab. 11, 1342–1371
(2006)
J. Warren, Dyson’s Brownian motions, intertwining and interlacing. Electron. J. Probab. 12, 573–
590 (2007)
M. Adler, E. Nordenstam, P. van Moerbeke, Consecutive minors for Dyson’s Brownian motions.
Stoch. Proc. Appl. 124, 2023–2051 (2014)
M. Adler, E. Nordenstam, P. van Moerbeke, The Dyson Brownian minor process. Annales de
l’Institut Fourier 64, 971–1009 (2014)
P.L. Ferrari, R. Frings, On the partial connection between random matrices and interacting particle
systems. J. Stat. Phys. 141, 613–637 (2010)
M. Adler, P. van Moerbeke, PDE’s for the joint distribution of the Dyson, Airy and Sine processes.
Ann. Probab. 33, 1326–1361 (2005)
F. Bornemann, P.L. Ferrari, M. Prähofer, The Airy1 process is not the limit of the largest eigenvalue
in GOE matrix diffusion. J. Stat. Phys. 133, 405–415 (2008)
J. Baik, P.L. Ferrari, S. Péché, Limit process of stationary TASEP near the characteristic line. Comm.
Pure Appl. Math. 63, 1017–1070 (2010)
P.L. Ferrari, H. Spohn, Scaling limit for the space-time covariance of the stationary totally asym-
metric simple exclusion process. Comm. Math. Phys. 265, 1–44 (2006)
T. Imamura, T. Sasamoto, Fluctuations of the one-dimensional polynuclear growth model with
external sources. Nucl. Phys. B 699, 503–544 (2004)
A.N. Borodin, P. Salminen, Handbook of Brownian Motion - Facts and Formulae (Birkhäuser,
Switzerland) (2012)
Another random document with
no related content on Scribd:
identify now Hamilton of Bangour, young David Malloch, a William
Crawford, and a William Walkinshaw,—contained about sixty songs
of Ramsay’s own composition. Similarly, among several mock-
antiques by modern hands inserted into The Evergreen, were two
by Ramsay himself, entitled The Vision and The Eagle and Robin
Redbreast.
The time had come for Ramsay’s finest and most characteristic
performance. More than once, in his miscellanies hitherto, he had
tried the pastoral form in Scotch, whether from a natural tendency to
that form or induced by recent attempts in the English pastoral by
Ambrose Philips, Pope, and Gay. Besides his pastoral elegy on the
death of Addison, and another on the death of Prior, he had written a
pastoral dialogue of real Scottish life in 162 lines, entitled Patie and
Roger, introduced by this description:—
“Beneath the south side of a craigy bield,
Where a clear spring did halesome water yield,
Twa youthfu’ shepherds on the gowans lay,
Tending their flocks ae bonny morn of May:
Poor Roger graned till hollow echoes rang,
While merry Patie hummed himsel a sang.”
This piece, and two smaller pastoral pieces in the same vein, called
Patie and Peggy and Jenny and Meggie, had been so much liked
that Ramsay had been urged by his friends to do something more
extensive in the shape of a pastoral story or drama. He had been
meditating such a thing through the year 1724, while busy with his
two editorial compilations; and in June 1725 the result was given to
the public in The Gentle Shepherd: A Scots Pastoral Comedy.
Here the three pastoral sketches already written were inwoven into a
simply-constructed drama of rustic Scottish life as it might be
imagined among the Pentland Hills, near Edinburgh, at that time,
still within the recollection of very old people then alive, when the
Protectorates of Cromwell and his son had come to an end and Monk
had restored King Charles. The poem was received with enthusiastic
admiration. There had been nothing like it before in Scottish
literature, or in any other; nothing so good of any kind that could be
voted as even similar; and this was at once the critical verdict. It is a
long while ago, and there are many spots in Edinburgh which
compete with one another in the interest of their literary
associations; but one can stand now with particular pleasure for a
few minutes any afternoon opposite that decayed house in the High
Street, visible as one is crossing from the South Bridge to the North
Bridge, where Allan Ramsay once had his shop, and whence the first
copies of The Gentle Shepherd were handed out, some day in June
1725, to eager Edinburgh purchasers.
The tenancy of this house by Ramsay lasted but a year longer.
He had resolved to add to his general business of bookselling and
publishing that of a circulating library, the first institution of the
kind in Edinburgh. For this purpose he had taken new premises, still
in the High Street, but in a position even more central and
conspicuous than that of “The Mercury” opposite Niddry’s Wynd.
They were, in fact, in the easternmost house of the Luckenbooths, or
lower part of that obstructive stack of buildings, already mentioned,
which once ran up the High Street alongside of St. Giles’s Church,
dividing the traffic into two narrow and overcrowded channels. It is
many years since the Luckenbooths and the whole obstruction of
which they formed a part were swept away; but from old prints we
can see that the last house of the Luckenbooths to the east was a tall
tenement of five storeys, with its main face looking straight down the
lower slope of the High Street towards the Canongate. The strange
thing was that, though thus in the very heart of the bustle of the town
as congregated round the Cross, the house commanded from its
higher windows a view beyond the town altogether, away to Aberlady
Bay and the farther reaches of sea and land in that direction. It was
into this house that Ramsay removed in 1726, when he was exactly
forty years of age. The part occupied by him was the flat immediately
above the basement floor, but perhaps with that floor in addition.
The sign he adopted for the new premises was one exhibiting the
heads or effigies of Ben Jonson and Drummond of Hawthornden.
Having introduced Ramsay into this, the last of his Edinburgh
shops, we have reached the point where our present interest in him
all but ends. In 1728, when he had been two years in the new
premises, he published a second volume of his collected poems,
under the title of Poems by Allan Ramsay, Volume II., in a
handsome quarto matching the previous volume of 1721, and
containing all the pieces he had written since the appearance of that
volume; and in 1730 he published A Collection of Thirty Fables.
These were his last substantive publications, and with them his
literary career may be said to have come to a close. Begun in the last
years of the reign of Queen Anne, and continued through the whole
of the reign of George I., it had just touched the beginning of that of
George II., when it suddenly ceased. Twice or thrice afterwards at
long intervals he did scribble a copy of verses; but, in the main, from
his forty-fifth year onwards, he rested on his laurels. Thenceforward
he contented himself with his bookselling, the management of his
circulating library, and the superintendence of the numerous
editions of his Collected Poems, his Gentle Shepherd, and his Tea
Table Miscellany that were required by the public demand, and the
proceeds of which formed a good part of his income. It would be a
great mistake, however, to suppose that, when Allan Ramsay’s time
of literary production ended, the story of his life in Edinburgh also
came to a close, or ceased to be important. For eight-and-twenty
years longer, or almost till George II. gave place to George III.,
Ramsay continued to be a living celebrity in the Scottish capital,
known by figure and physiognomy to all his fellow-citizens, and
Ramsay’s bookshop at the end of the Luckenbooths, just above the
Cross, continued to be one of the chief resorts of the well-to-do
residents, and of chance visitors of distinction. Now and then,
indeed, through the twenty-eight years, there are glimpses of him
still in special connections with the literary, as well as with the social,
history of Edinburgh. When the English poet Gay, a summer or two
before his death in 1732, came to Edinburgh on a visit, in the
company of his noble patrons, the Duke and Duchess of
Queensberry, and resided with them in their mansion of
Queensberry House in the Canongate,—now the gloomiest and
ugliest-looking house in that quarter of the old town, but then
reckoned of palatial grandeur,—whither did he tend daily, in his
saunterings up the Canongate, but to Allan Ramsay’s shop? One
hears of him as standing there with Allan at the window to have the
city notabilities and oddities pointed out to him in the piazza below,
or as taking lessons from Allan in the Scottish words and idioms of
the Gentle Shepherd, that he might explain them better to Mr. Pope
when he went back to London.
Some years later, when Ramsay had reached the age of fifty, and
he and his wife were enjoying the comforts of his ample success, and
rejoicing in the hopes and prospects of their children,—three
daughters, “no ae wally-draggle among them, all fine girls,” as
Ramsay informs us, and one son, a young man of three-and-twenty,
completing his education in Italy for the profession of a painter,—
there came upon the family what threatened to be a ruinous disaster.
Never formally an anti-Presbyterian, and indeed regularly to be seen
on Sundays in his pew in St. Giles’s High Kirk, but always and
systematically opposed to the unnecessary social rigours of the old
Presbyterian system, and of late under a good deal of censure from
clerical and other strict critics on account of the dangerous nature of
much of the literature put in circulation from his library, Ramsay
had ventured at last on a new commercial enterprise, which could
not but be offensive on similar grounds to many worthy people,
though it seems to have been acceptable enough to the Edinburgh
community generally. Edinburgh having been hitherto deficient in
theatrical accommodation, and but fitfully supplied with dramatic
entertainments, he had, in 1736, started a new theatre in Carrubber’s
Close, near to his former High Street shop. He was looking for great
profits from the proprietorship of this theatre and his partnership in
its management. Hardly had he begun operations, however, when
there came the extraordinary statute of 10 George II. (1737),
regulating theatres for the future all over Great Britain. As by this
statute there could be no performance of stage-plays out of London
and Westminster, save when the King chanced to be residing in some
other town, Ramsay’s speculation collapsed, and all the money he
had invested in it was lost. It was a heavy blow; and he was moved by
it to some verses of complaint to his friend Lord President Forbes
and the other judges of the Court of Session. While telling the story
of his own hardship in the case, he suggests that an indignity had
been done by the new Act to the capital of Scotland:—
“Shall London have its houses twa
And we doomed to have nane ava’?
Is our metropolis, ance the place
Where langsyne dwelt the royal race
Of Fergus, this gate dwindled down
To a level with ilk clachan town,
While thus she suffers the subversion
Of her maist rational diversion?”
However severe the loss to Ramsay at the time, it was soon tided
over. Within six years he is found again quite at ease in his worldly
fortunes. His son, for some years back from Italy, was in rapidly
rising repute as a portrait-painter, alternating between London and
Edinburgh in the practice of his profession, and a man of mark in
Edinburgh society on his own account; and, whether by a junction of
the son’s means with the father’s, or by the father’s means alone, it
was now that there reared itself in Edinburgh the edifice which at the
present day most distinctly preserves for the inhabitants the memory
of the Ramsay family in their Edinburgh connections. The
probability is that, since Allan had entered on his business premises
at the end of the Luckenbooths, his dwelling-house had been
somewhere else in the town or suburbs; but in 1743 he built himself a
new dwelling-house on the very choicest site that the venerable old
town afforded. It was that quaint octagon-shaped villa, with an
attached slope of green and pleasure-ground, on the north side of the
Castle Hill, which, as well from its form as from its situation, attracts
the eye as one walks along Princes Street, and which still retains the
name of Ramsay Lodge. The wags of the day, making fun of its
quaint shape, likened the construction to a goose-pie; and something
of that fancied resemblance may be traced even now in its extended
and improved proportions. But envy may have had a good deal to do
with the comparison. It is still a neat and comfortable dwelling
internally, while it commands from its elevation an extent of scenery
unsurpassed anywhere in Europe. The view from it ranges from the
sea-mouth of the Firth of Forth on the east to the first glimpses of the
Stirlingshire Highlands on the west, and again due north across the
levels of the New Town, and the flashing waters of the Firth below
them, to the bounding outline of the Fifeshire hills. When, in 1743,
before there was as yet any New Town at all, Allan Ramsay took up
his abode in this villa, he must have been considered a fortunate and
happy man. His entry into it was saddened, indeed, by the death of
his wife, which occurred just about that time; but for fourteen years
of widowerhood, with two of his daughters for his companions, he
lived in it serenely and hospitably. During the first nine years of
those fourteen he still went daily to his shop in the Luckenbooths,
attending to his various occupations, and especially to his circulating
library, which is said to have contained by this time about 30,000
volumes; but for the last five or six years he had entirely relinquished
business. There are authentic accounts of his habits and demeanour
in his last days, and they concur in representing him as one of the
most charming old gentlemen possible, vivacious and sprightly in
conversation, full of benevolence and good humour, and especially
fond of children and kindly in his ways for their amusement. He died
on the 7th of January 1758, in the seventy-second year of his age, and
was buried in Greyfriars Churchyard.
Ramsay had outlived nearly all the literary celebrities who had
been his contemporaries during his own career of active authorship,
ended nearly thirty years before. Swift and Pope were gone, after
Gay, Steele, Arbuthnot, and others of the London band, who had
died earlier. Of several Scotsmen, his juniors, who had stepped into
the career of literature after he had shown the way, and had attained
to more or less of poetic eminence under his own observation, three,
—Robert Blair, James Thomson, and Hamilton of Bangour,—had
predeceased him. Their finished lives, with all the great radiance of
Thomson’s, are wholly included in the life of Allan Ramsay. David
Malloch, who had been an Edinburgh protégé of Ramsay’s, but had
gone to London and Anglicised himself into “Mallet,” was about the
oldest of his literary survivors into another generation; but in that
generation, as Scotsmen of various ages, from sixty downwards to
one-and-twenty, living, within Scotland or out of it, at the date of
Ramsay’s death, we count Lord Kames, Armstrong, Reid, Hume,
Lord Monboddo, Hugh Blair, George Campbell, Smollett, Wilkie,
Blacklock, Robertson, John Home, Adam Smith, Adam Ferguson,
Lord Hailes, Falconer, Meikle, and Beattie. Such of these as were
residents in Edinburgh had known Allan Ramsay personally; others
of them had felt his influence indirectly; and all must have noted his
death as an event of some consequence.
The time is long past for any exaggeration of Allan Ramsay’s
merits. But, call him only a slipshod little Horace of Auld Reekie,
who wrote odes, epistles, satires, and other miscellanies in Scotch
through twenty years of the earlier part of the eighteenth century,
and was also, by a happy chance, the author of a unique and
delightful Scottish pastoral, it remains true that he was the most
considerable personality in Scottish literary history in order of time
after Drummond of Hawthornden, or, if we think only of the
vernacular, after Sir David Lindsay, and that he did more than any
other man to stir afresh a popular enthusiasm for literature in
Scotland after the Union with England. All in all, therefore, it is with
no small interest that, in one’s walks along the most classic
thoroughfare of the present Edinburgh, one gazes at the white stone
statue of Allan Ramsay, from the chisel of Sir John Steell, which
stands in the Gardens just below the famous “goose-pie villa.” It
looks as if the poet had just stepped down thence in his evening
habiliments to see things thereabouts in their strangely changed
condition. By the tact of the sculptor, he wears, one observes, not a
wig, but the true poetic night-cap or turban.
LADY WARDLAW AND THE BARONESS NAIRNE[6]
Old as he is, he will set out at once, taking his three eldest sons with
him, Robin, Thomas, and Malcolm, and telling his lady in his
farewell to her:—
“My youngest son sall here remain
To guaird these stately towers,
And shoot the silver bolt that keeps
Sae fast your painted bowers.”
And so we take leave of the high castle on the hill, with the lady,
her youngest son, and Fairly fair, in it, and follow the old lord and his
other three sons over the moors and through the glens as they ride to
the rendezvous. On their way they encounter a wounded knight,
lying on the ground and making a heavy moan:—
“‘Here maun I lie, here maun I die,
By treachery’s false guiles;
Witless I was that e’er gave faith
To wicked woman’s smiles.’”
Here the story might seem to end, and here perhaps it was intended
at first that it should end; but in the completer copies there are three
more stanzas, taking us back to Hardyknute’s castle on the high hill.
We are to fancy Hardyknute and his sons returning joyfully thither
after the great victory:—
“Loud and chill blew the westlin wind,
Sair beat the heavy shower;
Mirk grew the nicht ere Hardyknute
Wan near his stately tower:
His tower, that used with torches’ bleeze
To shine sae far at nicht,
Seemed now as black as mourning weed:
Nae marvel sair he sich’d.