Full download test bank at ebook textbookfull.

com

Macroscopic Limits of Quantum

Systems Munich Germany March 30

CLICK LINK TO DOWLOAD

https://textbookfull.com/product/macroscopic-
limits-of-quantum-systems-munich-germany-
march-30-april-1-2017-daniela-cadamuro/

textbookfull
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

National Geographic History March April 2017 2nd

Edition National Geographic

https://textbookfull.com/product/national-geographic-history-
march-april-2017-2nd-edition-national-geographic/

Cook s Illustrated 145 2017 march april 1st Edition

Cook'S Illustrated

https://textbookfull.com/product/cook-s-
illustrated-145-2017-march-april-1st-edition-cooks-illustrated/

Nonlinear problems in accelerator physics proceedings

of the INT workshop on Nonlinear Problems in
Accelerator Physics held in Berlin Germany 30 March 2
April 1992 Berz
https://textbookfull.com/product/nonlinear-problems-in-
accelerator-physics-proceedings-of-the-int-workshop-on-nonlinear-
problems-in-accelerator-physics-held-in-berlin-
germany-30-march-2-april-1992-berz/

Applications of Evolutionary Computation 19th European

Conference EvoApplications 2016 Porto Portugal March 30
April 1 2016 Proceedings Part I 1st Edition Giovanni
Squillero
https://textbookfull.com/product/applications-of-evolutionary-
computation-19th-european-conference-evoapplications-2016-porto-
portugal-march-30-april-1-2016-proceedings-part-i-1st-edition-
Database Systems for Advanced Applications 22nd
International Conference DASFAA 2017 Suzhou China March
27 30 2017 Proceedings Part I 1st Edition Selçuk Candan

https://textbookfull.com/product/database-systems-for-advanced-
applications-22nd-international-conference-dasfaa-2017-suzhou-
china-march-27-30-2017-proceedings-part-i-1st-edition-selcuk-
candan/

Mobility Analytics for Spatio Temporal and Social Data

First International Workshop MATES 2017 Munich Germany
September 1 2017 Revised Selected Papers 1st Edition
Christos Doulkeridis
https://textbookfull.com/product/mobility-analytics-for-spatio-
temporal-and-social-data-first-international-workshop-
mates-2017-munich-germany-september-1-2017-revised-selected-
papers-1st-edition-christos-doulkeridis/

Limits and Languages in Contemporary Irish Women's

Poetry Daniela Theinová

https://textbookfull.com/product/limits-and-languages-in-
contemporary-irish-womens-poetry-daniela-theinova/

Calling on all Foodies 30 1 New and Novel Ways to Enjoy

Your Chicken Blomgren April

https://textbookfull.com/product/calling-on-all-foodies-30-1-new-
and-novel-ways-to-enjoy-your-chicken-blomgren-april/

Macroscopic and Large Scale Phenomena Coarse Graining

Mean Field Limits and Ergodicity 1st Edition Adrian
Muntean

https://textbookfull.com/product/macroscopic-and-large-scale-
phenomena-coarse-graining-mean-field-limits-and-ergodicity-1st-
edition-adrian-muntean/
Springer Proceedings in Mathematics & Statistics

Daniela Cadamuro
Maximilian Duell
Wojciech Dybalski
Sergio Simonella Editors

Macroscopic
Limits of
Quantum
Systems
Munich, Germany, March 30 - April 1,
2017
Springer Proceedings in Mathematics & Statistics

Volume 270
Springer Proceedings in Mathematics & Statistics
This book series features volumes composed of selected contributions from
workshops and conferences in all areas of current research in mathematics and
statistics, including operation research and optimization. In addition to an overall
evaluation of the interest, scientific quality, and timeliness of each proposal at the
hands of the publisher, individual contributions are all refereed to the high quality
standards of leading journals in the field. Thus, this series provides the research
community with well-edited, authoritative reports on developments in the most
exciting areas of mathematical and statistical research today.

More information about this series at http://www.springer.com/series/10533

Daniela Cadamuro Maximilian Duell
•

Wojciech Dybalski Sergio Simonella

•

Editors

Macroscopic Limits
of Quantum Systems
Munich, Germany, March 30 - April 1, 2017

123
Editors
Daniela Cadamuro Wojciech Dybalski
Institut für Theoretische Zentrum Mathematik, M5
Physik Universität Leipzig Technische Universität München
Leipzig, Germany Garching, Germany

Maximilian Duell Sergio Simonella

Zentrum Mathematik, M5 UMPA UMR 5669 CNRS
Technische Universität München École normale supérieure
Garching, Germany Lyon, France

ISSN 2194-1009 ISSN 2194-1017 (electronic)

Springer Proceedings in Mathematics & Statistics
ISBN 978-3-030-01601-2 ISBN 978-3-030-01602-9 (eBook)
https://doi.org/10.1007/978-3-030-01602-9
Library of Congress Control Number: 2018957047

Mathematics Subject Classification (2010): 82C10, 35Q82, 81Qxx, 32A70, 11F72, 49-XX

© Springer Nature Switzerland AG 2018

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

“It is amazing how different the world appears on different scales.” Moving from
the subatomic up to the intergalactic world, we cross several physical levels
apparently ruled by very diverse laws. The connection between steps of this hier-
archy is a never-ending source of discoveries and mathematical challenges.
Our incipit is borrowed from the book by Herbert Spohn on the large-scale
dynamics of interacting particles, which is an outstanding reference in the field for
almost 30 years. Nowadays, research activity is intensively focused on quantum
systems. The present volume collects contributions from a number of experts and
aims to be a survey of current directions of investigation. Most of the authors
presented their research during the 3-day workshop “Macroscopic Limits of
Quantum Systems”, held in spring 2017 at the Technical University of Munich,
which was attended by nearly 60 people from several institutions and countries. The
conference was also an occasion to celebrate the achievements of Herbert and his
reception of the Max Planck Medal.
The main subject of the workshop was the mathematical investigation of col-
lective phenomena emerging from quantum theory at observable scales. On the side
of bosonic systems, the problem of Bose–Einstein condensation, ground-state
properties of Bogoliubov Hamiltonians and the Gross–Pitaevskii limit were cov-
ered. Concerning fermionic systems, the rigorous validation of Hartree–Fock
equations and the motion of tracer particles were discussed. Moreover, supercon-
ductivity and mathematical aspects of the BCS theory were presented, as well as the
statistical mechanics of the uniform electron gas and its relation to Jellium. Further
important topics were in the scope of the workshop: the justification and properties
of quantum Boltzmann-type equations, the theory of indirect measurements, the
strong-coupling limit for the polaron model, Peierls substitution in perturbed
periodic systems, and the adiabatic theorem for many-body systems.
This proceedings volume is a selection of works from leading scientists
including a large part of (regrettably, not all) the speakers of the workshop, and
proposing an overview of fields in which the current research is most active. The
book will serve both advanced graduate students as an introduction to research, and
specialists in mathematical physics, as it presents technical details in an accessible

v
vi Preface

way. As is well known, to prove rigorous results in quantum physics requires the
development of powerful tools within abstract mathematical theories. Here, we
encounter fruitful connections to functional analysis, spectral theory, statistical
mechanics, renormalization group techniques, and variational calculus. This book is
also an account of that remarkable effort.

Leipzig, Germany Daniela Cadamuro

Garching, Germany Maximilian Duell
Garching, Germany Wojciech Dybalski
Lyon, France Sergio Simonella
July 2018
Contents

Kinetic Theory and Thermalization of Weakly Interacting

Fermions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Jani Lukkarinen
The BCS Critical Temperature in a Weak External Electric
Field via a Linear Two-Body Operator . . . . . . . . . . . . . . . . . . . . . . . . . 29
Rupert L. Frank and Christian Hainzl
Effective Dynamics of Two Tracer Particles Coupled to a Fermi
Gas in the High-Density Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Maximilian Jeblick, David Mitrouskas and Peter Pickl
Mean-Field Evolution of Fermions with Singular Interaction . . . . . . . . . 81
Chiara Saffirio
Recent Advances in the Theory of Bogoliubov Hamiltonians . . . . . . . . . 101
Marcin Napiórkowski
Multi-scale Analysis in the Occupation Numbers of Particle States:
An Application to Three-Modes Bogoliubov Hamiltonians . . . . . . . . . . . 123
Alessandro Pizzo
Blow-Up Profile of Rotating 2D Focusing Bose Gases . . . . . . . . . . . . . . 145
Mathieu Lewin, Phan Thành Nam and Nicolas Rougerie
Gross–Pitaevskii Evolution for Bose–Einstein Condensates . . . . . . . . . . 171
Benjamin Schlein
Mean-Field Limits of Particles in Interaction with Quantized
Radiation Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Nikolai Leopold and Peter Pickl

vii
viii Contents

Taking Inspiration from Quantum-Wave Analogies—Recent

Results for Photonic Crystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
Max Lein
The Localization Dichotomy for Gapped Periodic Systems and Its
Relevance for Macroscopic Transport . . . . . . . . . . . . . . . . . . . . . . . . . . 243
Gianluca Panati
Kinetic Theory and Thermalization
of Weakly Interacting Fermions

Jani Lukkarinen

Abstract Weakly interacting quantum fluids allow for a natural kinetic theory
description which takes into account the fermionic or bosonic nature of the interact-
ing particles. In the simplest cases, one arrives at the Boltzmann–Nordheim equations
for the reduced density matrix of the fluid. We discuss here two related topics: the
kinetic theory of the fermionic Hubbard model, in which conservation of total spin
results in an additional Vlasov-type term in the Boltzmann equation, and the relation
between kinetic theory and thermalization.

1 Introduction

Kinetic theory describes motion which is transport dominated in the sense that typ-
ically the solutions to the kinetic equations correspond to constant velocity, i.e. bal-
listic, motion intercepted by collisions whose frequency is order one on the kinetic
space–time scales. Weakly interacting quantum fluids provide one such example
system, as discussed in detail in [1].
We focus here on one particular case of a weakly interacting quantum fluid, the
case of weakly interacting fermions hopping on a lattice. Such a model would arise
physically as a description of a fluid of electrons in a crystal background potential.
For our purposes, this model has also other attractive properties, namely, it has
interesting non-trivial kinetic theory with relatively few technical and mathematical
difficulties. Much of the discussion below can be straightforwardly adapted to bosonic
lattice systems, at least for initial data which exclude formation of Bose–Einstein
condensate. For more details about such extensions, we refer to [1]; for instance,
Remark 2.3 summarises the changes and new properties which arise for fermions
and bosons moving not on a lattice, but in the continuum R3 .

J. Lukkarinen (B)
Department of Mathematics and Statistics, University of Helsinki, P.O. Box 68,
00014 Helsingin yliopisto, Finland
e-mail: [email protected]

© Springer Nature Switzerland AG 2018 1

D. Cadamuro et al. (eds.), Macroscopic Limits of Quantum Systems,
Springer Proceedings in Mathematics & Statistics 270,
https://doi.org/10.1007/978-3-030-01602-9_1
2 J. Lukkarinen

The purpose of this contribution is not to provide a comprehensive review of

literature on kinetic theory and properties of fermionic systems. Instead, we focus
on building a bridge between mathematically rigorous results and the physics of
fermionic systems. To this end, we begin with a fairly detailed Sect. 2 on the defini-
tion of finite system of fermions hopping on a periodic lattice with pair interactions
between the particles, from the point of view of both a fixed particle number Hilbert
space and the full antisymmetric Fock space. In Sect. 3 we recall how the prob-
abilistic concepts of classical particle systems can be generalised into systems of
fermions, namely, the definition of quasifree states, reduced density matrices and
truncated correlation functions. The weak coupling limit of the first reduced density
matrix of a translation invariant system and its approximation by the appropriate
spatially homogeneous Boltzmann equation is reviewed in Sect. 4. As a conclusion,
we discuss in Sects. 5 and 6 the implications of the properties of the solutions to
the Boltzmann equation on thermalization in the original fermionic lattice system.
Acknowledgements and references can be found at the end of the text.

2 Dynamics of Lattice Fermions

We recall in this section the mathematical description of fermions, possibly with spin,
which are hopping on a finite periodic lattice of length L  1. The particles move
on a lattice whose points are labelled by Λ := Zd /(LZd ) which we parametrize by
a square centred at the origin. For instance, if L is even, we use the parametrisation
 
L L L d
Λ = − + 1, . . . , − 1, .
2 2 2

In particular, all arithmetic on Λ is performed ‘modulo L’: if x, y are in the above

parametrisation of Λ ⊂ Zd , then x + y ∈ Zd needs to be identified with its counter-
part in the parametrisation. Explicitly, ‘x + y’ is equal to x + y − Lm ∈ Λ where
m ∈ Zd is the unique vector for which x + y − Lm ∈ Λ.
To describe Fourier transforms, we employ the corresponding discrete dual lattice
Λ∗ := Λ/L = (L −1 Zd )/Zd . If needed, we use the parametrisation implied by the
 d
above notation; for instance for even L, we use Λ∗ = − 21 + L1 , . . . , 21 − L1 , 21 .
The arithmetic on Λ∗ is then performed modulo 1, i.e. using the arithmetic inherited
from the d-torus Td = Rd /Zd ⊃ Λ∗ .
Such periodic arithmetic is particularly well adapted for use of discrete Fourier
transforms. For a function f : Λ → C we take its Fourier transform to be the function

f : Λ∗ → C defined by the formula


f (k) := e−i2πk·x f (x) , k ∈ Λ∗ .
x∈Λ
Kinetic Theory and Thermalization of Weakly Interacting Fermions 3

The inverse transform of g : Λ∗ → C is then given by g̃ : Λ → C defined by


1  i2πk·x
g̃(x) := dk ei2πk·x g(k) = e g(k) , x ∈ Λ.
Λ∗ |Λ| k∈Λ∗

Here and in the following, we use the shorthand notation


1 
dk · · · = ··· .
Λ∗ |Λ| k∈Λ∗

On a finite lattice, the discrete Fourier transform is always pointwise invertible, i.e.
for all x ∈ Λ, k ∈ Λ∗ , ( f )˜(x) = f (x), (g̃)(k) = g(k).
We assume that the dominant free evolution is defined by giving the dispersion
relation ω : Td → R corresponding to free evolution after a thermodynamic limit
L → ∞ has been taken. More precisely, we let the periodic lattice hopping potential
α : Λ → R be defined by the inverse Fourier transform of the map ω|Λ∗ ,

α(x; L) := dk ei2πk·x ω(k) , x ∈ Λ. (1)
Λ∗

The function α determines the free n-particle Hamiltonian H0(n) by its action on
n-particle wave vectors ψ : Λn → C,


n 
H0(n) ψ(x1 , . . . , xn ) = α(x j − y)ψ(x1 , . . . , y, . . . , xn ) .
j=1 y∈Λ

The above construction allows an L-independent diagonalisation of H0(n) by taking

the discrete Fourier transform:


n
(H0(n) ψ)(k1 , . . . , kn ) = (k1 , . . . , kn ) .
ω(k j ) ψ
j=1

We assume that the dispersion relation is smooth and symmetric, ω(−k) = ω(k).
Then α(x; L) is always real, and denoting the inverse Fourier transform (i.e. the
Fourier series) of ω by α, we then have α(x; L) → α(x) for each fixed x ∈ Zd as
L → ∞. In addition, the range of α is finite, in the sense that |α(x)| decreases faster
than any power as |x| → ∞.
An explicit often considered example case is nearest neighbour hopping. This
corresponds to
 d
ω(k) = c − cos(2π kν ) ,
ν=1
4 J. Lukkarinen

where c ∈ R is any constant. For instance, choosing c = d, one obtains the standard
discrete Laplacian,

 1
d
α(x − y)ψ(y) = (2ψ(x) − ψ(x − eν ) − ψ(x + eν )) ,
y∈Λ
2 ν=1

where eν denotes the unit vector in direction ν. For vectors ψ which are obtained by
taking values of a slowly varying function ψ : Rd → C, the right-hand side can be
approximated by − 21 ∇ 2 ψ(x). Therefore, in this case one may also think of H0(n) as
a discrete approximation of the standard free n-particle Hamiltonian, with particle
mass normalised to one.
We construct a pair interaction potential V (x; L) analogously, starting from its
Fourier transform V : Td → C and defining

V (x; L) := (k) ,
dk ei2πk·x V x ∈ Λ. (2)
Λ∗

To make the potential real-valued and symmetric, we assume that V  is real-valued

(n)
and symmetric. The n-particle pair interaction potential V is then defined via the
formula
1 
n
V (n) (x1 , . . . , xn ; L) := V (xi − xi ; L) .
2
i ,i=1;i =i

The potential function acts as an multiplication operator on wave vectors, and we

do not make any distinction in the notation between the function and the operator.
Thus, if ψ : Λn → C is an n-particle wave vector, then

V (n) ψ(x1 , . . . , xn ) = V (n) (x1 , . . . , xn ; L)ψ(x1 , . . . , xn ) .

Naturally, if n = 1, we have V (n) = 0.

After these preliminaries, we define the full n-particle Hamiltonian by choosing
an interaction strength λ ≥ 0, setting Hλ(0) = 0, and for n ≥ 1 defining

Hλ(n) := H0(n) + λV (n) .

The corresponding evolution equation for n-particle wave vectors ψ(t) is

∂t ψ(t) = −iHλ(n) ψ(t) .

The n-particle Hilbert space is here finite-dimensional, Hn = (CΛ )⊗n = CΛ . By

n

construction, the evolution preserves particle number and each Hλ(n) is a bounded
∞
self-adjoint operator on Hn . Thus their direct sum Hλ := Hλ(n) defines a self-
n=0
Kinetic Theory and Thermalization of Weakly Interacting Fermions 5

∞
adjoint operator on the full Fock space F := Hn . More precisely, the domain
n=0
of the operator is

∞

D(Hλ ) := Ψ ∈ F Hλ(n) Ψn 2 < ∞ ,
n=0

and the action of Hλ on Ψ = (Ψ0 , Ψ1 , . . .) ∈ D(Hλ ) yields the vector (Hλ(n) Ψn )∞ n=0 ∈
F . (The proof of these properties can be found for instance in [2, Theorem 2.23].)
An analogous construction holds for the potential terms V (n) alone, and the corre-
sponding full Fock space operator is denoted by V ; clearly, λV = Hλ − H0 on the
domain of Hλ .
Each H0(n) and V (n) clearly commutes with permutations of particle labels (i.e. with
all of the operators Q π defined by (Q π ψ)(x1 , . . . , xn ) = ψ(xπ(1) , . . . , xπ(n) ), where
π is any permutation of {1, 2, . . . , n}). Thus Hλ leaves invariant both the fermionic
Fock space F− , containing those Ψ ∈ F for which each Ψn is antisymmetric under
permutations of particle labels, and the bosonic Fock space F+ , containing only
symmetric Ψn .
From now on, we focus on the corresponding fermionic lattice system which is
defined by wave vectors Ψ (t) ∈ F− and the semigroup generated by the restriction
of Hλ to F− . Since wave vectors with only finitely many non-zero particle sectors
belong to D(Hλ ) and form a dense set in F− , we find that for any Ψ (0) ∈ F− , the
n-particle sector of the time-evolved wave function can be obtained by solving the
matrix evolution equation

∂t Ψn (t) = −iHλ(n) Ψn (t) ,

with initial data Ψn (0).

2.1 Dynamics in Terms of Creation and Annihilation

Operators

Antisymmetry of wave vectors is one of the most important features of fermionic

quantum systems, and it can alter the properties of time evolution significantly.
Controlling the effect of antisymmetry is difficult in the above formulation of the
time evolution. A better alternative is offered by representing the time evolution
as an evolution equation of the corresponding fermionic creation and annihilation
operators. We summarise their main properties below and refer to [3, Section 5.2]
for more mathematical details.
6 J. Lukkarinen

In the present finite lattice case, the Fock space has been constructed using a one-
particle space h := CΛ and the corresponding (distinguishable) n-particle sectors
Hn := h⊗n = CΛ . Let P−(n) denote the orthogonal projection onto the subspace of
n

antisymmetric functions in Hn ; explicitly,

1 
(P−(n) ψ)(x1 , . . . , xn ) = (−1)π ψ(xπ(1) , . . . , xπ(n) ) ,
n! π∈S
n

where Sn denotes the group of permutations of the set {1, 2, . . . , n} and (−1)π is the
sign of the permutation π ∈ Sn . Since we consider a system of identical fermions,
at any time, a wave vector Ψ ∈ F− satisfies P−(n) Ψn = Ψn for all n.
Given a one-particle wave vector g ∈ h, we define the corresponding annihila-
tion operator a(g) as the map which takes a vector Ψ ∈ F− and removes the first
particle from each of its sectors, with a weight proportional to the overlap with g.
More precisely, for a fixed particle number n ≥ 1, there is a unique bounded linear
map An (g) : Hn → Hn−1 such that for any collection of one-particle wave vectors
f j ∈ h, ⎛ ⎞
n
√ 
n
An (g) ⎝ f j ⎠ = ng, f 1  fj ,
j=1 j=2

where g, f  is the one-particle 

scalar product, defined here conjugate linear in
the first argument, i.e. g, f  = x∈Λ g(x)∗ f (x). We then define the fermionic
annihilation operator a(g) : F− → F− by the rule

(a(g)Ψ )n = P−(n) An+1 (g)P−(n+1) Ψn+1 = P−(n) An+1 (g)Ψn+1 , n ≥ 0 , Ψ ∈ F− .

In general, annihilation operators are unbounded on the appropriate Fock space,

and one has to worry about the domain of the operator in its definition. However, it
is a remarkable consequence of the antisymmetrisation
√ that a(g) is in fact a bounded
operator on F− , and the normalisation n added above guarantees that its operator
norm is the same as the norm of the wave vector g, i.e. we always have a(g) =
gh .
The adjoint of a(g), which we denote here by a ∗ (g), is called the creation operator
at the vector g ∈ h. The creation operator can indeed be interpreted as creating a
particle with wave vector g at the first position (and hence shifting the labels of the
existing particles by one). This interpretation is based on a more direct construction
analogous to the one for a(g) above. Namely, there is a unique bounded linear
map Cn (g) : Hn → Hn+1 such that for any collection of one-particle wave vectors
f j ∈ h, ⎛ ⎞
n
√
Cn (g) ⎝ f j ⎠ = n + 1 g ⊗ f1 ⊗ · · · ⊗ fn .
j=1
Kinetic Theory and Thermalization of Weakly Interacting Fermions 7

We also set C0 (g)1 = g ∈ H1 . The fermionic creation operator is then given by

c(g) : F− → F− , and it satisfies (c(g)Ψ )0 = 0, and

(c(g)Ψ )n = P−(n) Cn−1 (g)P−(n−1) Ψn−1 = P−(n) Cn−1 (g)Ψn−1 , n ≥ 1 , Ψ ∈ F− .

One can check that then indeed c(g) = a ∗ (g) which implies that also c(g) = gh .
One important reason why working with the creation and annihilation operators
simplifies the analysis of time evolution is that they satisfy fairly simple algebraic
rules for swapping the order of any two such operators. Namely, they satisfy the
following canonical anticommutation relations: for any one-particle vectors f, g ∈
h, we have

a( f )a(g) + a(g)a( f ) = 0 = a( f )∗ a(g)∗ + a(g)∗ a( f )∗ ,

a( f )a(g)∗ + a(g)∗ a( f ) =  f, g1 , (3)

where ‘1’ denotes the identity operator on F− . In particular, a( f )2 = 0 = a ∗ ( f )2 ,

and if (e ) is any orthonormal basis of h, we have

a(e )a(e )∗ + a(e )∗ a(e ) = 1{ = }1 ,

with 1{P} denoting the generic characteristic function of the condition P: we define
1{P} = 1, if P is true, and 1{P} = 0, if P is false.
Moreover, tensor products in Hn are conveniently expressed in terms of products
of creation operators acting on the vacuum Ω = (1, 0, 0, . . .) ∈ F− . Namely, if g j ∈
h, j = 1, 2, . . . , n, are given, then ⊗ j g j ∈ Hn after antisymmetrisation defines a
vector Ψ ∈ F− by setting all other components to zero, i.e. setting Ψn = P−(n) (⊗ j g j )
and Ψm = 0, for m = n. This vector can also be obtained from

1
Ψ = √ a ∗ (g1 ) · · · a ∗ (gn )Ω . (4)
n!

The collection of creation and annihilation operators corresponding to the standard

unit vector orthonormal basis (ex )x∈Λ , where (ex ) y = 1{x=y} for all x, y ∈ Λ, is of
particular interest to us. We employ the following standard shorthand notations:

a(x) := a(ex ) , a ∗ (x) := a ∗ (ex ) = a(x)∗ , x ∈ Λ. (5)

These operators can be thought of as annihilating or creating a particle at the site x.

By (3), they satisfy the following simple anticommutation relations for any x, y ∈ Λ,

a(x)a(y) + a(y)a(x) = 0 = a(x)∗ a(y)∗ + a(y)∗ a(x)∗ ,

a(x)a(y)∗ + a(y)∗ a(x) = 1{x=y} 1 . (6)
8 J. Lukkarinen

We can also use the creation operators to generate an orthonormal basis for F− .
For this, first define

e(x1 , . . . , xn ) := a ∗ (x1 ) · · · a ∗ (xn )Ω , xi ∈ Λ , i = 1, 2, . . . , n .

The orthonormal basis may be constructed by collecting all non-repeating sequences

of arbitrary length and then choosing one representative for each collection of
sequences which differ by a permutation of particle labels. The actual choice does
not play much role: if (xi ) ∈ Λn and π ∈ Sn is some permutation, then by the anti-
commutation relations

e(xπ(1) , . . . , xπ(n) ) = (−1)π e(x1 , . . . , xn ) ,

and hence the choice merely affects signs of the basis vectors.
After these preliminaries, it is straightforward to check that wave vectors and
interaction potentials may also be represented using the creation and annihilation
operators. Namely, if Ψ ∈ F− , n ∈ N, and x ∈ Λn , we have

Ψn (x1 , . . . , xn ) = ⊗i=1
n
exi , Ψn H n = P−(n) (⊗i=1
n
exi ), Ψn H n ,

and hence by (4),

1
Ψn (x1 , . . . , xn ) = √ a ∗ (x1 ) · · · a ∗ (xn )Ω, Ψ F − .
n!

Moreover, the anticommutation relations imply that if x, y and xi ∈ Λ,

i = 1, 2, . . . , n, then

a ∗ (x)a(y)a ∗ (x1 ) · · · a ∗ (xn )Ω


n
= 1{y=xi } a ∗ (x1 ) · · · a ∗ (xi−1 )a ∗ (x)a ∗ (xi+1 ) · · · a ∗ (xn )Ω . (7)
i=1

Using these two properties it is now straightforward to check that the earlier defined
operators H0 and V on the fermionic Fock space have the following representations
in terms of creation and annihilation operators:

H0 = α(x − y; L)a(x)∗ a(y) , (8)
x,y∈Λ
1 
V = V (x − y; L)a(x)∗ a(y)∗ a(y)a(x) . (9)
2 x,y∈Λ
Kinetic Theory and Thermalization of Weakly Interacting Fermions 9

The above right-hand sides are finite sums in the Banach space of bounded operators
on F− , and thus H0 , V , and Hλ = H0 + λV are also bounded operators on the
fermionic Fock space.
The time evolution of any initial data Ψ (0) ∈ F− under the semigroup Ut :=
e−it Hλ can be solved if we can solve the time evolution of the annihilation operators,
i.e. it suffices to study
a(x, t) := eit Hλ a(x)e−it Hλ ,

and its adjoint

a ∗ (x, t) := eit Hλ a ∗ (x)e−it Hλ .

This follows from our definition that the Hamiltonian acts trivially on the vacuum
sector, (Hλ )0 = 0, and thus

a ∗ (x1 , t) · · · a ∗ (xn , t)Ω = eit Hλ a ∗ (x1 ) · · · a ∗ (xn )Ω ,

implying that
1
Ψn (x1 , . . . , xn , t) = √ a ∗ (x1 ) · · · a ∗ (xn )Ω, e−it Hλ Ψ (0)F −
n!
1
= √ a ∗ (x1 , t) · · · a ∗ (xn , t)Ω, Ψ (0)F − .
n!

Since the Hamiltonian is a bounded operator, we can directly differentiate the

definition and obtain

∂t a(x, t) = −ieit Hλ [a(x), Hλ ]e−it Hλ .

The computation of the commutator is straightforward using the anticommutation

relations, yielding
 
[a(x), Hλ ] = α(x − y; L)a(y) + λ V (x − y; L)a(y)∗ a(y)a(x) .
y∈Λ y∈Λ

Therefore, we find that in order to solve the original (linear) evolution equation in the
fermionic Fock space, it suffices to solve the following non-linear operator evolution
equation on the space of bounded operators on F− ,
 
∂t a(x, t) = −i α(x − y; L)a(y, t) − iλ V (x − y; L)a ∗ (y, t)a(y, t)a(x, t) .
y∈Λ y∈Λ
(10)

In Fourier variables, after defining



a (k, t) := e−i2π x·k a(x, t) ,
x∈Λ
10 J. Lukkarinen

we obtain

∂t
a (k, t) = −iω(k)a (k, t)

− iλ (k1 + k2 )a∗ (k1 , t)
dk1 dk2 dk3 δΛ (k − k1 − k2 − k3 )V a (k2 , t)
a (k3 , t) ,
(Λ∗ )3
(11)

a (k, t)]∗ =
where δΛ (k) := |Λ|1{k=0 mod Λ∗ } is a ‘discrete Dirac δ-function’ and [
a (−k, t).
∗

2.2 Fermionic Systems with Spin Interactions and the

Hubbard Model

Spin is an integral part of description of quantum mechanical particles. For instance,

by the spin–statistics relation, all fermionic particles possess a half-integer spin. In
particular, the spin cannot be zero, so the above fermionic description is not yet
completely adequate for physical fermions.
Spin is a one-particle property, and hence affects the definition of the one-particle
Hilbert space h above. It is determined by a half-integer value S ∈ N0 /2, resulting
in 2S + 1 new ‘internal’ degrees of freedom which are labelled by values in σ S :=
{−S, −S + 1, . . . , S}. There are several equivalent ways of defining the wave vector
of a particle with a non-zero spin: one can either think that they are multicomponent
wave vectors, ψ(x) ∈ CσS , or that each lattice site is augmented with D extra degrees
of freedom, ψ(x, σ ) ∈ C, σ ∈ σ S . These descriptions are quantum mechanically
equivalent since the identification

ψ(x)σ = φ(x, σ ) = ex ⊗ eσ , φ

provides a mapping ψ → φ which turns out to be a Hilbert space isomorphism

between ⊕σ ∈σS L 2 (Λ) and L 2 (Λ × σ S ). The second equality above also yields an
isomorphism, namely, the standard one between L 2 (Λ × σ S ) and L 2 (Λ) ⊗ L 2 (σ S ).
Hence, most of the discussion in the previous sections holds verbatim if we replace
x ∈ Λ by (x, σ ) ∈ Λ × σ S . The main differences come from the physical restrictions
for the spin interactions which have no need to be ‘translation invariant’ in the spin
degrees of freedom. Thus, Fourier transforming the spin degrees is not helpful and,
instead, one should try to aim at simplifications by finding other unitary transforma-
tions which diagonalise at least part of the Hamiltonian.
One case which reduces to the discussion without spin occurs when the total
Hamiltonian H can be diagonalised with respect to the spin degrees of freedom, i.e.
if there is a unitary transformation U for which U ∗ HU = ⊕σ ∈σS Hσ . Then after the
unitary transformation each spin component evolves independently from the others
and thus it satisfies the ‘spinless’ equations of the previous section.
Kinetic Theory and Thermalization of Weakly Interacting Fermions 11

Spatially translation invariant generalisations of the previous weakly interacting

Hamiltonians are determined by the operators
 
H0 = ασ σ (x − y; L)a(x, σ )∗ a(y, σ ) , (12)
x,y∈Λ σ,σ ∈σ S
1  
V = Vσ σ (x − y; L)a(x, σ )∗ a(y, σ )∗ a(y, σ )a(x, σ ) . (13)
2 x,y∈Λ σ,σ ∈σ
S

The functions ασ σ (x; L) and Vσ σ (x; L) are constructed as in (1) and (2), using some
given ωσ σ : Td → R and V σ σ : Td → R, for each σ, σ . We require H0 to be self-
adjoint and the interaction symmetric under spatial inversions, and this is guaranteed
by assuming that ω(−k) = ω(k) = ω(k)∗ , as S × S -matrices. Similarly, the self-
adjointness of V can be guaranteed by assuming that each V (k) is a Hermitian matrix
and that they satisfy an additional symmetry property V σ σ (−k) = V σ σ (k) related
to particle permutation invariance.
One well-studied example of this type is the Hubbard model which concerns
spin- 21 fermions like electrons. Then S = 21 and usually one simplifies the discussion
by labelling the spin degrees of freedom {− 21 , 21 } using the sign, i.e. using the set
2σ S = {−1, 1} for labelling. In the Hubbard model, the free evolution is taken to be
fully spin rotation invariant,
 
H0 = α(x − y; L)a(x, σ )∗ a(y, σ ) , (14)
x,y∈Λ σ =±1

and thus depending only on one dispersion

 relation function which is typically chosen
to be nearest neighbour, ω(k) = − dν=1 cos(2π kν ). The pair interactions in the
Hubbard model are taken to be onsite only,

1 
V = Vσ σ a(x, σ )∗ a(x, σ )∗ a(x, σ )a(x, σ ) , (15)
2 x∈Λ σ,σ =±1

and, since a(x, σ )2 = 0 and Vσ σ = Vσ σ (0; L), σ, σ ∈ {±1}, form a real symmetric
2 × 2 matrix, without loss of generality, we may set Vσ σ = 0 and use V+− = V−+ as
the sole real parameter. It is usually included in the definition of the coupling λ, and
thus the general fermionic spin- 21 onsite interactions are covered by the interaction1

VHubbard = a(x, +)∗ a(x, −)∗ a(x, −)a(x, +) . (16)
x∈Λ


x n(x, +)n(x, −) where
1 The most standard notation for the Hubbard model uses the potential U

n(x, σ ) :=a(x, σ )∗ a(x, σ ).

This is seen to be equivalent to the present case after setting U = λ
and using the anticommutation relations.
12 J. Lukkarinen

Let us point out that onsite potentials fall into the class of translation invariant
potentials studied in the previous subsection. Namely, they correspond to choosing
potentials whose Fourier transforms are constant, V σ σ (k) = Vσ σ for all k ∈ Td .
The main difficulties compared to deriving the evolution equations in the earlier
discussed case are notational. We skip the parts which are similar to the earlier
computations, and merely record the outcome in a form which is easy to use in
computations involving products of creation and annihilation operators.
We label annihilation operators with an additional label τ = −1 and creation
operators with τ = +1, and consider their dynamics after Fourier transform of the
spatial degrees of freedom. Explicitly, we define

a(k, σ, −1, t) := 
a (k, σ, t) = e−i2π x·k a(x, σ, t) , (17)
x∈Λ

a(k, σ, +1, t) := a∗ (k, σ, t) = e−i2π x·k a ∗ (x, σ, t) . (18)
x∈Λ

These operators are connected via operator adjoints, [a(k, σ, τ, t)]∗ = a(−k, σ,
−τ, t). Since now
 
∂t a(x, σ, t) = −i ασ σ (x − x ; L)a(x , σ , t)
x ∈Λ σ ∈σ S
 
− iλ Vσ σ (x − x ; L)a ∗ (x , σ , t)a(x , σ , t)a(x, σ, t) ,
x ∈Λ σ ∈σ S

the above operators satisfy the following closed evolution equations:


∂t a(k, σ, τ, t) = iτ ωσ σ (k; τ )a(k, σ , τ, t)
σ ∈σ S
 
+ iτ λ dk1 dk2 dk3 δΛ (k − k1 − k2 − k3 )
σ1 ,σ2 ,σ3 ∈σ S (Λ∗ )3

σ,σ1 ,σ2 ,σ3 (k1 , k2 , k3 ; τ )a(k1 , σ1 , 1, t)a(k2 , σ2 , τ, t)a(k3 , σ3 , −1, t) , (19)

×V

where ωσ σ (k; −1) := ωσ σ (k), ωσ σ (k; +1) := ωσ σ (k), and

σ,σ1 ,σ2 ,σ3 (k1 , k2 , k3 ; −1) = 1{σ1 =σ2 ,σ3 =σ } V

V σ σ2 (k1 + k2 ) ,
σ,σ1 ,σ2 ,σ3 (k1 , k2 , k3 ; +1) = 1{σ1 =σ,σ3 =σ2 } V
V σ σ2 (k2 + k3 ) .

Here we need the above equations only in two special cases. First, if there is no
spin, the equation reduces to
Kinetic Theory and Thermalization of Weakly Interacting Fermions 13

∂t a(k, τ, t) = iτ ω(k)a(k, τ, t) + iτ λ dk1 dk2 dk3 δΛ (k − k1 − k2 − k3 )
(Λ∗ )3
(k1 , k2 , k3 ; τ )a(k1 , 1, t)a(k2 , τ, t)a(k3 , −1, t) ,
×V (20)

(k1 , k2 , k3 ; −1) = V
with V (k1 + k2 ) and V
(k1 , k2 , k3 ; 1) = V
(k2 + k3 ). Second, for
the Hubbard model, the equations can be simplified into

∂t a(k, σ, τ, t) = iτ ω(k)a(k, σ, τ, t) + iτ λ dk1 dk2 dk3 δΛ (k − k1 − k2 − k3 )
(Λ∗ )3
× a(k1 , τ σ, 1, t)a(k2 , −σ, τ, t)a(k3 , −τ σ, −1, t) . (21)

3 States, Reduced Density Matrices and Truncated

Correlation Functions

A state in classical mechanics is a probability measure describing the distribution of

positions and velocities of the particles at some fixed time. Thus it can be used to
compute the statistics of all observables, i.e. measurable functions of the positions
and velocities at that time. In Hamiltonian mechanics, an initial state given at time
t = 0 determines the state at all times t ∈ R. Often, it is simpler to study the evolution
of physical properties of the system by inspecting the evolution starting from some
suitably chosen random initial state rather than from a deterministic state with fixed
values for the initial positions and velocities of the particles.
A state at time t in quantum mechanics is defined as a map ρt which associates
to each observable A a number ρt [A] which gives the limiting value for statistical
averages of this observable measured in repeated experiments. This is analogous
to the expectation value map under the probability measure which defines the state
in the classical case. The more precise mathematical definition of a state takes two
ingredients: the collection of observables A , which is assumed to be some subspace
of bounded operators, closed under adjoint and containing the identity operator, and
a positive linear functional ρ : A → C of norm 1.
For instance, a Borel probability measure μ of wave vectors Ψ ∈ H , Ψ  = 1,
generates a state by setting for any bounded operator A on H

ρ[A] := μ(dψ) ψ, Aψ .

Most often a state is determined by giving a trace-class operator ρ on H such that ρ

is positive, Tr ρ = 1, and setting ρ[A] = Tr[ρ A] for all A ∈ A . Such an operator ρ
is called the density matrix of the state (note that we do not make a distinction in the
notation between the state and its density matrix). If the Hilbert space is separable,
such as our Fock spaces are, then for instance all states given by the above Borel
probability measures have a density matrix associated with them.
14 J. Lukkarinen

The n-th reduced density matrix ρn is an analogous quantity which is obtained

from the full density matrix by taking a partial trace over the degrees of freedom
which concern particle labels higher than n. The general construction is discussed in
[3, Section 6.3.3] and in [1, Section 3], but there is a more direct definition available
for the present system of lattice fermions: Given a state ρ on the fermionic Fock
space, we first define

ρn (z 1 , z 1 , . . . , z n , z n ) := ρ[a ∗ (z 1 ) · · · a ∗ (z n )a(z n ) · · · a(z 1 )] . (22)

Here each z i and z i belongs to the one-particle label set, i.e. z i ∈ Λ in the spinless
case and z i ∈ Λ × σ S for spin-S particles. The collection of these complex numbers
defines the reduced density matrix ρn , which is a positive operator on h⊗n , via the
formula
⊗i z i , ρn (⊗i z i ) = ρn (z 1 , z 1 , . . . , z n , z n ) .

In quantum mechanics, given an initial density matrix ρ(0) = ρ, the expectation

of a time-evolved observable A(t) = Ut∗ AUt satisfies

ρ[A(t)] = Tr[ρUt∗ AUt ] = Tr[Ut ρUt∗ A] ,

by cyclicity of trace. Hence, we define the time-evolved density matrix ρ(t) :=

Ut ρUt∗ for which ρ(t)[A] = ρ[A(t)]. The reduced time-evolved density matrices
may thus be obtained as expectations of time-evolved creation and annihilation oper-
ators: by replacing each a(z) in (22) by a(z, t) = Ut∗ a(z)Ut , we obtain the reduced
density matrix ρ(t)n .
Considering the earlier observation that time-evolved annihilation operators suf-
fice to determine the time evolution of wave vectors, it is not surprising that reduced
density matrices play an important role in the physics of quantum fluids. For instance,
the expectation of the hopping Hamiltonian H0 may be computed from ρ(t)1 by the
formula
 
ρ(t)[H0 ] = ασ σ (x − y; L)ρ(t)1 ((x, σ ), (y, σ )) .
x,y∈Λ σ,σ ∈σ S

Indeed, for kinetic theory, the central goal is to describe the evolution of ρ(t)1 , a
positive operator on h, in the limit of weak coupling.
In fact, there is a class of fermionic states, called quasifree states, for which ρ1
uniquely determines all other reduced density matrices: if ρ is quasifree, then for all
n ≥ 1 the corresponding density matrix is given as a determinant of an n × n matrix,

ρn (z 1 , z 1 , . . . , z n , z n ) = det(ρ1 (z i , z j ))i, j=1,...,n .

To simplify analysis of states which are not quasifree but close to such, one can
introduce truncated correlation functions ρ T which are analogous to cumulants of
random variables in classical probability theory. The construction below applies to
Kinetic Theory and Thermalization of Weakly Interacting Fermions 15

a state ρ on a fermionic system which is even: it is assumed that an expectation of

any observable remains invariant if we change a(z) to −a(z) for all z. As explained
in more detail in [3, pp. 42–43], given an even state ρ to each even length sequence
(a1 , a2 , . . . , am ) of creation and annihilation operators one may associate a truncated
expectation ρ T [a1 , a2 , . . . , am ] such that the expectation of any product of even
length can be expressed as a sum over partitions. Explicitly,
 
ρ[a I ] = ε() ρ T [a S ] , (23)
∈P 2 (I ) S∈

where I = (1, 2, . . . , n), a I := a1 · · · an , P2 (I ) denotes the collection of partitions

of I into even length subsequences, ε() is the sign of the permutation which takes I
to  = (S1 , . . . , Sm ), and for a subsequence S = (s1 , . . . , sm ) of I we have used the
shorthand notation a S = (as1 , . . . , asm ). Note that odd sequences for even states have
always zero expectation, so this is the antisymmetrised analogue of the moments-to-
cumulants formula of probability.
The above definition requires careful consideration of the signs of each term. The
following identity can also serve as a basis for a recursive definition of the truncated
expectations:

ρ[a I ] = ε(S, I \ S)ρ T [a S ]ρ[a I \S ] , (24)
m∈S⊂I

where m ∈ I is any fixed label and ε(S, I \ S) is the sign of the permutation I →
(S, I \ S). (Note that all terms where S has an odd length are zero in the sum, since
then also I \ S is odd, so we could have restricted the sum to even subsequences
here.) For instance, ρ T [a1 , a2 ] = ρ[a1 a2 ], and for n = 4 we have

ρ[a1 a2 a3 a4 ] = ρ T [a1 , a2 , a3 , a4 ]
+ ρ T [a1 , a2 ]ρ[a3 a4 ] − ρ T [a1 , a3 ]ρ[a2 a4 ] + ρ T [a1 , a4 ]ρ[a2 a3 ] ,

and thus

ρ T [a1 , a2 , a3 , a4 ] := ρ[a1 a2 a3 a4 ]
− ρ[a1 a2 ]ρ[a3 a4 ] + ρ[a1 a3 ]ρ[a2 a4 ] − ρ[a1 a4 ]ρ[a2 a3 ] ,

and, in accordance with (23), also

ρ[a1 a2 a3 a4 ] = ρ T [a1 , a2 , a3 , a4 ]
+ ρ T [a1 , a2 ]ρ T [a3 , a4 ] − ρ T [a1 , a3 ]ρ T [a2 , a4 ] + ρ T [a1 , a4 ]ρ T [a2 , a3 ] . (25)

The truncated correlation functions can be used to characterise quasifree states:

an even state ρ is quasifree if and only if ρ T [a1 , a2 , . . . , an ] = 0 for all n > 2. This
16 J. Lukkarinen

is completely analogous with characterisation of Gaussian measures by vanishing

of their higher order cumulants. Even for states which are not quasifree, the trun-
cated correlation functions enjoy properties which are typically not valid for direct
expectations:
1. If n > 2, then ρ T [a1 , a2 , . . . , an ] is completely antisymmetric with respect to
permutation of its arguments: if π ∈ Sn , we have ρ T [aπ(1) , aπ(2) , . . . , aπ(n) ] =
(−1)π ρ T [a1 , a2 , . . . , an ]. (For a proof, consider a basic odd permutation which
swaps two neighbouring labels m and m , and then use (24) and the anticommu-
tation relations.)
2. If ρ is an equilibrium Gibbs state at sufficiently small activity and corresponding
to a short-range interaction, all reduced density matrices are typically decaying
summably in the separation of their spatial arguments. For a precise statement and
assumptions under which this result holds, see [3, Theorem 6.3.21], and further
discussion can be found in [4]. In particular, keeping one of the sites fixed, Fourier
transforms of the reduced density matrices are typically uniformly bounded in
the lattice size L, unlike those of the corresponding expectations.

4 Weak Coupling Limit and Quantum Kinetic Theory

For kinetic theory, we are interested in the evolution of the first truncated reduced
density matrix ρ1 (x , σ , x, σ ; t) = ρ T [a ∗ (x , σ , t), a(x, σ, t)]. There is no differ-
ence between the truncated and direct reduced density matrices for the first reduced
density matrix of an even state of fermions but for higher order density matrices
there is a difference in their properties. Most notably, for systems which are even-
tually well approximated by Gibbs states of the type discussed in item 3 at the end
of Sect. 3, one would expect the truncated correlation functions to decay in the dis-
tance. Then, Fourier transforms in these variables are given by ‘nice’ functions, for
instance, uniformly bounded in the lattice size or with a uniformly bounded L 2 (dk)-
norm. In contrast, the Fourier transform of the corresponding moments would be a
fairly complicated sum over ‘δΛ -distributions’.
Here, we consider only initial data which are both gauge invariant and trans-
lation invariant. The first condition means that the initial data does not contain
correlations between different particle sectors, and this property is preserved by
the present type of evolution. It simplifies the resulting analysis since for gauge
invariant states all moments, which do not have the same number of creation and
annihilation operators, are zero. For instance, then ρ[a(y, σ , t)a(x, σ, t)] = 0 =
ρ[a ∗ (y, σ , t)a ∗ (x, σ, t)].
For translation invariance, we require that all moments are invariant under periodic
spatial translations of the lattice Λ. For the present translation invariant H0 and V ,
also this property is preserved by the time evolution. As a consequence, any one of
the spatial arguments of the correlation functions can be translated to the origin. In
particular, there is a function F : Λ × R → C2×2 for which
Kinetic Theory and Thermalization of Weakly Interacting Fermions 17

ρ1 (x , σ , x, σ ; t) = Fσ σ (x − x, t) .

The Wigner function is defined as the discrete Fourier transform of F,

 
Wσ σ (k, t) := e−i2π y·k Fσ σ (y, t) = dk ρ[a(k, σ , 1, t)a(k , σ, −1, t)] .
y∈Λ Λ∗

(26)

Using the properties of adjoints, it is straightforward to check that the so-defined

σ S × σ S matrix W (k, t) is always Hermitian. In addition, translation invariance may
be invoked to prove that

ρ[a(k, σ , 1, t)a(k , σ, −1, t)] = Wσ σ (k, t)δΛ (k + k ) . (27)

We also introduce the related notation W̃ for the corresponding expectation where
the order of the operators has been swapped. More precisely, we define as matrices

W̃ (k, t) := 1 − W (k, t) , (28)

where 1 denotes the diagonal unit matrix. By the anticommutation relations, then

ρ[a(k , σ, −1, t)a(k, σ , 1, t)] = W̃σ σ (k, t)δΛ (k + k ) . (29)

The quantum kinetic equation will concern the time evolution of the above Her-
mitian matrix-valued Wigner functions. There are a number of differences in the
computations depending on whether there are spin interactions present or not, and
we have split the discussion accordingly below.

4.1 Fermionic Boltzmann–Nordheim Equation

We begin with a case in which the spin degrees of freedom evolve independently. As
mentioned above, this case can be handled ignoring the spin degrees of freedom and
thus we can use the spinless results and notations. We adapt here the method intro-
duced in [5] for derivation of a phonon Boltzmann equation for the weakly non-linear
discrete Schrödinger equation from the evolution hierarchy of truncated correlation
functions. For comparison, a derivation of the Boltzmann–Nordheim equation using
direct perturbation expansions of moments and their graph representations can be
found in [1].
It should be stressed that neither method currently produces a mathematically
rigorous derivation of fermionic kinetic theory. In particular, it is not yet known
which precise assumptions are needed for the kinetic approximation to work nor are
there any rigorous bounds for the accuracy of the approximation. From the point of
18 J. Lukkarinen

view of the truncated correlation function hierarchy, the key missing ingredient is a
control of the evolution of decay properties of correlation functions. Here, we do not
go into any detail about the role played by the terms ignored in the derivations below
but more details about why their effects are in general expected to be lower order in
the weak coupling limit λ → 0 can be found in [1, 5].
Let us also point out one case in which rigorous control has been possible: in [6],
the kinetic scaling limit of time correlations of equilibrium distributed fields with
discrete non-linear Schrödinger evolution are proven to follow the above scenario. In
this case, the state itself is stationary and the good decay properties of the truncated
correlation functions are provided by the initial data which can be studied with
methods from equilibrium statistical mechanics.
Differentiating (26) and recalling the adjoint relations yields the following
representation for the time derivative of the Wigner function of translation invariant
states:

∂t Wσ σ (k, t)

 
= dk ρ[∂t a(k, σ , 1, t)a(k , σ, −1, t)] + ρ[∂t a(−k , σ, 1, t)a(−k, σ , −1, t)]∗
∗
Λ
 
= dk ρ[∂t a(k, σ , 1, t)a(k , σ, −1, t)] + ρ[∂t a(k, σ, 1, t)a(k , σ , −1, t)]∗ . (30)
Λ∗

Thus for a translation invariant states of fermions without spin, we have

 
∂t W (k, t) = 2 Re dk ρ[∂t a(k, 1, t)a(k , −1, t)] . (31)
Λ∗

We use (20) to compute the derivative, yielding

 
dk ρ[∂t a(k, 1, t)a(k , −1, t)] = iω(k) dk ρ[a(k, 1, t)a(k , −1, t)]
Λ∗ Λ∗

+ iλ (k2 + k3 )δΛ (k − k1 − k2 − k3 )
dk1 dk2 dk3 dk4 V
(Λ∗ )4
× ρ[a(k1 , 1, t)a(k2 , 1, t)a(k3 , −1, t)a(k4 , −1, t)] . (32)

The first term on the right is purely imaginary and does not contribute to the real
part. In the second term, the expectation is antisymmetric with respect to the swap
k1 ↔ k2 , and thus we can conclude that
   
∂t W (k, t) = Re iλ dk1 dk2 dk3 dk4 V(k2 + k3 ) − V (k1 + k3 )
∗
(Λ )4

× δΛ (k − k1 − k2 − k3 )ρ[a(k1 , 1, t)a(k2 , 1, t)a(k3 , −1, t)a(k4 , −1, t)] .
(33)
Kinetic Theory and Thermalization of Weakly Interacting Fermions 19

We represent the remaining expectation in terms of truncated expectations using

 is real, all terms involving second-order truncated correlation functions
(25). Since V
produce terms which are purely imaginary and, hence, they do not contribute to the
derivative of the Wigner function. Therefore,
   
∂t W (k, t) = Re iλ dk1 dk2 dk3 dk4 V(k2 + k3 ) − V
(k1 + k3 )
(Λ∗ )4

× δΛ (k − k1 − k2 − k3 )ρ T [a(k1 , 1, t), a(k2 , 1, t), a(k3 , −1, t), a(k4 , −1, t)] .
(34)

Computation of derivatives of higher order truncated correlation functions would

be simplified by introducing the associated Wick polynomials, as was observed in
[5] for commuting fields. However, it is still possible to work out the necessary
combinatorics and cancellations by hand for the fourth-order terms which are needed
to compute the collision operator of kinetic theory. Namely, after a somewhat lengthy
computation employing the symmetry of the function V , one finds that
 
∂t e−it (ω1 +ω2 −ω3 −ω4 ) ρ T [a(k1 , 1, t), a(k2 , 1, t), a(k3 , −1, t), a(k4 , −1, t)]
 
= iλe−it (ω1 +ω2 −ω3 −ω4 ) δΛ (k1 + k2 + k3 + k4 ) V (k2 + k3 ) − V(k1 + k3 )

× W̃ (k2 )W (−k3 )W (−k4 ) − W (k1 )W (−k3 )W (−k4 )

+ W (k1 )W (k2 )W (−k4 ) − W (k1 )W (k2 )W̃ (−k3 )
+ (higher order truncated functions) , (35)

where we have introduced the shorthand notations ωi := ω(ki ), W̃ = 1 − W , and

each W and W̃ factor is evaluated at t.
We then integrate the above time derivatives from 0 to t. The terms involving
higher order truncated functions (4:th and 6:th in (35)), as well as the substitution
term involving the fourth-order truncated correlation at time 0, are expected to con-
tribute only terms which are subleading in λ at the kinetic timescales t ∝ λ−2 , due
to the ‘integrals’ over the oscillatory phase factors. The remaining terms yield the
approximation
 t  t  
W (k, t) − W (k, 0) ≈ dt ds Re −λ 2
dk1 dk2 dk3 dk4 ei(t −s)(ω1 +ω2 −ω3 −ω4 )
0 0 (Λ∗ )4
 
× V(k2 + k3 ) − V
(k1 + k3 ) 2 δΛ (k − k1 − k2 − k3 )δΛ (k1 + k2 + k3 + k4 )

× W̃ (k2 )W (−k3 )W (−k4 ) − W (k1 )W (−k3 )W (−k4 )

+ W (k1 )W (k2 )W (−k4 ) − W (k1 )W (k2 )W̃ (−k3 ) , (36)
20 J. Lukkarinen

where each W and W̃ factor is evaluated at s. Inside the integrand −k4 = k. Hence,
integration over k4 is straightforward and swapping the sign of k3 , the order of
time integrals, and denoting Wi := W (ki , s) and W̃i := 1 − Wi , we arrive at the
approximation
 t   t−s
W (k0 , t) − W (k0 , 0) ≈ λ 2
ds dk1 dk2 dk3 Re dr eir (ω1 +ω2 −ω3 −ω0 )
0 (Λ∗ )3 0
 2
× V(k2 − k3 ) − V
(k1 − k3 ) δΛ (k0
− k1 − k2 + k3 )
 
× −W̃2 W3 W0 + W1 W3 W0 − W1 W2 W0 + W1 W2 W̃3 , (37)

The real part of the remaining oscillatory time integral formally convergences
to π δ(ω0 − ω3 − ω1 − ω2 ) as t → ∞. In fact, the δ-function approximation should
only be used after the thermodynamic limit L → ∞ has been taken; for a finite lattice,
also values for which ω1 + ω2 − ω3 − ω0 is not exactly zero but close enough to zero
(e.g. o(L −2 )) will contribute to the collision term. Assuming that the thermodynamic
limit of the function W exists and using the same notation for the limit, we obtain
 t
W (k0 , t) − W (k0 , 0) ≈ ds CfBN [W (·, s)](k0 ) , (38)
0

where a relabelling k1 ↔ k3 yields the following more standard form of a fermionic

Boltzmann–Nordheim collision operator:

CfBN [W ](k0 ) := π λ 2
dk1 dk2 dk3 δ(ω0 + ω1 − ω2 − ω3 )
(Td )3
 
× V (k1 − k3 ) 2 δTd (k0 + k1 − k2 − k3 )
(k1 − k2 ) − V
 
× W̃1 W2 W3 − W0 W2 W3 − W0 W1 W̃2 + W0 W1 W3 . (39)

The kinetic equation obtained by replacing the approximation sign in (38) by an

equals sign is called the (spatially homogeneous) fermionic Boltzmann–Nordheim
equation. The term in square brackets in (39) is then usually written in a more
symmetric form as
W̃0 W̃1 W2 W3 − W0 W1 W̃2 W̃3 .

However, it should be noted that, since the highest order terms indeed cancel, the
collision operator has a non-linearity of third order, not of fourth order.

The above lattice kinetic
 theories have two conserved quantities, dk ω(k)W (k, t)
related to energy and dk W (k, t) related to particle density. The mathematical prop-
erties of their solutions have mainly been studied in the continuum case for which
instead of the lattice wave number k ∈ Td one uses the particle velocity v ∈ Rd and
the dispersion relation is ω(v) = v 2 in the nonrelativistic case. For the existence and
Another random document with
no related content on Scribd:
The Project Gutenberg eBook of The step on the
stair
This ebook is for the use of anyone anywhere in the United States
and most other parts of the world at no cost and with almost no
restrictions whatsoever. You may copy it, give it away or re-use it
under the terms of the Project Gutenberg License included with this
ebook or online at www.gutenberg.org. If you are not located in the
United States, you will have to check the laws of the country where
you are located before using this eBook.

Title: The step on the stair

Author: Anna Katharine Green

Release date: May 23, 2022 [eBook #68153]

Language: English

Original publication: United States: Dodd, Mead and Company, Inc,

1923

Credits: Charlene Taylor, Shaun Mudd and the Online Distributed

Proofreading Team at https://www.pgdp.net (This file was
produced from images generously made available by The
Internet Archive)

*** START OF THE PROJECT GUTENBERG EBOOK THE STEP

ON THE STAIR ***
A RUDE DRAWN DIAGRAM, LARGE ENOUGH TO BE SEEN
FROM ALL PARTS OF THE COURT ROOM, FELL INTO VIEW.
Page 146
 THE STEP ON
THE STAIR

BY
ANNA KATHARINE GREEN
AUTHOR OF

“THE LEAVENWORTH CASE,” “THE FILIGREE BALL,”

“THE MYSTERY OF THE HASTY ARROW,” ETC.

NEW YORK
DODD, MEAD AND COMPANY
1923

Copyright 1923
By DODD, MEAD AND COMPANY, Inc.

PRINTED IN THE U. S. A. BY
The Quinn & Boden Company
BOOK MANUFACTURERS
RAHWAY NEW JERSEY
 CONTENTS
Page
Book I The Three Edgars 3
Book II Hidden 93
Book III Which of Us Two? 191
Book IV Love 277
 BOOK I
THE THREE EDGARS

THE STEP ON THE STAIR

I
I had turned the corner at Thirty-fifth Street and was halfway down
the block in my search for a number I had just taken from the
telephone book when my attention was suddenly diverted by the
quick movements and peculiar aspect of a man whom I saw plunging
from the doorway of a large office-building some fifty feet or so
ahead of me.
Though to all appearance in a desperate hurry to take the taxi-cab
waiting for him at the curb, he was so under the influence of some
other anxiety almost equally pressing that he stopped before he
reached it to give one searching look down the street which, to my
amazement, presently centered on myself.
The man was a stranger to me, but evidently I was not so to him, for
his expression changed at once as our eyes met and, without waiting
for me to advance, he stepped hastily towards me, saying as we
came together:
“Mr. Bartholomew, is it not?”
I bowed. He had spoken my name.
“I have been waiting for you many interminable minutes,” he
hurriedly continued. “I have had bad news from home—a child hurt—
and must go at once. So, if you will pardon the informality, I will hand
over to you here and now the letter about which I telephoned you,
together with a key which I am assured you will find very useful. I am
sorry I cannot stop for further explanations; but you will pardon me, I
know. You can have nothing to ask which will not keep till to-
morrow?”
“No; but—”
I got no further, something in my tone or something in my look
seemed to alarm him for he took an immediate advantage of my
hesitation to repeat anxiously:
“You are Mr. Bartholomew, are you not? Edgar Quenton
Bartholomew?”
I smiled a polite acquiescence and, taking a card from my
pocketbook, handed it to him.
He gave it one glance and passed it back. The name corresponded
exactly with the one he had just uttered.
With a muttered apology and a hasty nod, he turned and fairly ran to
the waiting taxi-cab. Had he looked back—
But he did not, and I had the doubtful satisfaction of seeing him ride
off before I could summon my wits or pocket the articles which had
been so unceremoniously thrust upon me.
For what had seemed so right to him seemed anything but right to
me. I was Edgar Q. Bartholomew without question, but I was very
sure that I was not the Edgar Quenton Bartholomew he thought he
was addressing. This I had more than suspected when he first
accosted me. But when, after consulting my card, he handed me the
letter and its accompanying parcel, all doubt vanished. He had given
into my keeping articles meant for another man.
And I knew the man.
Yet I had let this stranger go without an attempt to rid him of his
misapprehension. Had seen him hasten away to his injured child
without uttering the one word which would have saved him from an
error the consequences of which no one, not even myself, could at
that moment foresee.
Why did I do this? I call myself a gentleman; moreover I believe
myself to be universally considered as such. Why, then—
Let events tell. Follow my next move and look for explanations later.
The man who had accosted me was a lawyer by the name of Miller.
Of that I felt assured. Also that he had been coming from his own
office when he first rushed into view. Of that office I should be glad to
have a momentary glimpse; also I should certainly be much more
composed in mind and ready to meet the possible results of my
inexcusable action if I knew whether or not the man for whom I had
been taken—the other Edgar Q. Bartholomew, would come for that
letter and parcel of which I had myself become the guilty possessor.
The first matter could be settled in no time. The directory just inside
the building from which I had seen Mr. Miller emerge would give me
the number of his office. But to determine just how I might satisfy
myself on the other point was not so easy. To take up my stand
somewhere in the vicinity—in a doorway, let us say—from which I
could watch all who entered the building in which I had located Mr.
Miller’s office seemed the natural and moreover the safest way. For
the passers-by were many and I could easily slip amongst them and
so disappear from view if by chance I perceived the other man of my
name approaching. Whereas, if once inside, I should find it difficult to
avoid him in case of an encounter.
Policy called for a watch from the street, but who listens to policy at
the age of twenty-three; and after a moment or two of indecision, I
hurried forward and, entering the building, was soon at a door on the
third floor bearing the name of
John E. Miller

attorney at law
Satisfied from the results of my short meeting with Mr. Miller in the
street below that he neither knew my person nor that of the other
Bartholomew (strange as this latter may seem when one considers
the character of the business linking them together), I felt that I had
no reason to fear being recognized by any of his clerks; and taking
the knob of the door in hand, I boldly sought to enter. But I found the
door locked, nor did I receive any response to my knock. Evidently
Mr. Miller kept no clerks or they had all left the building when he did.
Annoyed as I was at the mischance, for I had really hoped to come
upon some one there of sufficient responsibility to be of assistance
to me in my perplexity, I yet derived some gratification from the
thought that when the other Bartholomew came, he would meet with
the same disappointment.
But would he come? There seemed to be the best of reasons why he
should. The appointment made for him by Mr. Miller was one, which,
judging from what had just taken place between that gentleman and
myself, was of too great importance to be heedlessly ignored.
Perhaps in another moment—at the next stop of the elevator—I
should behold his gay and careless figure step into sight within
twenty feet of me. Did I wish him to find me standing in hesitation
before the lawyer’s closed door? No, anything but that, especially as
I was by no means sure what I might be led into doing if we thus
came eye to eye. The letter in my pocket—the key of whose
usefulness I had been assured—was it or was it not in me to hand
them over without a fuller knowledge of what I might lose in doing
so?
Honestly, I did not know. I should have to see his face—the far from
handsome face which nevertheless won all hearts as mine had
never done, good-looking though I was said to be even by those who
liked me least. If that face wore a smile—I had reason to dread that
smile—I might waver and succumb to its peculiar fascination. If on
the contrary its expression was dubious or betrayed an undue
anxiety, the temptation to leave him in ignorance of what I held would
be great and I should probably pass the coming night in secret
debate with my own conscience over the untoward situation in which
I found myself, himself and one other thus unexpectedly involved.
It would be no more than just, or so I blindly decided as I hastily
withdrew into a short hall which providentially opened just opposite
the spot where I stood lingering in my indecision.
It was an unnecessary precaution. Strangers and strangers only met
my eye as I gazed in anxious scrutiny at the various persons
hurrying by in every direction.
Five minutes—ten went by—and still a rush of strangers, none of
whom paused even for a moment at Mr. Miller’s door.
Should I waste any more time on such an uncertainty, or should I
linger a little while longer in the hope that the other Quenton
Bartholomew would yet turn up? I was not surprised at his being late.
If ever a man was a slave to his own temperament, that man was he,
and what would make most of us hasten, often caused him a
needless delay.
I would wait ten, fifteen minutes longer; for petty as the wish may
seem to you who as yet have been given no clew to my motives or
my reason for them, I felt that it would be a solace for many a bitter
hour in the past if I might be the secret witness of this man’s
disappointment at having through some freak or a culpable
indifference as to time, missed the interview which might mean
everything to him.
I should not have to use my eyes to take all this in; hearing would be
sufficient. But then if he should chance to turn and glance my way he
would not need to see my face in order to recognize me; and the
ensuing conversation would not be without its embarrassments for
the one hiding the other’s booty in his breast.
No, I would go, notwithstanding the uncertainty it would leave in my
mind; and impetuously wheeling about, I was on the point of carrying
out this purpose when I noticed for the first time that there was an
opening at the extreme end of this short hall, leading to a staircase
running down to the one beneath.
This offered me an advantage of which I was not slow to avail
myself. Slipping from the open hall on to the platform heading this
staircase, I listened without further fear of being seen for any
movement which might take place at door 322.
But without results. Though I remained where I was for a full half
hour, I heard nothing which betrayed the near-by presence of the
man for whom I waited. If a step seemed to halt before the office-
door upon which my attention was centered it went speedily on. He
whom I half hoped, half dreaded to see failed to appear.
Why should I have expected anything different? Was he not always
himself and no other? He keep an appointment?—remember that
time is money to most men if not to his own easy self? Hardly, if
some present whim, or promising diversion stood in the way. Yet
business of this nature, involving—But there! what did it involve?
That I did not know—could not know till what lay concealed in my
pocket should open up its secrets. My heart jumped at the thought. I
was not indifferent if he was. If I left the building now, the letter
containing these secrets would have to go with me. The idea of
leaving it in the hands of a third party, be he who he may, was an
intolerable one. For this night at least, it must remain in my keeping.
Perhaps on the morrow I should see my way to some other
disposition of the same. At all events, such an opportunity to end a
great perplexity seldom comes to any man. I should be a fool to let it
slip without a due balancing of the pros and cons incident to all
serious dilemmas.
So thinking, I left the building and in twenty minutes was closeted
with my problem in a room I had taken that morning at the Marie
Antoinette.
For hours I busied myself with it, in an effort to determine whether I
should open the letter bearing my name but which I was certain was
not intended for me, or to let it lie untampered with till I could
communicate with the man who had a legal right to it.
It was not the simple question that it seems. Read on, and I think you
will ultimately agree with me that I was right in giving the matter
some thought before yielding to the instinctive impulse of an honest
man.

II
My uncle, Edgar Quenton Bartholomew, was a man in a thousand. In
everything he was remarkable. Physically little short of a giant, but
handsome as few are handsome, he had a mind and heart
measuring up to his other advantages.
Had fortune placed him differently—had he lived where talent is
recognized and a man’s faculties are given full play—he might have
been numbered among the country’s greatest instead of being the
boast of a small town which only half appreciated the personality it
so ignorantly exalted. His early life, even his middle age I leave to
your imagination. It is of his latter days I would speak; days full of a
quiet tragedy for which the hitherto even tenor of his life had poorly
prepared him.
Though I was one of the only two male relatives left to him, I had
grown to manhood before Fate brought us face to face and his
troubles as well as mine began. I was the son of his next younger
brother and had been brought up abroad where my father had
married. I was given my uncle’s name but this led to little beyond an
acknowledgment of our relationship in the shape of a generous gift
each year on my birthday, until by the death of my mother who had
outlived my father twenty years, I was left free to follow my natural
spirit of adventure and to make the acquaintance of one whom I had
been brought up to consider as a man of unbounded wealth and
decided consequence.
That in doing this I was to quit a safe and quiet life, and enter upon
personal hazard and many a disturbing problem, I little realized. But
had it been given me to foresee this I probably would have taken
passage just the same and perhaps with even more youthful gusto.
Have I not said that my temperament was naturally adventurous?
I arrived in New York, had my three weeks of pleasure in town, then
started north for the small city from which my uncle’s letters had
invariably been post-marked. I had not advised him of my coming.
With the unconscious egotism of youth I wanted to surprise him and
his lovely young daughter about whom I had had many a dream.
Edgar Quenton Bartholomew sending up his card to Edgar Quenton
Bartholomew tickled my fancy. I had forgotten or rather ignored the
fact that there was still another of our name, the son of a yet younger
brother whom I had not seen and of whom I had heard so little that
he was really a negligible factor in the plans I had laid out for myself.
This third Edgar was still a negligible factor when on reaching C——
I stepped from the train and made my way into the station where I
proposed to get some information as to the location of my uncle’s
home. It was while thus engaged that I was startled and almost
thrown off my balance by seeing in the hand of a liveried chauffeur
awaiting his turn at the ticket office, a large gripsack bearing the
initials E. Q. B.—which you will remember were not only mine but
those of my unknown cousin.
There was but one conclusion to be drawn from this circumstance.
My uncle’s second namesake—the nephew who possibly lived with
him—was on the point of leaving town; and whether I welcomed the
fact or not, must at that very moment be somewhere in the crowd
surrounding me or on the platform outside.
More startled than gratified by this discovery, I impulsively reversed
the bag I was carrying so as to effectively conceal from view the
initials which gave away my own identity.
Why? Most any other man in my position would have rejoiced at
such an opportunity to make himself known to one so closely allied
to himself before the fast coming train had carried him away. But I
had my own conception of how and where my introduction to my
American relatives should take place. It had been my dream for
weeks, and I was in no mood to see it changed simply because my
uncle’s second namesake chose to take a journey just as I was
entering the town. He was young and I was young; we could both
afford to wait. It was not about his image that my fancies lingered.
Here the crowd of outgoing passengers caught me up and I was
soon on the outside platform looking about, though with a feeling of
inner revulsion of which I should have been ashamed and was not,
for the face and figure of a young man answering to my
preconceived idea of what my famous uncle’s nephew should be.
But I saw no one near or far with whom I could associate in any way
the initials I have mentioned, and relieved in mind that the hurrying
minutes left me no time for further effort in this direction, I was
searching for some one to whom I might properly address my
inquiries, when I heard a deep voice from somewhere over my head
remark to the chauffeur whom I now saw standing directly in front of
me, “Is everything all right? Train on time?” and turned, realizing in
an instant upon whom my gaze would fall. Tones so deliberate and
so rich with the mellowness of years never could have come from a
young man’s throat. It was my uncle, and not my cousin, who stood
at my back awaiting the coming train. One glance at his face and
figure made any other conclusion impossible.
Here then, in the hurry of departure from town where I had foolishly
looked upon him as a fixture, our meeting was to come off. The
surprise I had planned had turned into an embarrassment for myself.
Instead of a fit setting such as I had often imagined (how the dream
came back to me at that incongruous moment! The grand old parlor,
of the elegance of which strange stories had come to my ears—my
waiting figure, expectant, with eyes on the door opening to admit
uncle and cousin, he stately but kind, she curious but shy)—instead
of all this, with its glamour of hope and uncertainty, a station
platform, with but three minutes in which to state my claim and
receive his welcome.
Could any circumstances have been more prejudicial to my high
hopes? Yet must I make my attempt. If I let this opportunity slip, I
might never have another. Who knows! He might be going away for
weeks, perhaps for months. Danger lurks in long delays. I dared not
remain silent.
Meantime, I had been taking in his imposing personality. Though
anticipating much, I found myself in no wise disappointed. He was all
and more than my fancy had painted. If the grandeur of his
proportions aroused a feeling of awe, the geniality of his expression
softened that feeling into one of a more pleasing nature. He was
gifted with the power to win as well as to command; and as I noted
this and yielded to an influence such as never before had entered
my life, the hardihood with which I had contemplated this meeting
received a shock; and a warmth to which my breast was more or
less a stranger took the place of the pretense with which I had
expected to carry off a situation I was hardly experienced enough in
social amenities to handle with suitable propriety.
While this new and unusual feeling lightened my heart and made it
easy for my lips to smile, I touched him lightly on the arm (for he was
not noticing me at all), and quietly spoke his name.
Now I am by no means a short man, but at the sound of my voice he
looked down and meeting the glance of a stranger, nodded and
waited for me to speak, which I did with the least circumlocution
possible.
Begging him to pardon me for intruding myself upon him at such a
moment, I smilingly remarked:
“From the initials I see on the bag in the hand of your chauffeur, I
judge that you will not be devoid of all interest in mine, if only
because they are so strangely familiar to you.” And with a repetition
of my smile which sprang quite unbidden at his look of quick
astonishment, I turned my own bag about and let him see the E. Q.
B. hitherto hidden from view.
He gave a start, and laying his hand on my shoulder, gazed at me for
a moment with an earnestness I would have found it hard to meet
five minutes before, and then drew me slightly aside with the remark:
“You are James’ son?”
I nodded.
“You have crossed the ocean and found your way here to see me?”
I nodded again; words did not come with their usual alacrity.
“I do not see your father in your face.”
“No, I favor my mother.”
“She must have been a handsome woman.”
I flushed, not with displeasure, but because I had hoped that he
would find something of himself or at least of his family in my
personal traits.
“She was the belle of her village, when my father married her,” I
nevertheless answered. “She died six weeks ago. That is why I am
here; to make your acquaintance and that of my two cousins who up
till now have been little more than names to me.”
“I am glad to see you,”—and though the rumble of the approaching
train was every moment becoming more audible, he made no move,
unless the gesture with which he summoned his chauffeur could be
called one. “I was going to Albany, but that city won’t run away, while
I am not so sure that you will not, if I left you thus unceremoniously
at the first moment of our acquaintance. Bliss, take us back home
and tell Wealthy to order the fatted calf.” Then, with a merry glance
my way, “We shall have to do our celebrating in peaceful
contemplation of each other’s enjoyment. Both Edgar and Orpha are
away. But do not be concerned. A man of my build can do wonders
in an emergency; and so, I have no doubt, can you. Together, we
should be able to make the occasion a memorable one.”
The laugh with which I replied was gay with hope. No premonition of
mischief or of any deeper evil disturbed that first exhilaration. We
were like boys. He sixty-seven and I twenty-three.
It is an hour I love to look back upon.

III
I had always been told that my uncle’s home was one of unusual
magnificence but placed in such an undesirable quarter of the city as
to occasion surprise that so much money should have been lavished
in embellishing a site which in itself was comparatively worthless.
And yet while I was thus in a measure prepared for what I was to
see, I found the magnificence of the house as well as the
unattractiveness of the surroundings much greater than anything my
imagination had presumed to picture.
The fact that this man of many millions lived not only in the business
section but in the least prosperous portion of it was what I noted first.
I could hardly believe that the street we entered was his street until I
saw that its name was the one to which our letters had been