Ebook Direct and Inverse Scattering For The Matrix Schrodinger Equation 1St Edition Tuncay Aktosun Ricardo Weder Online PDF All Chapter

Direct and Inverse Scattering for the
Matrix Schrödinger Equation 1st

Edition Tuncay Aktosun Ricardo Weder
Visit to download the full and correct content document:
https://ebookmeta.com/product/direct-and-inverse-scattering-for-the-matrix-schroding
er-equation-1st-edition-tuncay-aktosun-ricardo-weder/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Chess Explained The c3 Sicilian 1st Edition Sam Collins
https://ebookmeta.com/product/chess-explained-
the-c3-sicilian-1st-edition-sam-collins/
Starting Out The c3 Sicilian 1st Edition John Emms
https://ebookmeta.com/product/starting-out-the-c3-sicilian-1st-
edition-john-emms/
The Paper Issue 83 1st Edition Origamiusa
https://ebookmeta.com/product/the-paper-issue-83-1st-edition-
origamiusa/
Structural Equation Modeling for Health and Medicine

1st Edition Douglas Gunzler
https://ebookmeta.com/product/structural-equation-modeling-for-
health-and-medicine-1st-edition-douglas-gunzler/
Direct Selling For Dummies Belinda Ellsworth
https://ebookmeta.com/product/direct-selling-for-dummies-belinda-
ellsworth/
Dilations Linear Matrix Inequalities the Matrix Cube

Problem and Beta Distributions 1st Edition J. William
Helton
https://ebookmeta.com/product/dilations-linear-matrix-
inequalities-the-matrix-cube-problem-and-beta-distributions-1st-
edition-j-william-helton/
Direct Selling For Dummies, 2nd Edition Belinda

Ellsworth
https://ebookmeta.com/product/direct-selling-for-dummies-2nd-
edition-belinda-ellsworth/
The American Revolution 1774 83 2nd Edition Daniel

Marston
https://ebookmeta.com/product/the-american-
revolution-1774-83-2nd-edition-daniel-marston/
Random Matrix Methods for Machine Learning 1st Edition

Romain Couillet
https://ebookmeta.com/product/random-matrix-methods-for-machine-
learning-1st-edition-romain-couillet/
Applied Mathematical Sciences
Tuncay Aktosun
Ricardo Weder
Direct and Inverse

Scattering
for the Matrix
Schrödinger
Equation
Volume 203
Series Editors
Anthony Bloch, Department of Mathematics, University of Michigan, Ann Arbor,
MI, USA
abloch@umich.edu
C. L. Epstein, Department of Mathematics, University of Pennsylvania,
Philadelphia, PA, USA
cle@math.upenn.edu
Alain Goriely, Department of Mathematics, University of Oxford, Oxford, UK
goriely@maths.ox.ac.uk
Leslie Greengard, New York University, New York, NY, USA
Greengard@cims.nyu.edu
Advisory Editors
J. Bell, Center for Computational Sciences and Engineering, Lawrence Berkeley
National Laboratory, Berkeley, CA, USA
P. Constantin, Department of Mathematics, Princeton University, Princeton, NJ,
USA
R. Durrett, Department of Mathematics, Duke University, Durham, CA, USA
R. Kohn, Courant Institute of Mathematical Sciences, New York University,
New York, NY, USA
R. Pego, Department of Mathematical Sciences, Carnegie Mellon University,
Pittsburgh, PA, USA
L. Ryzhik, Department of Mathematics, Stanford University, Stanford, CA, USA
A. Singer, Department of Mathematics, Princeton University, Princeton, NJ, USA
A. Stevens, Department of Applied Mathematics, University of Münster, Münster,
Germany
S. Wright, Computer Sciences Department, University of Wisconsin, Madison, WI,
USA
Founding Editors
F. John, New York University, New York, NY, USA
J. P. LaSalle, Brown University, Providence, RI, USA
L. Sirovich, Brown University, Providence, RI, USA
The mathematization of all sciences, the fading of traditional scientific boundaries,
the impact of computer technology, the growing importance of computer modeling
and the necessity of scientific planning all create the need both in education and
research for books that are introductory to and abreast of these developments. The
purpose of this series is to provide such books, suitable for the user of mathematics,
the mathematician interested in applications, and the student scientist. In particular,
this series will provide an outlet for topics of immediate interest because of the nov-
elty of its treatment of an application or of mathematics being applied or lying close
to applications. These books should be accessible to readers versed in mathematics
or science and engineering, and will feature a lively tutorial style, a focus on topics
of current interest, and present clear exposition of broad appeal. A compliment to
the Applied Mathematical Sciences series is the Texts in Applied Mathematics se-
ries, which publishes textbooks suitable for advanced undergraduate and beginning
graduate courses.
More information about this series at http://www.springer.com/series/34

Tuncay Aktosun • Ricardo Weder
Direct and Inverse Scattering

for the Matrix Schrödinger
Equation
Tuncay Aktosun Ricardo Weder
Department of Mathematics Instituto de Investigaciones en
University of Texas at Arlington Matemáticas Aplicadas y en Sistemas
Arlington, TX, USA Universidad Nacional Autónoma de México
Mexico City, Mexico
ISSN 0066-5452 ISSN 2196-968X (electronic)

ISBN 978-3-030-38430-2 ISBN 978-3-030-38431-9 (eBook)
https://doi.org/10.1007/978-3-030-38431-9
Mathematics Subject Classification: 34A55, 34L25, 34L40, 47A40, 81Uxx, 81U40
© Springer Nature Switzerland AG 2021

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, expressed 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
Ricardo Weder dedicates this monograph to
Teresa, Natalie Danitza, Ricardo Eugenio,
Robert, Fernanda, Anya, Mariel, and Maxine.
Preface
In this monograph, we present a complete and self-contained presentation of the

direct and inverse scattering theories for the matrix Schrödinger equation on the
half line with the general self-adjoint boundary condition. The results contained
here also apply to star graphs and to the matrix Schrödinger equation on the full
line.
The monograph, in particular, brings a comprehensive solution to the problem
of the characterization of the scattering data. This is done by defining the Faddeev
class of input data sets consisting of a potential and the boundary condition and by
defining the Marchenko class of scattering data sets consisting of a scattering matrix
and the bound-state information. Then, the direct scattering problem is viewed as a
mapping from the Faddeev class into the Marchenko class, and the inverse scattering
problem is viewed as a mapping from the Marchenko class into the Faddeev class.
The corresponding existence, uniqueness, and construction arising both in the di-
rect problem and in the inverse problem are established. The characterization of the
scattering data is provided by showing that each input data set in the Faddeev class
corresponds to a unique scattering data set in the Marchenko class and vice versa.
Furthermore, it is shown that the scattering data set used to construct the input data
set agrees with the scattering data set constructed from that input data set. We re-
mark that the boundary condition in an input data set is unique even though the two
boundary matrices describing that unique boundary condition are unique only up to
a postmultiplication by an invertible matrix.
The characterization of the scattering data in the monograph is provided in vari-
ous different but equivalent forms, and hence it is shown that the Marchenko class
can also be described in various different but equivalent forms. Thus, various char-
acterization theorems are presented, and some of these results have advantages over
others depending on the approach and solution technique used to analyze the rele-
vant direct and inverse problems. For example, in some cases it may be more advan-
tageous to solve an integral equation rather than a corresponding Riemann–Hilbert
problem or vice versa. It may be more convenient to look for a solution to an in-
vii
viii Preface
tegral equation in the class of bounded and integrable functions rather than in the
class of bounded and square-integrable functions. It may be more appropriate to
look for a solution to a homogenous Riemann–Hilbert problem in a more restricted
class rather than in a Hardy space. By having different but equivalent forms of the
characterization of the scattering data, we also understand better how different ap-
proaches and solution techniques are related to each other in solving the direct and
inverse problems.
The theory presented is supplemented by various explicitly solved examples.
Some of these examples illustrate the independence of the characterization condi-
tions by showing the failure of one characterization condition despite the satisfaction
of the remaining conditions. Some examples illustrate how Levinson’s theorem is
applied in one form of the characterization. Some other examples illustrate how the
boundary condition is reconstructed with the help of the high-energy analysis of the
scattering matrix or by using the unitarity of the generalized Fourier map.
We hope that our monograph will benefit a wide variety of readers. For exam-
ple, scientists and engineers, who are interested in using and applying the results
without worrying about the mathematical details or proofs, can benefit by reading
the summaries of the direct and inverse problems and seeing how various relevant
quantities are constructed, by viewing various characterization theorems provided,
and by looking at some of the explicitly solved examples. This can be done by
checking Sects. 2.6, 2.7, 2.8, 3.1, 5.2, and Chap. 6. Beginners in mathematics and
mathematical physics who are interested in the scattering and spectral theories of
the direct and inverse problems can try to read the entire monograph, at times by
consulting the mathematical notation and preliminaries provided in Appendix A.
Advanced researchers can at any order look at the new results presented, techniques
used, connections made among different techniques, and proofs provided.
One of the goals in our monograph is to generalize the characterization result pro-
vided by Agranovich and Marchenko [11] from the Dirichlet boundary condition to
the general boundary condition. Motivated by quantum physics, the scattering ma-
trix is traditionally normalized to one at high energies in the scalar case and to the
identity matrix in the matrix case. The same normalization in the non-Dirichlet case
becomes problematic and prevents formulating a characterization. Our monograph
clarifies this issue so that the scattering and inverse scattering problems can be ana-
lyzed properly regardless of the boundary condition used.
Arlington, TX, USA Tuncay Aktosun

Mexico City, Mexico Ricardo Weder
November 2019
Acknowledgments
The content of this monograph is based on research supported in part by CONACYT

under project CB2015, 254062, Project PAPIIT-DGAPA-UNAM IN103918, and by
Coordinación de la Investigación Cientı́fica, UNAM.
ix
Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 An Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Notes on the Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2 The Matrix Schrödinger Equation and the Characterization

of the Scattering Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.1 Outline of the Chapter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2 The Matrix Schrödinger Equation on the Half Line . . . . . . . . . . . . . 22
2.3 Star Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.4 The Schrödinger Equation on the Full Line . . . . . . . . . . . . . . . . . . . . 28
2.5 The Faddeev Class and the Marchenko Class . . . . . . . . . . . . . . . . . . 31
2.6 A First Characterization of the Scattering Data . . . . . . . . . . . . . . . . 38
2.7 Alternate Characterizations of the Scattering Data . . . . . . . . . . . . . . 40
2.8 Another Characterization of the Scattering Data . . . . . . . . . . . . . . . 45
3 Direct Scattering I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.1 Outline of the Solution to the Direct Problem . . . . . . . . . . . . . . . . . . 49
3.2 Special Solutions to the Schrödinger Equation . . . . . . . . . . . . . . . . . 51
3.3 The Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
3.4 Equivalence of the Formulations of the Boundary Condition . . . . . 94
3.5 The Quadratic Form of the Hamiltonian . . . . . . . . . . . . . . . . . . . . . . 103
3.6 Transformations of the Jost and Scattering Matrices . . . . . . . . . . . . 109
3.7 The Jost and Scattering Matrices with Zero Potential . . . . . . . . . . . 112
3.8 Low-Energy Analysis with Potentials in the Faddeev Class . . . . . . 114
3.9 Low-Energy Analysis with Potentials of Finite
Second Moment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
3.10 High-Energy Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
3.11 Bound States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
3.12 Levinson’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
3.13 Further Properties of the Scattering Data . . . . . . . . . . . . . . . . . . . . . . 226
3.14 The Marchenko Integral Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
xi
xii Contents
3.15 The Boundary Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237

3.16 The Existence and Uniqueness in the Direct Problem . . . . . . . . . . . 240
4 Direct Scattering II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261

4.1 Basic Principles of the Scattering Theory . . . . . . . . . . . . . . . . . . . . . 261
4.2 The Limiting Absorption Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
4.3 The Generalized Fourier Maps for the Absolutely Continuous
Subspace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
4.4 The Wave Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
4.5 The Scattering Operator and the Scattering Matrix . . . . . . . . . . . . . 298
4.6 The Spectral Shift Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
4.7 Trace Formulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
4.8 The Number of Bound States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322
5 Inverse Scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339

5.1 Nonuniqueness Due to the Improperly Defined Scattering
Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
5.2 The Solution to the Inverse Problem . . . . . . . . . . . . . . . . . . . . . . . . . 346
5.3 Bounds on the Constructed Solutions . . . . . . . . . . . . . . . . . . . . . . . . 380
5.4 Relations Among the Characterization Conditions . . . . . . . . . . . . . . 390
5.5 The Proof of the First Characterization Theorem . . . . . . . . . . . . . . . 400
5.6 Equivalents for Some Characterization Conditions . . . . . . . . . . . . . 403
5.7 Inverse Problem Using Only the Scattering Matrix as Input . . . . . . 413
5.8 Characterization via Levinson’s Theorem . . . . . . . . . . . . . . . . . . . . . 417
5.9 Parseval’s Equality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 445
5.10 The Generalized Fourier Map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 448
5.11 An Alternate Method to Solve the Inverse Problem . . . . . . . . . . . . . 459
5.12 Characterization with Potentials of Stronger Decay . . . . . . . . . . . . . 475
5.13 The Dirichlet Boundary Condition . . . . . . . . . . . . . . . . . . . . . . . . . . . 478
6 Some Explicit Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485

6.1 Illustration of the Theory with Explicit Examples . . . . . . . . . . . . . . 485
6.2 Some Methods Yielding Explicit Examples . . . . . . . . . . . . . . . . . . . 486
6.3 Explicit Examples in the Characterization of the Scattering Data . . 496
6.4 Explicit Examples of Particular Solutions . . . . . . . . . . . . . . . . . . . . . 536
Appendix A Mathematical Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545

A.1 Vectors, Matrices, and Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545
A.2 Banach and Hilbert Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
A.3 Inequalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 553
A.4 Mollifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 555
A.5 Equicontinuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 556
A.6 Distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 556
A.7 Absolute Continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 559
A.8 Sobolev Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 561
A.9 The Fourier Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 562
Contents xiii
A.10 Hardy Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564

A.11 Other Banach Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 566
A.12 Linear Operators Between Banach and Hilbert Spaces . . . . . . . . . . 567
A.13 Operators Between Finite Dimensional Hilbert Spaces . . . . . . . . . . 572
A.14 Self-adjoint Operators and Symmetric Quadratic Forms . . . . . . . . . 573
A.15 Trace-Class and Hilbert–Schmidt Operators . . . . . . . . . . . . . . . . . . . 576
A.16 Resolvent and Spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 580
A.17 The Spectral Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 580
A.18 The Spectral Shift Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 587
A.19 Deficiency Indices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 589
A.20 Self-adjoint Extensions of Matrix Schrödinger Operators . . . . . . . . 590
A.21 Hermitian Symplectic Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . 592
A.22 Integral Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596
A.23 The Gronwall Inequalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 602
A.24 Miscellaneous Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 604
A.25 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 607
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 619
List of Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 623

Chapter 1
Introduction
Abstract In this introductory chapter the goals of the monograph are described, the
contents of the remaining chapters and Appendix A are outlined, and the relevant
general references in the literature are mentioned. The direct and inverse scatter-
ing problems for the matrix Schrödinger equation on the half line with the general
self-adjoint boundary condition are viewed as two mappings, and it is indicated how
these two mappings become inverses of each other by specifying their domains ap-
propriately. The traditional definition of the scattering matrix in one way with the
Dirichlet boundary condition and in a different way with a non-Dirichlet boundary
condition is criticized, and it is indicated that such a practice makes it impossible
to formulate a well-posed inverse scattering problem unless the boundary condition
is already known and the Dirichlet and non-Dirichlet boundary conditions are not
mixed. It is emphasized that the recovery of the boundary condition should be a part
of the solution to the inverse scattering problem rather than a part of the inverse
scattering data.
1.1 An Overview
In this monograph we analyze the direct and inverse scattering problems for the
matrix Schrödinger equation on the half line with the general self-adjoint bound-
ary condition. In the direct scattering problem, given the potential and the boundary
condition we determine the scattering matrix and the bound-state information con-
sisting of the bound-state energies and the bound-state normalization matrices. In
the associated inverse problem, given the scattering matrix and the bound-state in-
formation, we determine the corresponding potential and the boundary condition.
We refer to the data set consisting of the potential and the boundary condition as the
input data set and denote it by D. We refer to the data set consisting of the scatter-
ing matrix and the bound-state information as the scattering data set and denote it
by S. Thus, we can mathematically view the direct scattering problem as the map
© Springer Nature Switzerland AG 2021 1

T. Aktosun, R. Weder, Direct and Inverse Scattering for the Matrix
Schrödinger Equation, Applied Mathematical Sciences 203,
https://doi.org/10.1007/978-3-030-38431-9 1
2 1 Introduction
D : D → S with the domain D[D] and view the inverse scattering problem as the
map D : S → D with the domain D[D].
There are four aspects in the direct and inverse problems. These are the exis-
tence, uniqueness, construction, and characterization. For the existence in the direct
problem, given D in a specified class D[D] we determine whether a corresponding S
exists in some specific class. The uniqueness is concerned with whether there exists
a unique S corresponding to a given D in the domain of D, or two or more distinct
sets S may correspond to the same D. The construction deals with the recovery of
S from D. Similarly, for the existence in the inverse problem, given S in a speci-
we determine whether a corresponding D exists in some specific
fied class D[D],
class. The uniqueness in the inverse problem is concerned with whether there exists
a unique D corresponding to a given S in the domain of D, or two or more distinct
sets D may correspond to the same S. The construction in the inverse problem deals
with the recovery of D from S. The characterization problem is the most delicate
one. It consists of finding a domain D[D] for the direct map D and a domain D[D]
for the inverse map D in such a way that we have

D D[D] = D[D], D]
DD[ = D[D],
i.e., the range of D[D] under D coincides with the domain of D and that the range of

D[D] under D coincides with the domain of D. If the characterization holds then we
have D = D −1 , i.e., the map D
coincides with the inverse of the map D. Informally,
we can describe the characterization problem as the identification of a class of input
data sets and a class of scattering data sets in such a way that there is a one-to-one
correspondence between those two classes.
Our goal in this monograph is to provide a complete solution to the direct and in-
verse scattering problems for the matrix Schrödinger equation on the half line with
the general self-adjoint boundary condition and also to present a complete character-
ization. We establish our characterization result in such a way that it certainly holds
in the scalar case, it holds for any self-adjoint boundary condition, it yields the con-
struction in the corresponding direct and inverse problems, and it reveals how the
individual conditions in the characterization affect the existence, uniqueness, and
construction. Through various comments and explicit examples, we clarify various
issues arising in our characterization so that our approach can be useful in establish-
ing characterizations for some other direct and inverse problems.
The only viable characterization in the literature for the matrix Schrödinger oper-
ator on the half line can be found in the seminal work by Agranovich and Marchenko
in [8, 9, 11]. However, the analysis in [8, 9, 11] is restricted to the Dirichlet boundary
condition, and hence our study can be viewed as a generalization of the characteri-
zation given in [8, 9, 11]. It is remarkable that a characterization in the scalar case
valid for a general self-adjoint boundary condition does not exist and cannot ex-
ist in the way a scattering matrix is defined traditionally. As indicated in Section 4
of [27], as a result of defining [30, 78, 186, 192] the scattering matrix in one way
with the Dirichlet boundary condition and in a different way with a non-Dirichlet
1.1 An Overview 3
boundary condition, it is impossible to have the uniqueness in the inverse problem

unless separate characterizations are developed in the Dirichlet case and in the non-
Dirichlet case, respectively. In the scalar case, it is known [27] that, in the absence of
bound states, an input data set consisting of a real-valued potential with the Dirichlet
boundary condition and another input data set consisting of a real-valued potential
with the Neumann boundary condition may correspond to the same scattering ma-
trix. For further details on this issue the reader is referred to Sect. 5.1 of our present
monograph and to Section 4 and Example 6.3 of [27]. In our own solution to the
characterization problem we do not encounter such a nonuniqueness issue because
we define the scattering matrix in a unique way, without defining it in one way in
the Dirichlet case and in another way in the non-Dirichlet case. Actually, we define
the scattering matrix in such a way that the associated Schrödinger operator for the
unperturbed problem has the Neumann boundary condition. This definition is moti-
vated by the theory of quantum graphs [138, 163, 164, 278], about which we have
included some discussions in Sects. 4.4 and 4.5. In the matrix case a boundary con-
dition could be partly Dirichlet and partly non-Dirichlet, and the Dirichlet boundary
condition in the matrix case is really a very special case and can be referred to as the
purely Dirichlet case [32]. For further details on the self-adjoint boundary condition,
we refer the reader to Proposition 3.4.5 and Sect. 3.6.
As already mentioned in the preface we analyze the existence, uniqueness, con-
struction, and characterization issues related to the relevant direct and inverse prob-
lems under the assumption that the input data set D belongs to the Faddeev class and
the scattering data set S belongs to the Marchenko class. The Faddeev class consists
of input data sets D as in (2.5.1), where the potential V (x) and the boundary ma-
trices A and B are as specified in Definition 2.5.1. The Marchenko class consists of
scattering data sets S as in (2.5.2), where the scattering matrix S and the bound-state
data {κ j , M j }Nj=1 are as specified in Definition 2.5.5.
In [8, 9, 11] Agranovich and Marchenko studied the inverse problem for (2.2.1)
in the special case when A = 0 and B = −I, with 0 being the n × n zero matrix
and I denoting the n × n identity matrix, and when the potential
V (x) appearing
in (2.2.1) is Lebesgue measurable, Hermitian, and satisfying 0∞ dx x |V (x)| < +∞
instead of (2.2.3). Thus, without assuming the integrability of the potential, in [8,
9, 11] Agranovich and Marchenko provided a characterization of the corresponding
scattering data. Our own work provides a generalization of the characterization of
[8, 9, 11] to the case with the general self-adjoint boundary condition. When a non-
Dirichlet boundary condition is used at x = 0, the integrability of the potential is
necessary, and that is why the integrability of V (x) in (2.2.3) is crucial. In particular,
to be able to define the regular solution ϕ (k, x) appearing in (3.1.2), it is necessary
that the potential is integrable at x = 0. For further details on this point we refer the
reader to Theorem 1.2.1 of [11] and also [281]. In Chap. 6 we illustrate this issue
explicitly in Example 6.2.3.
We also mention the relevant references by Harmer [137–139], where the direct
and inverse problems for (2.2.1) are formally studied with the general self-adjoint
boundary condition, not as in (2.2.4)–(2.2.6) but in an equivalent form. However,
the work in [137–139] lacks the large-k analysis beyond the leading term and also
4 1 Introduction
lacks the small-k analysis of the scattering data, both of which are essential for the
analysis of the relevant inverse problem. Thus, our study can also be considered as
a complement to Harmer’s work in [137–139]. In our monograph we use the results
from previous work presented in [11, 23, 24, 32, 276–278].
The theory of matrix Schrödinger equations has origins in the beginning of quan-
tum mechanics. It was realized that the matrix equations could be used to take
into account internal properties such as spin and to consider systems of particles
such as a collection of atoms, molecules, and nuclei. A well-known example is
the Pauli equation, which is the formulation of the Schrödinger equation for half-
spin particles. For further applications on this issue, we refer the reader to [8–
11, 78, 102, 172, 180, 201, 204]. In this context the matrix Schrödinger equation
on the half line appears after applying the technique of separation of variables on
a Schrödinger system with a spherically symmetric potential. Once the variables
are separated one obtains a matrix equation that only depends on the radial vari-
able. In standard quantum mechanical applications the Dirichlet boundary condition
naturally arises. This is because the Dirichlet boundary condition is related to the
finiteness of the wavefunction at the origin.
The study of quantum wires has physical relevance in the design of elementary
gates in quantum computing and in nanotubes for microscopic electronic devices.
A quantum graph is an idealization of quantum wires with very small cross sectional
areas, where such wires meet at the vertices. The quantum graph is obtained in the
zero limit of the cross sectional area of quantum wires. With the advent of quantum
graphs, the interest in matrix Schrödinger equations has been expanded drastically
and the consideration of general boundary conditions has become crucial. The ma-
trix Schrödinger equation on the half line with the general self-adjoint boundary
condition has now many applications in the scattering theory on quantum graphs
[59–61, 65, 79, 97, 98, 124, 135, 145, 161, 163, 164, 174–178, 183].
A particular phenomenon governed by a matrix Schrödinger equation is a star
graph, i.e., a one-vertex graph with a finite number of semi-infinite edges. In this
case, a homogeneous boundary condition linear in the wavefunction and its deriva-
tive is imposed at the vertex, and the dynamics on each edge is governed by the
Schrödinger equation. Physically, a star graph represents a finite number of very
thin quantum wires connected at the vertex. For example, a string of atoms may
form a star graph.
The consideration of the general self-adjoint boundary condition at a vertex
rather than only the Dirichlet boundary condition is relevant. For quantum graphs it
is crucial that the boundary condition at each vertex links the values of the wave-
function and its derivative arriving from different edges. An important case is the
Kirchhoff boundary condition, which amounts to the continuity of the wavefunction
at a vertex and also that at that vertex the sum of the derivatives of the wavefunctions
from all the edges is zero. The boundary condition at each vertex in the quantum
graph depends on how the limit is taken, and we refer the reader to Section 7.5 of
[60] for a detailed discussion of the related extensive literature. Thus, it is relevant
to consider all possible self-adjoint boundary conditions to take into account various
limiting procedures in quantum graphs.
1.1 An Overview 5
There are also other relevant applications of non-Dirichlet boundary conditions

in wave propagation problems. For example, consider the inverse problem of deter-
mining the shape of the human vocal tract during the utterance of a vowel, where the
sound pressure measurements at the lips are used as input. Such a problem [15, 25]
can be analyzed as an inverse problem for the Schrödinger equation on the half line
with a general self-adjoint boundary condition. The propagation of sound waves
through human lungs [150, 151] is a technique used in medical diagnostics, and
its analysis involves propagation of acoustic waves on an infinite tree. The inverse
scattering problems for Schrödinger operators over star-shaped graphs have also
applications to detect faults in electrical networks [244].
Our monograph is organized as follows. It consists of six chapters, an appendix,
a bibliography, a subject index, and a symbol index. The section titles listed in the
Contents provide some information about each chapter and Appendix A.
In Chap. 2 we introduce the matrix Schrödinger equation on the half line, the
n × n matrix potential V (x), and the general self-adjoint boundary condition de-
scribed by two constant n × n matrices A and B. In Sect. 2.3 we show that the scat-
tering and inverse scattering problems on a star graph can be analyzed by studying
a matrix Schrödinger equation on the half line with a diagonal-matrix potential. In
Sect. 2.4 we show that a 2n × 2n matrix Schrödinger equation on the half line with
the general self-adjoint boundary condition is unitarily equivalent to an n × n matrix
Schrödinger equation on the full line with a point interaction at x = 0. Thus, the
results on the direct and inverse scattering problems in our monograph can also be
applied on the corresponding problems on the full line. In Sect. 2.5 we introduce
the Faddeev class of input data sets D and the Marchenko class of scattering data
sets S. As indicated in (2.5.1) an input data set consists of the potential V (x) and
the two boundary matrices A and B, and the restrictions on D specified in Defini-
tion 2.5.1 determine when D belongs to the Faddeev class. As indicated in (2.5.2)
a scattering data set consists of the scattering matrix S(k) and the bound-state in-
formation containing the number of bound states N, the bound-state energies −κ 2j ,
and the bound-state normalization matrices M j . The description of the Marchenko
class is given in Definition 2.5.5. A main result in our monograph is the establish-
ment of a one-to-one correspondence between the Faddeev class of input data sets
and the Marchenko class of scattering data sets. In Sects. 2.6, 2.7, and 2.8 we list
a number of properties that can be satisfied by S. With the help of such properties,
various equivalent descriptions of the Marchenko class are possible. Such equiva-
lent descriptions allow us to present various different but equivalent results for the
characterization of the scattering data in the solution to the inverse problem. We
state ten characterization theorems, but their proofs are postponed until Chap. 5.
We note that the proofs of our characterization results with the general self-
adjoint boundary condition are considerably more involved than the case of the
Dirichlet boundary condition studied in [11]. The Dirichlet boundary condition in-
volves only the solution to the matrix Schrödinger equation, and consequently the
corresponding inverse problem only requires the analysis of the Marchenko equa-
tion (3.14.1). On the contrary, the general self-adjoint boundary condition involves
6 1 Introduction
both the solution to the matrix Schrödinger equation and the spatial derivative of
that solution. Consequently, the study of the corresponding inverse problem requires
the analysis of both (3.14.1) and the derivative Marchenko equation (3.14.6). The
Dirichlet boundary condition is singled out among all self-adjoint boundary con-
ditions as the only one such that the high-energy limit S∞ of the scattering matrix
appearing in (3.10.33) is equal to −I. The general self-adjoint boundary condition
is more complicated. As elaborated in Proposition 3.15.1 and Proposition 5.2.9, in
our characterization theorems in Sects. 2.6 and 2.7 we construct the boundary ma-
trices A and B by solving (3.15.2) with the help of the input from the first two terms
in the high-energy behavior of the scattering matrix. In the proof of our character-
ization theorem in Sect. 2.8 we use a different method exploiting the unitarity of
the generalized Fourier map. The latter method of characterization does not exist in
[11].
In Chap. 3 we present the solution to the direct scattering problem for the matrix
Schrödinger equation on the half line with the general self-adjoint boundary con-
dition. Using an input data set in the Faddeev class we establish the properties of
various quantities such as the Jost solution, the Jost matrix, the physical solution,
and the regular solution. Various equivalent formulations of the general self-adjoint
boundary condition are considered. We construct the self-adjoint realizations of the
corresponding matrix Schrödinger operator. We determine the low-energy and high-
energy behavior of the Jost matrix and the scattering matrix. We show how the gen-
eral self-adjoint boundary condition is related to the high-energy behavior of the
scattering matrix, and we provide an explicit construction of the boundary matrices
A and B from the scattering data set. We analyze the bound states, and we prove
Levinson’s theorem by showing how the phase of the determinant of the scattering
matrix is related to the total number of bound states including the multiplicities. We
also analyze the Marchenko integral equation and the derivative Marchenko integral
equation.
In Chap. 4 we continue our analysis of the direct scattering problem for the ma-
trix Schrödinger equation on the half line with the general self-adjoint boundary
condition. We describe the scattering process physically and mathematically, and
we provide the definition of the scattering operator in terms of the wave operators.
We introduce the limiting absorption principle, and we show how the Hamiltonian
and its resolvent are related to the potential and the two boundary matrices A and B.
We introduce the generalized Fourier maps associated with the absolutely continu-
ous spectrum and provide their basic properties. We express the wave operators and
the scattering operator in terms of the generalized Fourier maps and their adjoints.
We prove that the scattering matrix defined in (2.2.28) in terms of the Jost matrix
coincides with the scattering matrix derived from the scattering operator. We con-
sider various other topics such as the properties of the spectral shift function, trace
formulas of Buslaev–Faddeev type, and a Bargmann–Birman–Schwinger bound on
the number of bound states.
In Chap. 5 we provide the solution to the inverse scattering problem by construct-
ing the corresponding input data set in the Faddeev class from a scattering data set
in the Marchenko class. We discuss the nonuniqueness arising in the inverse scatter-
1.1 An Overview 7
ing problem if the scattering matrix is defined one way with the Dirichlet boundary
condition and in a different way with the non-Dirichlet boundary condition, as done
traditionally. We present the solution to the inverse scattering problem by analyzing
the solvability of the Marchenko integral equation and the derivative Marchenko
integral equation. We present a proof of the main characterization result presented
in Theorem 2.6.1. By establishing the equivalents among various characterization
properties, we prove the alternate characterization theorems stated in Sect. 2.7. We
study the inverse scattering problem when the input scattering data set lacks the
bound-state information. From the given scattering matrix we show how to construct
a scattering data set belonging to the Marchenko class so that the constructed scat-
tering data set can be used as input into a properly posed inverse scattering problem.
With the help of Levinson’s theorem, we prove the two equivalent characterization
results stated in Theorems 2.7.9 and 2.7.10. We provide a proof of Parseval’s equal-
ity, expressing the completeness of the set consisting of the physical solution and
the bound-state solutions. We present the generalized Fourier map and establish its
properties. Using the generalized Fourier map we prove the characterization result
stated in Theorem 2.8.1. We also consider the characterization of the scattering data
when the potentials are restricted to the class L1p (R+ ) for p > 1 instead of only p = 1
in the Faddeev class. Finally, we formulate the characterization of the scattering data
in the special case of the purely Dirichlet boundary condition.
In Chap. 6 we illustrate the theory presented earlier via explicitly solved exam-
ples. We show how the Marchenko integral equation can yield explicitly solved ex-
amples when its kernel contains a matrix exponential and hence becomes separable.
We demonstrate the necessity of the integrability of the potential when a general
self-adjoint boundary condition is used rather than only the Dirichlet boundary con-
dition. We illustrate the characterization of the scattering data by various examples
where all the characterization conditions are satisfied or one or more of the con-
ditions are not satisfied. The examples where only one characterization condition
fails indicate the independence of the characterization conditions applied. In Exam-
ples 6.3.23, 6.3.24, 6.3.25, and 6.3.27 we consider potentials outside the Faddeev
class because they decay too slowly at infinity. These examples show that outside
the Faddeev class the uniqueness aspect of the inverse scattering problem fails when
we use as input a scattering data set in the Marchenko class. This proves that the
Faddeev class is optimal for the inverse scattering problem with the scattering data
sets in the Marchenko class. In Chap. 6 we also provide some examples to illus-
trate how a solution to the zero-energy Schrödinger equation is affected by various
restrictions on the scattering data. We illustrate the solution to the inverse scatter-
ing problem with various explicit examples, and we demonstrate how the potential,
boundary condition, and other relevant quantities are constructed from a given scat-
tering data set. We also illustrate the use of Levinson’s theorem and the generalized
Fourier map through some explicit examples.
Appendix A provides the mathematical preliminaries and it includes some ba-
sic definitions, notations, and results used in our monograph. The subject index and
the symbol index may be useful to determine the exact location of each prelim-
8 1 Introduction
inary result. The preliminaries include a brief description of Banach and Hilbert
spaces, Sobolev spaces, Hardy spaces, and some other functions spaces used in the
monograph, as well as distributions, the Fourier transform, basic inequalities, lin-
ear integral equations, and the Gronwall inequalities. A brief review is included
for particular linear operators such as those that are self-adjoint, compact, Hilbert–
Schmidt, unitary, or trace class. Also reviewed are various concepts from functional
analysis such as the resolvent, spectrum, spectral shift function, spectral theorem,
self-adjoint extensions, absolute continuity, quadratic forms, polar decomposition,
and Hermitian symplectic geometry. A glimpse of Appendix A reveals how the di-
rect and inverse scattering problems for the matrix Schrödinger equation on the half
line with the general self-adjoint boundary condition are related to a host of areas in
functional analysis and theory of ordinary differential equations.
1.2 Notes on the Bibliography
The bibliography on the topics studied in this monograph and in related areas is vast
and we have no claim for completeness. Our aim is to provide the readers with a
general perspective of the literature so that they can see how the results presented in
this monograph are related to the direct and inverse scattering and spectral theories
and also appreciate how broad the subject is. We remark that the choice of references
listed is also influenced by our personal scientific interests.
We refer the reader to the excellent historical review by Roger Newton in the
foreword to [78]. The first attempts to obtain information on potentials from the
continuous spectrum, i.e., from the scattering data, appeared in the physics literature
in late 1930s. The scattering matrix was introduced by Wheeler [283] and Heisen-
berg [140–142]. The introduction of the scattering matrix gave the quantum me-
chanical inverse scattering problem a great impetus and importance, and the inverse
problem was started to be viewed as obtaining information about the interactions
from the scattering matrix. In [140–142] Heisenberg conjectured that all relevant
physical information was encoded in the scattering matrix, and he proposed to for-
mulate all physical theories in terms of the scattering matrix rather than, for exam-
ple, the Hamiltonian. Heisenberg’s proposal was intensively studied and it came to
be known as the S-matrix theory in high-energy physics. Heisenberg also discussed
the inverse scattering problem and made the conjecture that the uniqueness holds,
i.e., that the scattering matrix uniquely determines the potential and consequently
the scattering matrix uniquely determines the bound states. He also discussed the
construction of the potential from the scattering matrix and the characterization of
the scattering data.
The Jost solution, which plays a central role in scattering on the half line and on
the line, was introduced by Res Jost [155]. Fröberg [114] investigated the inverse
problem by starting from the radial Schrödinger equation with a central potential
and tried to construct the potential from the knowledge of one scattering phase shift
of a given angular momentum, without using the JWKB approximation. However,
1.2 Notes on the Bibliography 9
Bargmann [47, 48] proved that, in the presence of bound states, the potential in the
radial Schrödinger equation cannot uniquely be determined by the phase shift of one
angular momentum known at all positive energies. Levinson [184, 185] proved that
if there are no bound states then the uniqueness holds, and he also proved what is
now known as Levinson’s theorem, which makes the connection between the bound
states and the phase shift of a given angular momentum. Marchenko [189, 192]
considered the uniqueness question studied by Levinson and used the method of
transformation operators to investigate the relevant inverse problems.
The method of transformation operators continues to play an important role in the
inverse scattering and spectral theories. Gel’fand and Levitan [121, 122, 186, 192]
analyzed the inverse problem by using the spectral function as the input data. They
obtained a characterization of the spectral function for the Schrödinger equation
on a finite interval and on the half line by using the method of transformation op-
erators. This approach came to be known as the Gel’fand–Levitan method. Mark
Krein [165–170] used a different but related approach to study the inverse prob-
lem. Marchenko [186, 190, 192] solved the inverse scattering problem for the
Schrödinger equation on the half line via the method of transformation operators,
and he obtained a characterization of the scattering data. This approach to the solu-
tion of the inverse scattering problem came to be known as the Marchenko method.
We refer the reader to [156] for a related result and Krein’s work [171] for a different
approach. An important outcome of these general methods to solve the inverse prob-
lem is the construction of families of potentials for which the Schrödinger equation
with radially symmetric potentials has explicit solutions at a given angular momen-
tum. These are now known as Bargmann potentials [47, 48] and they correspond to
a scattering matrix S(k) which is a rational function of k.
The Gel’fand–Levitan method and the Marchenko method were soon general-
ized to the matrix cases, i.e., to analyze systems of differential equations. Newton
and Jost [204] extended the Gel’fand–Levitan method to the matrix Schrödinger
equation with the Dirichlet boundary condition, and we also mention Krein’s related
work [172]. Newton extended [201] this method to study particles with spin. The
Marchenko method was generalized to the matrix Schrödinger equation with the
Dirichlet boundary condition by Agranovich and Marchenko in [8, 9] and to parti-
cles with spin in [10]. For a detailed presentation of the results of Agranovich and
Marchenko, we refer the reader also to [11, 102]. For further results in the inverse
scattering theory for systems of differential equations on the half line, we refer the
reader to Section 3 of Chapter IX of the monograph by Chadan and Sabatier [78].
A characterization of the spectral function for the Dirac system with the Dirichlet
boundary condition was given by Gasymov and Levitan in [120], and a characteriza-
tion of the scattering data for the Dirac system with the Dirichlet boundary condition
was given by Gasymov in [117] and by Gasymov and Levitan in [119]. For the Dirac
system with a radially symmetric potential Blancarte, Grébert, and Weder obtained
[64] high-energy and low-energy estimates and proved the Parzen theorem for all
angular momenta.
Using the Marchenko method Faddeev proposed [103] a characterization of the
scattering data for the Schrödinger equation on the full line for real-valued poten-
10 1 Introduction
tials in the L11 (R) class, i.e., potentials that are integrable and that have a finite first
moment. This class of potentials is now known as the Faddeev class. However, the
validity of Faddeev’s characterization was questioned first by Chadan and Sabatier
in the 1977 edition of their monograph [77] and then by Deift and Trubowitz [83].
As a remedy, Deift and Trubowitz [83] provided a modification of Faddeev’s char-
acterization for potentials in the smaller class of L21 (R), which consists of potentials
that are integrable and that have a finite second moment. In [83] Deift and Trubowitz
raised the uncertainty issue on the continuity of the scattering matrix at zero energy
under the L11 (R) condition. This turned out to be a rather delicate issue. We remark
that Marchenko’s important monograph [192] was originally published in 1977 in
Russian, but its English translation was first published in 1986. It turned out that in
[192] Marchenko had already given a characterization in the L11 (R) class, by care-
fully avoiding making an assumption about the continuity of the scattering matrix
at zero energy. Another characterization in the L11 (R) class avoiding the continuity
of the scattering matrix at zero energy was presented by Melin [196]. Finally, Klaus
in [162] and Aktosun and Klaus in [19] settled the issue by proving the continuity
of the scattering matrix at zero energy for potentials in the L11 (R) class. A charac-
terization for potentials in the L11 (R) class, by using the Marchenko method and the
continuity of the scattering matrix at zero energy, can also be found in [19].
For the matrix Schrödinger equation on the line Aktosun, Klaus, and van der Mee
proved in [22] the continuity of the scattering matrix at zero energy for potentials in
the L11 (R) class. For the matrix Schrödinger equation on the half line, the continuity
of the scattering matrix at zero energy for potentials in the class L11 (R+ ) was proved
by Agranovich and Marchenko [11] in the case of the Dirichlet boundary condition
and by Aktosun, Klaus, and Weder [23] in the case of the general self-adjoint bound-
ary condition. We remark that the continuity result by Agranovich and Marchenko
with the Dirichlet boundary condition can be found in Theorem 5.6.2 on p. 137
of [11].
The problem of the reconstruction of the entire potential on the full line from
some appropriate scattering data, when the potential is partially known has been
studied by various researchers such as Novikova and Markushevich [214]; Aktosun,
Klaus, and van der Mee [20]; Aktosun [13, 14]; Rundell and Sacks [227]; Grébert
and Weder [132]; Gesztesy and Simon [127]; Aktosun and Weder [29]; and Akto-
sun and Papanicolaou [26]. For the inverse resonance problem where the potential
is constructed from some data containing the resonances, we refer the reader to
the studies by Marletta, Shterenberg, and Weikard [194] and Matthew and Weikard
[195] and the references quoted in the aforementioned papers. For various further
results in nonhomogeneous media, energy-dependent potentials, and other related
problems on the inverse scattering on the line, we refer the reader to the monograph
by Chadan and Sabatier [78] and the review by Aktosun and Klaus [18]. For direct
and inverse scattering problems for systems of first-order differential equations on
the line, we refer the reader to the work by Beals and Coifman [53, 54]. For the
analysis of the inverse scattering for the Dirac system on the line the reader can
consult the study by Hinton, Jordan, Klaus, Shaw [143]. For the study of the direct
and inverse scattering for higher-order equations on the line we refer the reader to
the studies by Beals [52] and by Beals, Deift, Tomei [55].
In 1967 Gardner, Greene, Kruskal, and Miura [116] found a remarkable and as-
tonishing application of the inverse scattering problem for the Schrödinger equation
on the line to the solution of the Cauchy problem for the Korteweg–de Vries equa-
tion. There is an extensive literature on this method, which is now known as the in-
verse scattering transform, and of its relation to the theory of solitons and the theory
of integrable nonlinear evolution equations. For a presentation of these results we
refer the reader to the works by Ablowitz and Clarkson [1]; Ablowitz, Kaup, Newell,
and Segur [2]; Ablowitz and Segur [3]; Aktosun [16]; Faddeev and Takhtajan [107];
Fokas [110]; Lax [181]; Marchenko [192]; Novikov, Manakov, Pitaevskii, and Za-
kharov [213], and many references contained in these works. For the application of
the inverse scattering transform to nonlinear evolution equations on star graphs the
reader can consult the paper by Caudrelier [76] and the references therein.
For a generalization of the Gel’fand–Levitan method and the Marchenko method
to discrete systems on the half line and on the line we refer the reader to many
references such as those by Case [71–73], Case and Chiu [74], and Case and Kac
[75].
A problem closely related to an inverse scattering problem is an inverse spectral
problem. In an inverse spectral problem one uses the spectrum of a related opera-
tor to obtain information on the coefficients in an equation and the corresponding
boundary conditions. We can refer to a coefficient to be recovered as the potential
by using the analogy with the Schrödinger equation. An inverse spectral problem
can be formulated in various ways depending on the input data available. One pos-
sible input data set to use is the spectral function, which is the case [121, 122] in
the Gel’fand–Levitan method. The method by Krein [165–170] is of this nature. In
inverse spectral problems involving only a discrete spectrum, as the input data set
one can use the entire spectrum and an appropriate set of norming constants or use
the spectra corresponding to two different boundary conditions. Another possibility
is to use a mixed input data set consisting of information related to a part of the
discrete or continuous spectrum, some norming constants, and partial knowledge of
the potential. The use of mixed input data may be appropriate if one is interested in
determining a part of the spectrum, the norming constants, or the boundary condi-
tions not known a priori. In using as input the spectra from two different boundary
conditions, one can instead use a mixed input data set by combining some infor-
mation from the two spectral data sets. There are other possibilities in the choice
of input data, such as using the Weyl–Titchmarsh m-function or the spectral shift
function.
The first inverse spectral problem was probably considered by Lord Rayleigh,
who briefly discussed the possibility of constructing the variable density of a string
from the frequencies of its vibration, for which we refer the reader to p. 172 of
Section 141 in Chapter VI of [238]. However, with the development of quantum
mechanics and specially with the introduction of the Schrödinger equation in 1925,
the spectra of differential equations acquired a much greater importance and a dif-
ferent physical significance. The equations used to describe mechanical vibrations
12 1 Introduction
were applied to the dynamics of atoms and molecules and the spectrum was related
to energy levels. It is remarkable that the first result in the inverse spectral problem
for the Schrödinger equation was obtained only four years after its creation by Vic-
tor Ambartsumian [34]. Ambartsumian proved that if the eigenvalues corresponding
to the Schrödinger operator in a finite interval with the Neumann boundary condi-
tions at both ends coincide with the eigenvalues when the potential is identically
zero, then it is necessary that the potential has to be identically zero. This is actu-
ally a very special case, and Borg [66] proved that in general one spectrum is not
enough to determine the corresponding Sturm–Liouville operator. Borg considered
two Schrödinger operators on a finite interval for which the boundary condition at
one end is the same and the boundary condition at the other end is different. He
proved that if the eigenvalues for the two Schrödinger operators coincide then the
potentials and the boundary conditions must be the same for both operators.
For the Schrödinger operator on the half line, Marchenko proved [191] that the
potential and the boundary condition are uniquely determined either by the spectral
function or by the Weyl–Titchmarsh m-function. Borg [67] and Marchenko [191]
studied the Schrödinger operator on the half line with a boundary condition at the
origin when there is no continuous spectrum. They proved that two sets of dis-
crete spectra associated with two different boundary conditions at the origin (and
a fixed boundary condition at infinity if any) uniquely determine the potential and
the boundary condition at the origin. Gesztesy and Simon obtained [126] a gener-
alization of the Borg–Marchenko theorem to the case with a continuous spectrum,
and they obtained a uniqueness result by using a spectral shift function as input;
in fact, their paper contained various other uniqueness results for inverse spectral
problems. Aktosun and Weder established [30, 31] another generalization of the
Borg–Marchenko theorem to the case with a continuous spectrum by using the in-
put data given in terms of the spectral measure, and they proved that the input data
set consisting of the discrete eigenvalues for one boundary condition at the origin,
the continuous part of the spectral measure for that boundary condition, and a subset
of the discrete eigenvalues for a different boundary condition uniquely determines
the potential and the boundary conditions at the origin. Levitan and Gasymov [187]
solved the characterization problem for two spectra by providing the necessary and
sufficient conditions for two sets of real numbers to be the spectra of a Schrödinger
operator in a finite interval with different boundary conditions. Gasymov and Dzha-
biev [118] solved the characterization problem using two spectra for the Dirac equa-
tion on a finite interval.
Hochstadt and Lieberman [144] proved that if two Schrödinger operators on the
interval [0, 1] satisfy the same boundary conditions and their potentials coincide in
the interval (1/2, 1) then their potentials must also coincide in the interval (0, 1/2).
The same result and its various generalizations were proved by Gesztesy and Simon
[130] by using another method. The reader is referred to [130] for various other
results and a detailed discussion of the literature on the Hochstadt–Lieberman the-
orem. Simon introduced a new method [235] for the inverse spectral theory, and its
elaborations were given by Gesztesy and Simon [129, 131] and Ramm and Simon
[220]. One of the new results [235] obtained with this method is a local version
of the Borg–Marchenko theorem. In their monograph [219] Pöschel and Trubowitz

presented a detailed study of the inverse spectral theory for the Schrödinger operator
on a finite interval and they solved the characterization problem with their method.
The monographs [286, 287] by Yurko and [111] by Freiling and Yurko present
an extensive account of the direct and inverse spectral theories for differential op-
erators on the half line and on the full line, in particular an account of the inverse
spectral theory for higher-order operators using the Weyl matrix as the input data.
Eckhardt, Gesztesy, Nichols, Sakhnovich, and Teschl considered [88] the inverse
spectral problems for the matrix Schrödinger operators with distributional poten-
tials by using the Weyl–Titchmarsh matrix-valued function as input data.
There is a discrete version of the inverse spectral problem on the half line and
on the full line, where one studies the inverse spectral problem for Jacobi operators.
The reader is referred to the references by Akhiezer [12], Simon [234], and Teschl
[241] for basic results on Jacobi matrices. Using the m-function techniques Gesztesy
and Simon [128] proved several uniqueness results in the inverse spectral problem
for finite and semi-infinite Jacobi matrices.
The inverse problem for semi-infinite Jacobi matrices using two spectra was stud-
ied by various authors such as Donoghue Jr. [85], Fu and Hochstadt [115], Gu-
seinov [134], Halilova [136], and Teschl [240]. Using two spectra Silva and Weder
[232, 233] presented a comprehensive and detailed solution to the inverse prob-
lem for semi-infinite Jacobi matrices; in particular, they solved the characterization
problem by providing a set of necessary and sufficient conditions in order for two
sets of real numbers to be the eigenvalues of a unique Jacobi operator with differ-
ent boundary conditions. Weikard [282] proved a local Borg–Marchenko theorem
for difference equations. The inverse resonance problem for Jacobi operators was
studied [193] by Marletta, Naboko, Shterenberg, and Weikard.
For further information on the inverse scattering and spectral problems for differ-
ential and discrete operators on a finite interval, on the half line, and on the full line,
the reader can consult the monographs [11, 55, 78, 111, 186, 192, 219, 286, 287],
the review by Gesztesy [125], and the references therein.
There is also a large literature on the multidimensional inverse scattering the-
ory. In the inverse scattering problems in two or more dimensions, it is necessary
to deal with partial differential equations rather than ordinary differential equations.
The scattering problem on the half line is related to the scattering problem in three
dimensions as follows. Suppose that in three dimensions we have a potential that
is spherically symmetric. Using spherical harmonics, we can decompose the scat-
tering problem as a direct sum of partial waves each with a fixed angular momen-
tum. Each partial wave with a fixed angular momentum corresponds to a separate
scattering problem on the half line. Thus, the inverse scattering problem in three
dimensions with a spherically symmetric potential corresponds to obtaining infor-
mation on the potential when the scattering matrix or the phase shift is known for
each angular momentum. A natural question from the three-dimensional viewpoint
is whether the potential can be determined by the knowledge of the phase shifts
at all energies and all angular momenta. This was actually the question raised by
Heisenberg [140–142] in the particular case of potential scattering. For clarifica-
14 1 Introduction
tion, we note that the nonuniqueness [47, 48] indicated by Bargmann occurs when
the phase shift is known at all energies but at one angular momentum value. Thus,
Bargmann’s nonuniqueness result does not bring a conclusive answer to Heisen-
berg’s question. The breakthrough came from Faddeev [100] who proved that in
three dimensions the high-energy limit of the scattering matrix uniquely determines
the potential and hence also the bound-state energies. Faddeev considered potentials
that were not necessarily spherically symmetric, but using his result in the particular
case of potentials with spherical symmetry one concludes that the high-energy limit
of the phase shifts known for all angular momenta uniquely determines the potential
and hence also the bound-state energies. Thus, Faddeev’s result [100] proved that
Heisenberg’s conjecture [140–142] was correct.
At a formal level, modulo an assumption about exceptional points, a character-
ization for the inverse scattering problem in two or more dimensions is given by
Faddeev [106]. In this characterization there are two key elements playing a central
role. One of them is a new Green’s function known as the Faddeev Green’s function,
and the other is a new class of exponentially growing solutions known as the Fad-
deev solutions, which were previously introduced by Faddeev in [104, 105]. Using
a nonlinear ∂¯ -equation, some related characterizations in two or more dimensions
were given by Novikov and Henkin in [212] and by Weder in [246]. The charac-
terization given by Novikov and Henkin utilizes some results from several complex
variables, and the characterization by Weder is based on a limiting absorption prin-
ciple for the Faddeev Green’s function [247].
Let us mention some uniqueness and nonuniqueness results for the multidimen-
sional inverse Schrödinger scattering at fixed energy. Novikov [210] and Eskin
and Ralston [95] proved that the potential is determined uniquely by the scatter-
ing matrix at a fixed energy if that potential decays exponentially at infinity. On
the other hand, if we consider general short-range potentials decaying at infinity as
C(1 + |x|)−1−ε or faster, where C and ε are some fixed positive parameters, it is
known that the scattering matrix at a fixed energy does not uniquely determine the
potential. In fact, some examples [78] of radially oscillating potentials in three di-
mensions are known in such a way that those potentials decay proportional to |x|−2
at infinity with the scattering matrix being equal to the identity at some positive en-
ergy. In two dimensions, there are also examples [133] of potentials decaying pro-
portional to |x|−2 at infinity with the scattering matrix being equal to the identity at
some positive energy. Weder [248] proved that if two potentials have the same scat-
tering amplitude at some positive energy and they coincide outside some ball then
they necessarily coincide everywhere. In fact, the same problem also appears in dif-
ferent settings. We refer the reader to the work by Weder [254] and the references
therein for the proof that the scattering matrix at a fixed energy uniquely determines
an exponentially decreasing perturbation in stratified media. Weder [268, 269] also
proved that the scattering matrix at a fixed quasi-energy uniquely determines time-
periodic potentials that decay exponentially at infinity.
It is a remarkable fact that the uniqueness in the inverse scattering problem at a
fixed energy is restored for short-range potentials having a regular behavior at infin-
ity. It was proved by Weder and Yafaev [279] that if a short-range potential is equal
to a sum of homogeneous terms outside a ball then the scattering matrix at a fixed
energy uniquely determines the potential. In [279] a method for the reconstruction
of the potential was also given, and more generally it was proved that if a short-
range potential is asymptotic to an infinite sum of homogenous terms at infinity
then the scattering matrix at a fixed energy uniquely determines each one of the ho-
mogeneous terms. For similar results in the case of long-range potentials, including
the Coulomb potential, we refer the reader to the work by Weder and Yafaev [280].
Similar problems in the short-range case were studied using a different method by
Joshi [152] and by Joshi and Sá Barreto [153, 154].
For more information on multidimensional inverse scattering problems, we refer
the reader to [78, 203, 211, 245] and the references therein.
There are two main approaches in the analysis of scattering problems. One of
them is the stationary approach and the other is the time-dependent approach. All
the results so far mentioned in this section related to the inverse scattering theory
have been obtained by using methods in the stationary approach. These methods are
based on the study of time-periodic solutions to the evolution equations describing
the scattering process. In a scattering process, if the particles are far apart from each
other then the interaction between them is negligible, and if the particles are close
to each other then they interact with each other. The scattering process describes
the time evolution of particles that are initially, i.e., at large negative times, far apart
from each other in the region where the interaction is weak. As time evolves the
particles come close to each other in the region where the interaction is strong. Fi-
nally, at large positive times the particles are again far apart from each other in the
region where the interaction is weak. At any time, the particles are localized in a
well-defined region of space. These two properties, i.e., the propagation in time and
the localization in space, form an essential part of any scattering process. In the sta-
tionary approach, one uses time-periodic solutions that actually do not propagate
in time and hence such solutions are not localized in space. In other words, in the
stationary approach the time-propagation and space-localization properties are lost.
Of course, these two properties are recuperated by using the Fourier analysis after
the scattering problem is solved. The main point is that these two properties are not
used in the stationary approach in the analysis of scattering problems. Nevertheless,
the stationary methods make up a powerful tool, and a lot of information in the
scattering of waves and particles is obtained by using the stationary approach.
On the other hand, the applicability of stationary methods has its limitation in
situations where it is not practical to discard the two aforementioned properties. For
example, the use of stationary methods is not practical in the scattering of three or
more particles where the configuration space is large and at each time the solution
is localized in a small region of space. Another such example is when there are
topological effects in quantum mechanics. For example, in the magnetic Aharonov–
Bohm effect a charged particle, e.g., an electron, interacts with a toroidal magnet and
it is essential to know if the electron travels through the hole of the torus or outside of
the hole. For such scattering problems it is essential to use time-dependent methods
that consider the scattering of finite-energy wave packets so that we can analyze the
two aforementioned properties in a precise manner.
16 1 Introduction
In [91–93] Enss and Weder pioneered the use of time-dependent methods (scat-
tering of wave packets) in the inverse scattering theory, and in particular they ini-
tiated a time-dependent method to estimate the high-velocity (or high-energy) be-
havior of the scattering solutions and of the scattering matrix. In this method all
the physical properties of the scattering process including the time-propagation and
space-localization are utilized in the proofs. Consequently, this method has a lot of
flexibility, which makes it possible to solve many problems by using proofs that
closely follow the physical intuition. We now briefly review some of the results that
have been obtained by using this method.
Using this method Enss and Weder [91–93] analyzed the non-relativistic
Schrödinger equation for an N-body system with long-range potentials. The Stark
effect (scattering by a decaying electric potential and an electric field that is con-
stant in space) for an N-body system was considered by Weder [250], Nicoleau
[206], Adachi and Maehara [5], and Valencia and Weder [243]. The Stark effect
was analyzed by Nicoleau [207] with a time-periodic electric field and by Adachi,
Kamada, Kazuno, and Toratani [6] with electric fields that are asymptotically zero in
time. Enss and Weder [94] analyzed the inverse problem for two-cluster scattering.
Arians [37, 38] and Jung [159] considered two-body systems with decaying elec-
tromagnetic potentials in the entire space. Arians [39] considered the inverse prob-
lem with decaying electromagnetic potentials and a constant magnetic field. Weder
[252] considered an N-body system in the entire space with time-dependent poten-
tials. The relativistic case of the Dirac equation was studied by Jung [157, 158] with
time-independent potentials and by Ito [148, 149] with time-dependent potentials.
Weder analyzed the nonlinear Schrödinger equation in [253, 255, 258–262] and the
nonlinear Klein–Gordon equation in [256, 264]. Weder [271, 272] studied the forced
nonlinear Schrödinger equation on the half line. Sandoval Romero and Weder [229]
considered a nonlinear Schrödinger equation with a potential and a nonlocal nonlin-
earity. Nicoleau [209] considered the Schrödinger equation with a strong repulsive
potential, and Nicoleau [208] studied the Schrödinger equation in a semiclassical
process. Daudé and Nicoleau [81, 82] studied the scattering in a black-hole metric.
Nicoleau [205], Weder [265], and Ballesteros and Weder [41–46] considered the
magnetic Aharonov–Bohm effect and Weder [273] studied the electric Aharonov–
Bohm effect. We refer the reader to the review papers [251, 257, 263, 266, 270, 275]
for further information.
Another approach in the inverse scattering and spectral theories is the boundary
control method introduced by Belishev [56]. For reviews of this method we refer the
reader to [57, 58] by Belishev. For the relation between the boundary control method
and Simon’s method [235], the reader can consult the paper [228] by Rybkin. The
scattering theory for nonself-adjoint operators has been extensively studied, and we
refer the reader to the studies by Falconi, Faupin, Fröhlich, and Schubnel [108], by
Faupin and Fröhlich [109], and the references in these papers. For scattering on the
line with energy-dependent potentials, the reader can consult the work by Aktosun,
Klaus, van der Mee [21] and the references therein. There is also an extensive lit-
erature on the direct and inverse problems for canonical systems, and we refer the
reader to [40] and the references quoted there. For some additional references on
the literature for the scattering process, the reader can consult Sect. 4.1.
Chapter 2
The Matrix Schrödinger Equation and
the Characterization of the Scattering
Data
Abstract In this chapter we describe the basic ingredients of the direct and inverse
scattering problems for the matrix Schrödinger equation on the half line with the
general self-adjoint boundary condition. We show how the analysis of star graphs
and the Schrödinger scattering problem on the full line can be reduced to the study of
the matrix Schrödinger equation on the half line with some appropriate self-adjoint
boundary conditions. To analyze the direct and inverse problems on the half line,
we introduce the input data set consisting of a potential and two constant boundary
matrices describing the boundary condition. We define the Faddeev class of input
data sets by imposing some appropriate restrictions on the input data sets. We in-
troduce the scattering data set consisting of a scattering matrix and the bound-state
information. We define the Marchenko class of scattering data sets by imposing
some appropriate restrictions on the scattering data sets. The unique solutions to the
direct and inverse scattering problems are achieved by establishing a one-to-one cor-
respondence between the Faddeev class of input data sets and the Marchenko class
of scattering data sets. Various equivalent descriptions of the Marchenko class are
introduced. Such equivalent descriptions allow us to present various different but
equivalent results for the characterization of the scattering data in the solution to the
inverse problem. For the reader’s convenience various equivalent characterization
theorems are stated but their proofs are postponed until Chap. 5.
2.1 Outline of the Chapter
In this chapter we present the matrix Schrödinger equation on the half line contain-
ing an n × n matrix-valued potential V (x). We also introduce the general self-adjoint
boundary condition at x = 0 in terms of a pair of constant n × n matrices denoted
by A and B, which are uniquely determined up to a postmultiplication by an invert-
ible matrix. An input data set D is identified, as in (2.5.1), as the set consisting of
© Springer Nature Switzerland AG 2021 19

T. Aktosun, R. Weder, Direct and Inverse Scattering for the Matrix
Schrödinger Equation, Applied Mathematical Sciences 203,
https://doi.org/10.1007/978-3-030-38431-9 2
20 2 The Matrix Schrödinger Equation and the Characterization of the Scattering Data
the potential and the two boundary matrices. The Faddeev class of input data sets
is described in Definition 2.5.1 by imposing the restrictions (2.2.2) and (2.2.3) on
the potential V (x) and the restrictions (2.2.5) and (2.2.6) on the boundary matri-
ces A and B. Corresponding to each input data set in the Faddeev class, we then
have a self-adjoint Schrödinger operator on the half line. Associated with that self-
adjoint Schrödinger operator, we introduce various relevant quantities to analyze
the corresponding direct and inverse problems. Such quantities include the Jost so-
lution f (k, x) appearing in (2.2.26), the Jost matrix J(k) appearing in (2.2.27), the
scattering matrix S(k) appearing in (2.2.28), the physical solution Ψ (k, x) appear-
ing in (2.2.29), the bound-state energies −κ 2j , the multiplicity of each bound state
m j , the normalization matrix M j , and the normalized bound-state matrix solution
Ψj (x) appearing in (2.2.30). A scattering data set S is described, as in (2.5.2), as the
set consisting of the scattering matrix, the bound-state energies, the multiplicity of
each bound state, and the normalization matrix for each bound state. The Marchenko
class of scattering data sets is described in Definition 2.5.5 by introducing the appro-
priate restrictions on the scattering data sets. A first characterization of the scattering
data sets is achieved by establishing a one-to-one correspondence between the Fad-
deev class of input data sets and the Marchenko class of scattering data sets. Several
equivalent formulations of this characterization are presented, and this is done by
introducing various lists of properties that are satisfied by a scattering data set if and
only if it belongs to the Marchenko class. Moreover, we give a characterization of
the scattering data set that differs from the other characterization formulations in the
way in which the boundary condition is obtained from the scattering data set.
The chapter is organized as follows. In Sect. 2.2 we introduce the matrix
Schrödinger equation, the self-adjoint boundary condition (2.2.4), the boundary ma-
trices A and B, and other relevant quantities used in the analysis of the corresponding
direct and inverse problems. We also show that A and B are uniquely determined up
to a postmultiplication by an invertible matrix. In Sect. 2.3 we show that the di-
rect and inverse scattering problems on a star graph can be analyzed as a special
case of the corresponding problems for the half-line matrix Schrödinger operator. In
Sect. 2.4 it is shown that the full-line Schrödinger equation can also be analyzed as a
special case of the 2×2 matrix Schrödinger equation. In Sect. 2.5, the input data sets
and the scattering data sets are introduced and the Faddeev class of input data sets is
defined. Then, various properties identified by Arabic and Roman numerals are in-
troduced as (1), (2), (3), (4), and (I), (III), (V), (VI). In fact, there are two versions
of (3) denoted by (3a ) and (3b ); there are five versions of (4) denoted by (4a ), (4b ),
(4c ), (4d ), (4e ); there are three versions of (III) denoted by (IIIa ), (IIIb ), (IIIc );
and there are eight versions of (V) denoted by (Va ), (Vb ), (Vc ), (Vd ), (Ve ), (V f ),
(Vg ), (Vh ). Having defined various properties for a scattering data set S, we are able
to formulate various characterizations for S so that it corresponds to a unique input
data set D in the Faddeev class, where we recall that the uniqueness of the bound-
ary matrices is understood up to postmultiplication by an invertible matrix. Such a
typical characterization has the form (1, 2, 3, 4), by which we mean S satisfies the
properties (1), (2), either of the two properties (3a ) or (3b ), and any one of the five
properties listed as (4a ), (4b ), (4c ), (4d ), (4e ). Another typical characterization has
2.1 Outline of the Chapter 21
the form (1, 2, III + V, 4), by which we mean S satisfies (1, 2, 4), and that instead
of satisfying either of the two properties (3a ) or (3b ), it satisfies two other proper-
ties, the first of which being one of the three properties (IIIa ), (IIIb ), (IIIc ) and the
second being one of the eight properties (Va ), (Vb ), (Vc ), (Vd ), (Ve ), (V f ), (Vg ),
(Vh ). The choice of notation enables us to streamline all the characterizations of S.
In Sect. 2.6 we provide a formulation of the characterization of a scattering data set
in the Marchenko class in the form (1, 2, 3a , 4a ). We also indicate that, by replacing
(2.2.3) with (2.6.1) in the definition of the Faddeev class and by replacing (2.5.8)
with (2.6.2) in the property (2), we obtain a characterization of the scattering data
sets in a subclass of the Marchenko class. In Sect. 2.7 we provide several formula-
tions of the characterization in the form (1, 2, III + V, 4). We then present another
◦ ◦
form of the characterization in the form (1, 2, L + 5, 4). Here, the property (L) is
introduced in Definition 2.7.7 and it is related to Levinson’s theorem; the property
◦ ◦
(5) is described in Theorem 2.7.9; and the property (4) has three versions denoted
◦ ◦ ◦
by (4c ), (4d ), (4e ), which are all described in Definition 2.7.8. Finally, in Sect. 2.8
still another characterization is presented in the form (I + VI, 2, A + V, 4), where
the property (A) is described in Theorem 2.8.1. This final characterization is based
on the solution to the inverse problem via the method summarized in Sect. 5.11 and
utilizing the generalized Fourier map studied in Chap. 5.
We note that one of the characterization formulations, namely (1, 2, 3a , 4a ), en-
ables us to define the Marchenko class of scattering data sets, which is given in
Definition 2.5.5. Let us also remark on the construction of the boundary matrices A
and B in the solution to the inverse problem. Related to the characterization formula-
tions (1, 2, 3, 4) and (1, 2, III + V, 4), we mention that the matrices A and B are con-
structed by using the data related to the high-energy limit of the scattering matrix.
On the other hand, related to the characterization formulations (I + VI, 2, A + V, 4),
the boundary matrices A and B are constructed by exploiting the unitarity of the
generalized Fourier map, which is studied in Chap. 5.
Note that some of the characterization formulations may have certain advantages
over others. For example, the characterization formulations in the form (1, 2, 3, 4)
require that we have at hand the solution to the Marchenko integral equation (3.14.1)
in order to check if (3) is satisfied. On the other hand, the characterization formula-
tions in the form (1, 2, III+V, 4) and (I+VI, 2, A+V, 4) do not require that we have
at hand the solution to the Marchenko integral equation. Having various equivalent
formulations of the characterization allows us to use various different techniques to
solve the corresponding inverse scattering problem, and it also allows us to under-
stand how all these different techniques are related to each other. For example, in
some cases it may be advantageous to solve an integral equation rather than a corre-
sponding Riemann–Hilbert problem or vice versa. It may also be more convenient
to look for a solution to an integral equation in the class of bounded and integrable
functions rather than in the class of bounded and square-integrable functions. It may
be more convenient to look for a solution to a homogenous Riemann–Hilbert prob-
lem in a more restricted class rather than in a Hardy space. In Sects. 2.6, 2.7, and 2.8
we state several theorems, each presenting a characterization formulation using a
specific set of conditions. The equivalent but different forms of the characterization
allow us to choose a specific set of conditions among many possibilities, and this of-
fers a great flexibility in choosing the solution method to solve the inverse problem
depending on the available data type.
For the convenience of the reader, with the exception of the proof of Proposi-
tion 2.2.1, the details of the proofs in this chapter are presented in later chapters.
2.2 The Matrix Schrödinger Equation on the Half Line
In this section we introduce the matrix Schrödinger equation (2.2.1), the matrix po-
tential V (x), and the boundary matrices A and B used to describe the self-adjoint
boundary condition at x = 0. In specifying the self-adjoint boundary condition,
we show that the boundary matrices A and B are uniquely determined modulo a
postmultiplication by an invertible matrix. We also introduce other main quanti-
ties associated with the corresponding direct scattering problem, i.e., the Jost solu-
tion f (k, x), the Jost matrix J(k), the scattering matrix S(k), the physical solution
Ψ (k, x), the bound-state normalization matrices M j , and the bound-state matrix so-
lutions Ψj (x).
Consider the matrix Schrödinger equation on the half line given by
− ψ (k, x) +V (x) ψ (k, x) = k2 ψ (k, x), x ∈ R+ . (2.2.1)
We note that R+ := (0, +∞), the prime denotes the derivative with respect to the
spatial coordinate x, the scalar k2 is the complex-valued spectral parameter, the po-
tential V (x) is an n × n self-adjoint matrix-valued function of x and belongs to class
L11 (R+ ), and n is any positive integer. We assume that n is fixed and is known. The
self-adjointness of V (x) is expressed as
V (x) = V (x)† , x ∈ R+ , (2.2.2)
where the dagger denotes the matrix adjoint (complex conjugate and matrix trans-
pose). We equivalently say Hermitian to describe a self-adjoint matrix. We remark
that, unless we are in the scalar case, i.e., unless n = 1, the potential is not necessar-
ily real valued. The condition V ∈ L11 (R+ ) means that each entry of the matrix V (x)
is Lebesgue measurable in R+ and
∞
dx (1 + x) |V (x)| < +∞, (2.2.3)
0
where |V (x)| denotes the operator matrix norm. Clearly, a matrix-valued function
belongs to L11 (R+ ) if and only if each entry of that matrix belongs to L11 (R+ ).
The wavefunction ψ (k, x) appearing in (2.2.1) may be an n × n matrix-valued
function or it may be a column vector with n components. We are interested in
studying (2.2.1) with an n × n self-adjoint potential V in L11 (R+ ) under the gen-
2.2 The Matrix Schrödinger Equation on the Half Line 23
eral self-adjoint boundary condition at x = 0. There are various equivalent formu-

lations [23, 32, 137–139, 163, 164] of the general self-adjoint boundary condition
at x = 0, where further discussions are presented in Sect. 3.4. We find it convenient
[23, 32] to present the boundary condition in terms of two constant n × n matrices
A and B as
− B† ψ (0) + A† ψ (0) = 0, (2.2.4)
where A and B satisfy
− B† A + A† B = 0, (2.2.5)
A† A + B† B > 0. (2.2.6)
We will refer to A and B as the boundary matrices. In Sect. 3.8 we provide some
illustrative examples for particular cases of the general self-adjoint boundary condi-
tion described in (2.2.4)–(2.2.6), including the so-called Kirchhoff boundary condi-
tion, the δ -type boundary condition, and the δ -type boundary condition. The con-
dition in (2.2.6) means that the n × n matrix (A† A + B† B) is positive definite. We
refer to a positive-definite matrix as a positive matrix.
One can directly verify that (2.2.4) remains invariant if the boundary matrices
A and B are replaced with AT and BT, respectively, where T is an arbitrary n × n
invertible matrix. The details of this invariance are provided in the next proposition.
We express this fact by saying that the self-adjoint boundary condition in (2.2.4) is
uniquely determined by the matrix pair (A, B) modulo an invertible matrix T, and
we equivalently state that (2.2.4) is equivalent to the knowledge of (A, B) modulo T.
Proposition 2.2.1. Given n × n matrices A1 , B1 , A2 , B2 , assume that
A†1 A1 + B†1 B1 > 0. (2.2.7)
Let
1 := (Z1 , Z2 ) ∈ C2n : −B†1 Z1 + A†1 Z2 = 0 , (2.2.8)

2 := (Z1 , Z2 ) ∈ C2n : −B†2 Z1 + A†2 Z2 = 0 . (2.2.9)
Then, 1 = 2 if and only if there exists an invertible matrix T such that
A2 = A1 T, B2 = B1 T. (2.2.10)
Proof. If (2.2.10) holds, then using (2.2.10) in (2.2.8) and (2.2.9) we conclude that
1 = 2 . Let us now prove that (2.2.10) holds if 1 = 2 . Let D1 and D2 be the
operators from C2n into Cn defined as
D1 (Z1 , Z2 ) := −B†1 Z1 + A†1 Z2 , Z1 , Z2 ∈ Cn , (2.2.11)
D2 (Z1 , Z2 ) := −B†2 Z1 + A†2 Z2 , Z1 , Z2 ∈ Cn . (2.2.12)

We have 1 = 2 if and only if
Ker [D1 ] = Ker [D2 ], (2.2.13)

where Ker denotes the kernel. Let us use Ran to denote the range. In order to prove
that (2.2.10) is satisfied, we first prove that Ran[D1 ] is equal to Cn . Suppose that W
is a vector in Cn orthogonal to Ran [D1 ], i.e.,
W, −B†1 Z1 + A†1 Z2 = 0, Z1 , Z2 ∈ Cn , (2.2.14)
where ·, · is the standard scalar product in Cn . In (2.2.14) choosing Z1 = −B1 Z

and Z2 = A1 Z with Z ∈ Cn , we obtain
(A†1 A1 + B†1 B1 )W, Z = 0, Z ∈ Cn . (2.2.15)
Choosing Z as Z = (A†1 A1 + B†1 B1 )W, from (2.2.15) we get
(A†1 A1 + B†1 B1 )W, (A†1 A1 + B†1 B1 )W = 0. (2.2.16)
Using (2.2.7) in (2.2.16), we conclude that W = 0. Thus, we have shown that the
only vector in Cn that is orthogonal to Ran [D1 ] is the zero vector. Because Cn is the
direct sum of Ran [D1 ] and its orthogonal complement, we conclude that Ran [D1 ] =
Cn . Since the range of D1 is Cn , for any U ∈ Cn there exist Z1 and Z2 in Cn such
that
U = −B†1 Z1 + A†1 Z2 . (2.2.17)
Let us use E to denote the operator from Cn into Cn defined as
E U = −B†2 Z1 + A†2 Z2 . (2.2.18)
Let us confirm that E U is well defined, i.e., it is independent of the particular

(Z1 , Z2 ) that we use in (2.2.17) to represent U. So, suppose that for some other
Z̃1 and Z̃2 in Cn we have
U = −B†1 Z̃1 + A†1 Z̃2 . (2.2.19)
Equating the right-hand sides of (2.2.17) and (2.2.19) we obtain

−B†1 Z1 − Z̃1 + A†1 Z2 − Z̃2 = 0,
which indicates that

(Z1 − Z˜1 , Z2 − Z˜2 ) ∈ Ker [D1 ]. (2.2.20)
Then, from (2.2.13) and (2.2.20) it follows that
(Z1 − Z̃1 , Z2 − Z̃2 ) ∈ Ker[D2 ]. (2.2.21)
Using (2.2.21) in (2.2.11) and (2.2.12) we get
− B†2 Z1 + A†2 Z2 = −B†2 Z̃1 + A†2 Z̃2 . (2.2.22)
Comparing (2.2.18) and (2.2.22) we observe that E U is well defined. Let us also use
E to denote the n × n matrix, in the canonical basis, representing the linear operator
2.2 The Matrix Schrödinger Equation on the Half Line 25
E appearing in (2.2.18). The reader is referred to Sect. A.13 for the matrix repre-
sentation of a linear operator. We would like to show that the matrix E is invertible.
For this, it is enough to prove that the only solution to the linear system E U = 0 is
the zero vector in Cn , i.e., U = 0. Using the representation of U as in (2.2.17), from
(2.2.18) and E U = 0 we get
E U = −B†2 Z1 + A†2 Z2 = 0,
which indicates that (Z1 , Z2 ) ∈ Ker [D2 ]. On the other hand, as a result of (2.2.13)
we have (Z1 , Z2 ) ∈ Ker [D1 ]. Thus, with the help of (2.2.11), (2.2.12), and (2.2.17)
we obtain
U = −B†1 Z1 + A†1 Z2 = D1 (Z1 , Z2 ) = 0. (2.2.23)
As seen from (2.2.23) we have U = 0, and thus the matrix E is invertible. Letting
Z2 = 0 in (2.2.17) we get
E U = E (−B†1 Z1 ) = −B†2 Z1 , Z1 ∈ Cn ,
which implies
E B†1 = B†2 . (2.2.24)
In a similar way, letting Z1 = 0 in (2.2.17), we obtain
E A†1 = A†2 . (2.2.25)
From (2.2.24) and (2.2.25) we see that (2.2.10) holds with T = E † .
Next, we introduce the main quantities associated with the direct scattering prob-
lem for the matrix Schrödinger equation on the half line. The elaborations on such
quantities are provided in Chap. 3. We assume that the potential V (x) satisfies (2.2.2)
and (2.2.3) and that the boundary matrices A and B satisfy (2.2.5) and (2.2.6).
(a) Regardless of the boundary matrices A and B, the matrix Schrödinger equation
(2.2.1) possesses the n × n matrix-valued Jost solution f (k, x) satisfying the
asymptotic condition
f (k, x) = eikx [I + o(1)], x → +∞, (2.2.26)
where we recall that I is the n × n identity matrix. The existence and uniqueness
of f (k, x) as well as its relevant properties are summarized in Proposition 3.2.1.
(b) In terms of the boundary matrices A and B and the Jost solution f (k, x), the
corresponding n × n Jost matrix J(k) is given by
J(k) := f (−k∗ , 0)† B − f (−k∗ , 0)† A, k ∈ R, (2.2.27)
where the asterisk denotes complex conjugation. The redundant appearance of

k∗ instead of k in (2.2.27) when k ∈ R is useful in extending the domain of the
Jost matrix from k ∈ R to k ∈ C+ .
(c) In terms of the Jost matrix J(k) given in (2.2.27), the n × n scattering matrix
S(k) is given by
S(k) := −J(−k) J(k)−1 , k ∈ R. (2.2.28)
(d) In terms of the Jost solution f (k, x) appearing in (2.2.26) and the scattering
matrix S(k) appearing in (2.2.28), the n × n matrix-valued physical solution
Ψ (k, x) is given by
Ψ (k, x) := f (−k, x) + f (k, x) S(k), k ∈ R. (2.2.29)
In Proposition 3.2.12 it is shown that Ψ (k, x) is a solution to (2.2.1) and it satis-

fies the boundary condition (2.2.4).
(e) The extension of the Jost matrix J(k) from k ∈ R to k ∈ C+ is invertible in
C+ except perhaps at a finite number of k-values on the positive imaginary
axis. Thus, there exists a nonnegative number N and a set of N distinct positive
numbers κ j such that, as shown in Proposition 3.11.13, at k = iκ j for 1 ≤ j ≤ N
the Schrödinger equation (2.2.1) has m j linearly independent square-integrable
column-vector solutions satisfying the boundary condition (2.2.4), where each
m j is a positive integer less than or equal to n. It is possible to rearrange those
linearly independent column-vector solutions into an n × n matrix Ψj (x) in such
a way that Ψj (x) can be uniquely constructed as
Ψj (x) := f (iκ j , x) M j , 1 ≤ j ≤ N, (2.2.30)
where M j is an n × n nonnegative Hermitian matrix of rank m j . We will refer

to Ψj (x) as the normalized bound-state matrix solution. The unique construc-
tion of M j is given in (3.11.91). The values −κ 2j for 1 ≤ j ≤ N represent the
bound-state energies of the matrix Schrödinger equation (2.2.1) with the bound-
ary condition (2.2.4), the columns of Ψj (x) are eigenfunctions corresponding to
−κ 2j , and the integer m j is the multiplicity of the bound state at k = i κ j . In the
particular case when N = 0, the matrix Schrödinger operator corresponding to
(2.2.1) and (2.2.4) has no bound states, and hence there are no corresponding
κ j and m j values. We will refer to N as the number of bound states without
including the multiplicities.
2.3 Star Graphs
In this section we show that the half-line matrix Schrödinger equation with a
diagonal-matrix potential is unitarily equivalent to the Schrödinger equation on a
star graph. Consequently, the results that we obtain in this monograph for the direct
and inverse scattering problems on the half-line matrix Schrödinger equation also
apply to the corresponding direct and inverse problems on star graphs.
A star graph is a quantum graph with only one vertex and a finite number, say n,
of semi-infinite edges. The Hilbert space relevant to a star graph is given by
2.3 Star Graphs 27
n
H := L2 (R+ , C),
j=1
with L2 (R+ , C) denoting the Hilbert space of square-integrable functions on R+

taking values in C. For clarity, in this section at times we use L2 (R+ , C) instead of
the equivalent notation L2 (R+ ). An element Y of H is a finite sequence of the form
Y(x) := {Y1 (x), · · · ,Yn (x)}, (2.3.1)
and the scalar product of two vectors Y and W in H is expressed in terms of their
sequence elements as
n
(Y, W) := ∑ (Y j ,W j )L2 (R+ ,C) . (2.3.2)
j=1
The Schrödinger equation on the star graph is given by
(LY) (x) = k2 Y(x), (2.3.3)
where
(LY) (x) := {−Y1 (x) +V1 (x)Y1 (x), · · · , −Yn (x) +Vn (x)Yn (x)},
with the potentials V j (x) for 1 ≤ j ≤ n being real-valued functions satisfying (2.2.3).
The relevant boundary condition for the star graph is given by
− B† Y(0) + A† Y (0) = 0, (2.3.4)
where the n × n matrices A and B satisfy (2.2.5) and (2.2.6). As in the case of the
matrix Schrödinger equation, this is the general self-adjoint boundary condition for
a star graph. For example, the Kirchhoff boundary condition corresponds to a special
case of (2.3.4) We refer the reader to [23, 32, 137–139, 163, 164] and Sect. 3.4 for
further information on the boundary condition for star graphs.
Let us now prove that a star graph is unitarily equivalent to a half-line matrix
Schrödinger operator with the diagonal-matrix potential given by
V (x) := diag{V1 (x), · · · ,Vn (x)}, (2.3.5)
and with the boundary condition (2.2.4) involving the boundary matrices A and B
appearing in (2.3.4). We define the unitary operator U from H onto L2 (R+ ) via
⎡ ⎤
Y1 (x)
⎢ ⎥
U Y(x) := ⎣ ... ⎦ . (2.3.6)
Yn (x)
With the help of (2.3.2), using the appropriate norms induced by the corresponding
scalar products, we get
||U Y|| = ||Y||H , Y∈H , (2.3.7)
and we stress that U is onto L2 (R+ ). Let us use ψ (x) to denote column vector
appearing on the right-hand side of (2.3.6). With the help of (2.3.1) and (2.3.5) we
observe that Y is a solution to (2.3.3) if and only if ψ is a solution to (2.2.1). In
a similar way, we see that Y satisfies the boundary condition (2.3.4) if and only if
ψ satisfies the boundary condition (2.2.4) with the same boundary matrices A and
B appearing in (2.3.4). Thus, we conclude the unitary equivalence between a star
graph and a half-line Schrödinger operator.
2.4 The Schrödinger Equation on the Full Line
In this section we first show that a 2 × 2 matrix Schrödinger equation on the half line
is unitarily equivalent to a scalar Schrödinger equation on the full line with a point
interaction at x = 0. After that, we generalize this result to systems and we prove
that a 2n × 2n matrix Schrödinger equation on the half line is unitarily equivalent
to an n × n matrix Schrödinger equation on the full line with a point interaction at
x = 0. Hence, the results in this monograph apply, in particular, to the scalar and
matrix Schrödinger equations on the full line.
The relevant Hilbert space for the scalar Schrödinger equation on the full line
is L2 (R, C), where by L2 (R, C) we denote the Hilbert space of square-integrable
functions on R taking values in C. In terms of the scalar quantities ψ1 (x) and ψ2 (x),
let us define the column vector-valued quantity ψ (x) with two components as
ψ (x) := (ψ1 (x), ψ2 (x))T ,
with the superscript T denoting the matrix transpose. Next, we define the unitary
operator U from L2 (R+ ) onto L2 (R, C) as

ψ1 (x), x ≥ 0,
U ψ (x) := (2.4.1)
ψ2 (−x), x < 0.
We then have
U ψ L2 (R,C) = ψ L2 (R+ ) , ψ ∈ L2 (R+ ),
and U is onto L2 (R, C). Suppose that the matrix potential V in (2.2.1) is diagonal,
i.e.,
V (x) := diag{V1 (x),V2 (x)},
for some scalar-valued potentials V1 (x) and V2 (x). Let us use Y (x) to denote the
right-hand side of (2.4.1), i.e., we let
2.4 The Schrödinger Equation on the Full Line 29

ψ1 (x), x ≥ 0,
Y (x) :=
ψ2 (−x), x < 0,
so that we have

ψ1 (x), x > 0,
Y (x) :=
−ψ2 (−x), x < 0.
We remark that
Y (0+ ) = ψ1 (0), Y (0− ) = ψ2 (0), Y (0+ ) = ψ1 (0), Y (0− ) = −ψ2 (0).
We conclude that ψ (x) satisfies the Schrödinger equation (2.2.1) if and only Y (x)
satisfies the Schrödinger equation on the full line given by
−Y (x) + Q(x)Y (x) = k2 Y (x), x ∈ R \ {0},
with the real-valued scalar potential

V1 (x), x ≥ 0,
Q(x) :=
V2 (−x), x < 0.
Moreover, ψ satisfies the boundary condition (2.2.4) if and only if Y satisfies the
point-interaction condition
⎧
⎨−[B† ]11 Y (0+ ) − [B† ]12 Y (0− ) + [A† ]11 Y (0+ ) − [A† ]12 Y (0− ) = 0,
(2.4.2)
⎩
−[B† ]21 Y (0+ ) − [B† ]22 Y (0− ) + [A† ]21 Y (0+ ) − [A† ]22 Y (0− ) = 0,
where [A† ]i j and [B† ]i j denote the (i, j) entries of the matrices A† and B† , respec-
tively. For example, suppose that ψ satisfies the so-called δ -type boundary condition
ψ1 (0) = ψ2 (0), ψ1 (0) + ψ2 (0) = c ψ1 (0),
where c is a real parameter, for which the special case c = 0 corresponds to the
Kirchhoff boundary condition. In this case, the matrices appearing in (2.4.2) are
given by
0 1 −1 c
A= , B= . (2.4.3)
0 1 1 0
The boundary conditions (2.4.2) corresponding to the matrices A and B given in
(2.4.3) are
Y (0− ) = Y (0+ ), Y (0+ ) −Y (0− ) = cY (0+ ),
which is related to the Schrödinger equation on the line with a δ -point interaction
with the coupling constant c, which occurs when the potential contains the compo-
nent c δ (x). In particular, when c = 0 we obtain the Schrödinger equation on the full
line without point interaction at x = 0.
Let us now consider a 2n × 2n matrix Schrödinger equation on the half line. In

terms of the column vector-valued quantities ψ1 (x) and ψ2 (x) each with n com-
ponents belonging to L2 (R+ , Cn ) , let us define the column vector-valued quantity
ψ (x) with 2n components as
ψ (x) := (ψ1 (x), ψ2 (x))T ,
where we recall that the superscript T denotes the matrix transpose.

Analogous to
the scalar case we define the unitary operator U from L2 R+ , C2n onto L2 (R, Cn )
by letting

ψ1 (x), x ≥ 0,
Uψ (x) := (2.4.4)
ψ2 (−x), x < 0.
Let the potential in (2.2.1) be the block diagonal matrix given by
V (x) := diag{V1 (x),V2 (x)},
where V1 (x) and V2 (x) are some self-adjoint n × n matrices. As in the scalar case,
let us use Y (x) to denote the right-hand side of (2.4.4). Then, we see that ψ satisfies
the Schrödinger equation (2.2.1) if and only if Y satisfies the matrix Schrödinger
equation on the line given by
−Y (x) + Q(x)Y (x) = k2 Y (x), x ∈ R \ {0},
with the n × n matrix potential

V1 (x), x ≥ 0,
Q(x) :=
V2 (−x), x < 0.
Let us decompose the 2n × 2n matrices A and B appearing in (2.2.4) as

A1 B1
A= , ,
A2 B2
with A1 , A2 , B1 , B2 being n× 2n matrices. Hence, ψ satisfies the boundary condition

(2.2.4) if and only if Y satisfies the point-interaction condition at x = 0 given by
− B†1 Y (0+ ) − B†2 Y (0− ) + A†1 Y (0+ ) − A†2 Y (0− ) = 0. (2.4.5)
For example, let us consider the special choice

0 I −I Λ
A= , B= , (2.4.6)
0 I I 0
where Λ is a self-adjoint n × n matrix. The matrices in (2.4.6) satisfy (2.2.5) and

(2.2.6). The point-interaction condition in (2.4.5) is then given by
2.5 The Faddeev Class and the Marchenko Class 31
Y (0+ ) = Y (0− ), Y (0+ ) −Y (0− ) = Λ Y (0), (2.4.7)
where Y (0) denote the common value in the first equality in (2.4.7). The point-
interaction condition in (2.4.7) corresponds to a δ -type interaction at x = 0 with the
coupling matrix Λ . If Λ = 0, which is analogous to having c = 0 in (2.4.3), then Y (x)
and Y (x) are continuous at x = 0, and in that case we obtain the matrix Schrödinger
equation on the full line without a point interaction at x = 0.
2.5 The Faddeev Class and the Marchenko Class
In this section, in preparation for the characterization of the scattering data, we

introduce the Faddeev class of input data sets and the Marchenko class of scattering
data sets. The description of the Faddeev class is given in Definition 2.5.1, and the
description of the Marchenko class is given in Definition 2.5.5. As mentioned in
Sect. 2.1, the characterization of the scattering data essentially consists of showing
that there is a one-to-one correspondence between the Marchenko class and the
Faddeev class.
In Definition 2.5.2 we introduce various properties identified by some Arabic
numerals and in Definition 2.5.3 we introduce various properties identified by some
Roman numerals. Various versions of some of those properties are indicated by us-
ing some subscripts. We later establish various equivalents among some of those
properties. We define the Marchenko class in terms of the properties (1), (2), (3a ),
and (4a ). By using the equivalents among various properties, we are able to describe
the Marchenko class in various equivalent forms, and this is the idea behind present-
ing various forms of the characterization of the scattering data. Such equivalents are
presented in Sect. 5.6. Later in Proposition 5.4.2(a) we show that if the physical
solution to the Schrödinger equation satisfies the boundary condition, then the nor-
malized bound-state matrix solutions automatically satisfy the boundary condition.
In the direct scattering problem related to (2.2.1) and (2.2.4), our input data set
D is given by
D := {V, A, B}, (2.5.1)
with the understanding that V is equivalent to the knowledge of the n × n matrix
V (x) for x ∈ R+ , the boundary matrices A and B are two constant n × n matrices,
and the matrices A and B are uniquely specified up to a postmultiplication by an
invertible n × n matrix T.
In the inverse scattering problem our scattering data set S is given by

S := S, {κ j , M j }Nj=1 , (2.5.2)
with the understanding that S is equivalent to the knowledge of the n × n scattering

matrix S(k) specified for k ∈ R, the positive constants κ j are N distinct numbers
related to the bound-state energies −κ 2j , the n × n matrices M j are N constant matri-
ces related to the normalizations of matrix-valued bound-state wavefunctions, and
each M j is nonnegative Hermitian and has rank m j for some integer between 1 and
n. Thus, we use N to denote the number of bound states without counting the mul-
tiplicities. We remark that the possibility N = 0 is included in our consideration, in
which case the scattering data set consists of the scattering matrix alone. The finite-
ness of N is guaranteed by Theorem 3.11.1(g). The integer m j corresponds to the
multiplicity of the bound state associated with k = iκ j , and hence the integer N
defined as the sum of the ranks of the matrices M j , i.e.,
N
N := ∑ m j, (2.5.3)
j=1
corresponds to the total number of bound states including the multiplicities.

In our analysis of the direct problem related to (2.2.1) and (2.2.4), we assume
that our input data set D belongs to the Faddeev class defined below.
Definition 2.5.1. The input data set D given in (2.5.1) is said to belong to the Fad-
deev class if the potential V (x) satisfies (2.2.2) and (2.2.3) and the matrices A and B
satisfy (2.2.5) and (2.2.6). Equivalently, D belongs to the Faddeev class if the n × n
matrix-valued potential V (x) appearing in (2.2.1) is Hermitian and belongs to class
L11 (R+ ) and the constant n × n matrices A and B appearing in (2.2.4) satisfy (2.2.5)
and (2.2.6).
In our monograph, we provide various equivalent formulations of the character-
ization of the scattering data set S given in (2.5.2) when the corresponding D in
(2.5.1) belongs to the Faddeev class. In order to state those characterization proper-
ties in an efficient manner, we first introduce a set of properties that are all indicated
with Arabic numerals. Such individual properties are later used in order to form a
set of conditions to identify the Marchenko class and are also used to present various
formulations of the characterization of the scattering data sets so that those sets in
the Marchenko class are in a one-to-one correspondence with the input data sets in
the Faddeev class. Without always mentioning explicitly, we assume that the scatter-
ing matrix S(k) is an n × n matrix whose entries are Lebesgue-measurable functions
for k ∈ R.
Definition 2.5.2. The properties (1), (2), (3a ), (3b ), (4a ), (4b ), (4c ), (4d ), (4e ) for
the scattering data set S in (2.5.2) are defined as follows:
(1) The scattering matrix S(k) satisfies
S(−k) = S(k)† = S(k)−1 , k ∈ R, (2.5.4)
and there exist constant n × n matrices S∞ and G1 in such a way that

G1 1
S(k) = S∞ + +o , k → ±∞. (2.5.5)
ik k
Furthermore, we have Fs (y) bounded in y ∈ R and integrable in y ∈ R+ . Here,

Fs (y) is the n × n matrix-valued quantity defined as
2.5 The Faddeev Class and the Marchenko Class 33
∞
1
Fs (y) := dk [S(k) − S∞ ] eiky , y ∈ R. (2.5.6)
2π −∞
We remark that the constant matrix S∞ can be obtained from the scattering
matrix S(k) via
S∞ := lim S(k). (2.5.7)
k→±∞
We also remark that, as a result of (2.5.5), the matrix S(k) − S∞ is square inte-
grable in k ∈ R and consequently Fs (y) belongs to L2 (R).
(2) For the matrix Fs (y) defined in (2.5.6), the derivative Fs (y) satisfies
∞
dy (1 + y) |Fs (y)| < +∞, (2.5.8)
0
where we recall that the norm used in (2.5.8) is the operator norm of a matrix.
We remark that for this property to hold the matrix S∞ defined in (2.5.7) should
exist and the quantity S(k) − S∞ should be square integrable in k ∈ R.
(3a ) The physical solution Ψ (k, x) to the Schrödinger equation (2.2.1) satisfies the
boundary condition (2.2.4), i.e.,
− B†Ψ (k, 0) + A†Ψ (k, 0) = 0, k ∈ R. (2.5.9)
We clarify this property as follows. The scattering data set S yields a particular
n × n matrix-valued solution Ψ (k, x) to (2.2.1) known as the physical solution
given in (2.2.29) and also yields a pair of matrices A and B (modulo an invert-
ible matrix) satisfying (2.2.5) and (2.2.6). Our statement (3a ) is equivalent to
saying that (2.2.4) is satisfied if we use in (2.2.4) the quantities Ψ (k, x), A, and
B constructed from S.
(3b ) The Jost matrix J(k) satisfies
J(−k) + S(k) J(k) = 0, k ∈ R. (2.5.10)
We clarify this property as follows: The scattering data set S yields a Jost matrix
J(k) constructed as in (2.2.27), unique up to a postmultiplication by an invert-
ible matrix. Using the scattering matrix S(k) given in S and the Jost matrix
constructed from S we must have (2.5.10) satisfied.
(4a ) The Marchenko equation (3.14.1) at x = 0 given by
∞
K(0, y) + F(y) + dz K(0, z) F(z + y) = 0, y ∈ R+ , (2.5.11)
0
has a unique solution K(0, y) in L1 (R+ ). Here, F(y) is the n × n matrix related
to Fs (y) given in (2.5.6) as
N
F(y) := Fs (y) + ∑ M 2j e−κ j y , y ∈ R+ . (2.5.12)
j=1
In the case where N = 0 the sum on the right-hand side of (2.5.12) is absent and
F(y) = Fs (y) for y ∈ R+ . We remark that, in order for this property to hold, the
matrix S∞ defined in (2.5.7) should exist and that the quantity S(k) − S∞ should
be square integrable in k ∈ R.
(4b ) The only solution in L1 (R+ ) to the integral equation
∞
K(0, y) + dz K(0, z) F(z + y) = 0, y ∈ R+ , (2.5.13)
0
is the trivial solution K(0, y) ≡ 0. We remark that (2.5.13) is the homogeneous

version of the Marchenko integral equation at x = 0 given by (2.5.11) and that
F(y) appearing in (2.5.13) is the n × n matrix defined in (2.5.12). We also re-
mark that, in order for this property to hold, the matrix S∞ defined in (2.5.7)
should exist and the quantity S(k) − S∞ should be square integrable in k ∈ R.
(4c ) The only integrable solution X(y), which is a row vector with n integrable com-
ponents in y ∈ R+ , to the linear homogeneous integral equation
∞
X(y) + dz X(z) F(z + y) = 0, y ∈ R+ , (2.5.14)
0
is the trivial solution X(y) ≡ 0. Note that F(y) appearing in (2.5.14) is the n × n
matrix defined in (2.5.12). We remark that, in order for this property to hold,
the matrix S∞ defined in (2.5.7) should exist and that the quantity S(k) − S∞
should be square integrable in k ∈ R.
(4d ) The only solution X̂(k) to the system
⎧
⎨X̂(iκ j ) M j = 0, 1 ≤ j ≤ N,
(2.5.15)
⎩
X̂(−k) + X̂(k) S(k) = 0, k ∈ R,
where X̂(k) is a row vector with n components belonging to the class L̂1 (C+ ),
is the trivial solution X̂(k) ≡ 0. For the definition of L̂1 (C+ ) we refer the reader
to Sect. A.11.
(4e ) The only solution h(k) to the system
⎧
⎨M j h(iκ j ) = 0, 1 ≤ j ≤ N,
(2.5.16)
⎩
h(−k) + S(k) h(k) = 0, k ∈ R,
where h(k) is a column vector with n components belonging to the class

L̂1 (C+ ), is the trivial solution h(k) ≡ 0.
Let us make some comments on the property (1) of Definition 2.5.2 above. The
property (2.5.5) implies that S(k)−S∞ is square integrable in k ∈ R, and hence it also
implies that Fs (y) belongs to L2 (R). In fact, (2.5.5) contains even more information.
The quantity G1 and hence (2.5.5) itself are used to construct the boundary matrices
A and B appearing in (2.2.4)–(2.2.6). On the other hand, for the construction of the
Another random document with
no related content on Scribd:
CHAPITRE IV
LES HAUTS PLATEAUX
La ville de Tcharkalik commence par un asile, vetus urbes

condentium consilium. « Mot profond que la situation de toutes les
vieilles villes de l’antiquité et du moyen âge commente
éloquemment… » On parle de Rome ; l’histoire (?) de sa fondation
m’est soudainement revenue à la mémoire, et je me surprends à
citer quelques lignes de Michelet.
La faute en est à ce que la géographie, les choses présentes,
pour mieux dire, jettent des lueurs sur les choses passées et font
entrevoir ou deviner la préhistoire.
Ce que je vois à Tcharkalik, ce que l’on me conte, pourrait fournir
matière à des rapprochements bien intéressants. Peut-être pourrais-
je vous donner une idée, vague sans doute, de la manière que les
villes naissaient dans l’antiquité. Je dis « naissaient dans
l’antiquité », car il ne se passe ici rien de comparable à ce que j’ai vu
dans les colonies russes de l’Asie, rien de comparable à ses
jaillissements de villes américaines qui commencent par un hôtel
éclairé à la lumière électrique et l’installation du téléphone.
En revanche, nous pourrions vous offrir le spectacle d’une
agglomération de réfugiés, chasseurs, chercheurs d’or, cultivateurs
de la terre ; vous les montrer en lutte avec des autochtones qui leur
sont intellectuellement inférieurs, puisque, avant l’arrivée des
Khotanlis, les Lobis n’avaient pas encore jeté la première mue de la
civilisation, puisqu’ils étaient encore chasseurs et qu’ils ignoraient ou
ne pratiquaient pas le labour.
Nous pourrions vous montrer ce qu’est une question agraire à sa
première phase, ce qu’est la politique intérieure la plus rudimentaire,
et la politique extérieure telle qu’elle vient de naître du voisinage
d’un plus fort et d’un besoin d’alliance, et tout cela, cher lecteur, sans
aucun ministère.
Mais laissons cela.
Voici novembre, le 1er novembre, et nous n’avons pas terminé
notre besogne. Ce que nous venons de faire n’a été qu’une agréable
promenade semée d’inconvénients si minimes, qu’ils en étaient
comme les condiments, la rendant encore plus agréable.
Nous vous avons dit que la première grande étape était Kourla ;
la seconde est Tcharkalik ; la troisième sera Batang, si nous
continuons à réaliser notre programme avec le même bonheur. Et
Batang est loin, des déserts nous en séparent, l’inconnu est en
travers de nous. Et après Batang, c’est le Tonkin, à l’autre bout de
l’Asie, au bord de l’Océan. Heureusement qu’en voyage on n’a pas
de temps à perdre et qu’en règle générale les voyageurs, rêvassant
peu, ne cherchent pas à s’imaginer des raisons de ne pas agir ; sans
quoi, nous pourrions nous effrayer de notre entreprise, quoique nous
ne puissions pas dire que notre projet soit au-dessus des forces de
l’homme, car nous n’avons encore rien essayé. Les circonstances
peuvent nous être plus favorables qu’à nos devanciers, et nous
réussirons. Qui sait ?
Il nous faut donc trouver à tout prix cette route, que nous
appelons, dans nos conversations, « la route du Sud ». Nous
mettons nos gens en campagne, et chacun cherche à découvrir
l’homme précieux qui la connaîtra et voudra nous la montrer. Il suffit
que l’un de nos gens ait questionné maladroitement, et voilà que
nous ne pouvons obtenir un renseignement précis. Au reste, peu de
nos serviteurs se soucient de poursuivre le voyage ; les personnes
aimant l’exploration ailleurs que dans leur chambre ou dans un bon
campement sont plus rares qu’on ne pense. La route du Sud est
notre grande préoccupation, elle le sera longtemps.
Nos trois Sibériens vont nous quitter. Ils devaient venir jusqu’au
Lob Nor, et j’essaye vainement de les entraîner plus loin. Ils ne s’en
soucient pas.
Le chamelier doungane veut aussi retourner sur ses pas, nous
ne le retenons que par l’appât du gain.
Nous cherchons des volontaires dans les gens du pays pour
remplacer ceux qui partiront. Deux se présentent, l’un connaît le
chemin de Bogalik, suivi autrefois par l’Anglais Carey. Nous leur
promettons de bons gages, et leur entrée dans la troupe relève un
peu le moral des Dounganes.
Le chef de nos chameliers le vieil Imatch, bien que marchant
difficilement, ira jusqu’au bout, jusqu’où nos Khotanlis iront.
Ce brave homme au rude parler n’a pas de crainte, son seul
désir serait d’avoir de meilleures jambes. Il est affectueux, il aime
ses chameaux et ne les veut pas quitter.
Il est Kirghiz d’origine, de la tribu des Kizaï, qui habitent la Sibérie
et la province de Kouldja.
Souvent, le soir, autour du feu, nous faisons causer le vieil
Imatch. Il aime à parler de son pays et du grand événement de sa
jeunesse, qui fut l’insurrection contre les Chinois.
« Je n’ai pas oublié ces choses-là, dit-il, et aussi longtemps que
je vivrai elles me resteront dans la tête. J’avais vingt ans, il y a de
cela vingt-deux ans. Les Dounganes et les Tarantchis se soulevèrent
les premiers. Nous autres Kizaï vivons loin des villes et nous
n’avions que peu de rapports avec les Chinois : aussi, dès l’abord,
nous ne voulions pas nous soulever. Mais les Dounganes nous
envoyèrent des émissaires avec ce message : « Si vous refusez de
nous aider, lorsque nous en aurons fini avec les infidèles (Chinois),
nous tournerons toutes nos forces contre vous et nous vous
anéantirons ». Alors les anciens de nos tribus tinrent conseil et ils
dirent : « Nous ne risquons rien à aider les Dounganes et les
Tarantchis, car il est clair, d’après ce qui se passe, que pas un
Chinois ne survivra. Il y aura beaucoup de butin pour les braves.
Nous avons une belle occasion de nous enrichir aux dépens des
infidèles. Marchons avec l’aide d’Allah et nous reviendrons riches. »
« Alors nous nous sommes armés de sabres et de haches, car
nous avions peu de fusils, et, montés sur de bons étalons, nous
avons attaqué les Chinois.
— Étaient-ils braves ?
— Plus faciles à tuer que des moutons.
— En as-tu tué aussi ?
— Oui, j’en ai tué le plus que j’ai pu. A chaque pas nous
rencontrions des fuyards, nous leur prenions leurs chevaux, leurs
vêtements s’ils en valaient la peine. Parfois on laissait la vie aux
jeunes, mais ceux qui les rencontraient plus loin la leur enlevaient.
— Vraiment, les Chinois n’étaient pas braves ?
— Braves comme… dit Imatch en crachant après s’être servi
d’une comparaison malpropre. Seuls les Solons ont montré un peu
plus de cœur. Quel Kirghiz pourra oublier Baïan-taï. On avait cerné
cette ville, et nous avons tout tué, à l’exception des enfants et des
femmes. Nous nous les sommes partagés et nous avons été punis
de notre faiblesse, car ces femmes ont gâté le sang de notre race. »
Imatch conte ces exploits d’une voix rauque, avec les gestes de
frapper de grands coups de taille. Aussi lui donnons-nous un grand
sabre russe de cavalerie avec lequel il se propose de pourfendre les
ennemis.
Nous recrutons deux indigènes ; je crois que nous aurons lieu de

nous en louer plus tard.
L’aîné s’appelle Timour. Il a été pâtre, il est chercheur d’or et
chasseur lorsqu’il a des loisirs. Il est marié et il cultive un coin de
terre. Il a souvent erré dans l’Altyn Tagh, le Tchimène Tagh, et les
hauts plateaux ne l’effrayent point. Il exécute les ordres sans
broncher, il travaille vite, on le dit infatigable marcheur, il sait soigner
les chevaux et les chameaux. Il rit volontiers, il est d’humeur égale,
et, qualité précieuse, il est content de son sort à Tcharkalik. Un tout
petit morceau de sucre en fait le plus heureux des hommes. Tout ce
que nous faisons l’intéresse : il regarde les armes avec plaisir, les
oiseaux préparés avec attention, il les reconnaît, dit leur nom. C’est
un curieux. Le soir on l’entend chanter, raconter des légendes ;
quand Rachmed ou un autre débite une histoire, il en suit toutes les
péripéties avec soin, riant, s’exclamant ; bref, c’est un poète, un
aventurier, un amoureux du nouveau.
Lorsqu’on lui demande s’il fera froid dans les montagnes du Sud :
« Oui », répond-il, et il cache ses mains dans les manches de son
vêtement, et les réchauffe sous ses aisselles, puis il agite les bras
en disant : « Ce n’est rien », et il rit. Avec cela il n’est pas trop grand,
pas trop gros, et très alerte, il danse légèrement. Bien plus, il sait les
prières, on le tient pour un mollah, et il possède des paroles contre
les maladies. Un homme complet, comme vous voyez.
L’autre, plus jeune, âgé d’une vingtaine d’années, a nom Iça. Il
est très vigoureux. Il dépouille un mouton avec la plus grande
dextérité et sait parfaitement cuire le riz. Avouons qu’il mange l’un et
l’autre avec une non moins grande dextérité. Tout ce qui a trait à la
cuisine l’intéresse : fendre le bois, allumer le feu, l’entretenir, aller
quérir de l’eau, nettoyer la marmite, ce sont là nobles besognes dont
il s’acquitte à souhait. Il se souvient le lendemain de ce qu’on lui a dit
la veille. Il a un rire éclatant, mais tellement naturel qu’on l’entend
avec plaisir. D’habitude il est assez sombre. Il aurait le défaut de
fumer le hachich, mais en petite quantité. Ceux qu’il a servis sont
contents de lui. Une nuit, je l’ai vu se coucher sur une simple natte
posée près du feu, sans autre vêtement qu’un kalat déchiré. Il a
dormi fort bien à cette place, quoique le feu se fût éteint et que le
minimum de la nuit fût de − 19 degrés. Le lendemain il s’est levé très
gai et sans le moindre rhume de cerveau. Vous comprenez que nous
ayons arrêté là l’examen qu’on lui faisait subir à son insu, et que
nous lui ayons pardonné d’avoir fui la maison paternelle, après avoir
cassé les deux bras à sa belle-mère.
En effet, Iça avait reçu une femme de la main de son père.
L’union était, paraît-il, heureuse. Mais la seconde femme du père
d’Iça avait pris en haine son beau-fils et sa bru, et, raconte-t-il, il n’y
avait pas d’avanies, de méchancetés qu’elle ne fît subir à la jeune
femme en l’absence des hommes.
Iça résolut de se venger. Un jour que la marâtre était seule à la
maison, il la roua irrespectueusement de coups de bâton, si bien
qu’il la laissa pour morte dans la cour et les deux bras cassés.
Il renvoya sans tarder sa propre femme à sa famille, il réunit ce
qu’il put du butin, et, ayant conté l’accident à quelques amis, il monta
à cheval et se sauva de Kiria. Après des aventures diverses il vint
échouer à Tcharkalik, qui paraît être un lieu de rendez-vous pour les
originaires de la province de Khotan pressés d’un subit besoin de
prendre l’air.
Nous renouvelons les provisions, nous en achetons encore que
nous comptons faire transporter par les indigènes jusqu’à ce qu’elles
soient épuisées. Car, je le répète, c’est le point important pour nous,
il nous faut assurer la subsistance des hommes et des bêtes.
Le jour de la fête de la naissance de Mahomet, les autorités en
corps viennent nous rendre visite et nous offrent des présents. Ils
veulent que nous participions à leurs réjouissances, car nous
sommes loin de notre patrie, de nos foyers, et il serait malséant à
eux de ne pas nous inviter. Je les remercie, je leur répète que nous
n’avons pas de mauvaises intentions au fond du cœur, et je leur
affirme que toujours nos actes seront conformes à nos paroles. Ils
nous croient : « Vous êtes des hommes vrais, nous le voyons bien »,
disent-ils. Ils demandent l’autorisation de prendre nos serviteurs à
leurs tables. Tout cela est accordé, bien entendu, et toute la journée
on fête Mahomet par des repas, par des chants, par des danses, par
des luttes, où Rachmed, qui est très adroit, obtient un grand succès.
Deux moutons offerts par nous sont cuits dans la marmite immense
de la mosquée. Cette marmite finit mal, car, l’ayant employée pour
raffiner du sel cristallisé, nous la faisons éclater : accident de très
mauvais augure qu’un cadeau fait supporter sans murmure.
Le 7 novembre une épouvantable tempête de nord-est hurle
toute la nuit et nous oblige à construire un abri pour notre cuisine. La
température s’abaisse subitement, et le matin les indigènes nous
arrivent déguisés en gens du Nord. Tous sont vêtus de peaux de
moutons ou de fourrures de bêtes sauvages, telles que les renards,
les loups. Notre troupe profite de cette bonne occasion pour essayer
ses costumes d’hiver, et c’est une véritable mascarade.
Une nouvelle intéressante est que quatre Kalmouks sont arrivés
à Abdallah. Ils formeraient l’avant-garde du khan des Kalmouks qui
revient de Lhaça, où il est allé en pèlerinage. Il ne tarderait pas à
arriver, en assez piteux état. Sa caravane a été décimée : deux
cents chameaux et vingt hommes sont morts. Le retour s’est
effectué surtout avec des koutasses (des yaks) et en passant par le
Tsaïdam. Car d’après le messager, vingt ans auparavant, le khan
des Kalmouks ayant essayé de se rendre à la « Ville des Esprits »
par la route du Kizil Sou aurait dû rebrousser chemin, parce que les
montagnes sont infranchissables.
L’aksakal des Khotanlis m’ayant apporté de la graisse de
marmotte afin de me guérir d’une attaque de rhumatismes, je le
questionne au sujet de la route du Kizil Sou, et sans se prononcer
franchement il me donne à entendre que l’on ne doit pas attacher
grande importance aux paroles de ce Lobi. « Quant aux difficultés de
la route, ajoute-t-il, elles sont réelles. Une fois, nous sommes allés
du côté de Bogalik avec cent cinquante ânes afin de rapporter de l’or
et des peaux, car la chasse est bonne, et nous avons perdu du
monde et beaucoup d’ânes.
— Pour quelles raisons ?
— Par le froid et surtout par les odeurs mauvaises qui
s’échappent du sol [2] . Elles vous tuent en vous empoisonnant. Les
ânes y résistent moins que les hommes. »
[2] Mal de montagne.
Après la tempête, l’atmosphère est moins empoussiérée ; le ciel

brumeux devient clair ; pas un souffle n’agite l’air, mais il gèle plus
fort que ne le voudraient les indigènes. Sous la tente le minimum a
été de − 12 degrés. Cet abaissement considérable de la température
a jeté l’alarme parmi la population. Tous ont quitté leurs maisons :
tous ceux qui peuvent porter un fagot sur le dos se sont égaillés
dans la brousse. Et c’est un va-et-vient continuel de femmes, de
vieillards et surtout d’enfants chargés. Pas une fillette qui ne soit
courbée par un poids plus lourd que son corps ; l’hiver est là, le
grand aryk est gelé, les champs en jachère où l’eau s’était répandue
sont blancs de gelée.
Il nous tarde aussi de partir.
Le 9 novembre le minimum est de − 19 degrés avec une petite
brise nord-ouest rafraîchissante, le maximum est encore de + 20
degrés, mais au soleil, où l’on se trouve fort bien. Les indigènes ont
dirigé l’eau de l’aryk vers les citernes, ils font leurs provisions d’eau
pour l’hiver. Depuis une semaine tous les moulins tournaient en
prévision de cette sécheresse, chacun faisait sa provision de farine.
Un artiste qui me paraît remarquable s’accompagne d’une
guitare à deux cordes et nous chante près du feu une chanson
pleine de philosophie. Elle marque cependant la cadence aux
danseurs et aux danseuses. Tandis que les femmes marchent à
petits pas, yeux baissés, et qu’elles cherchent des attitudes
gracieuses du torse, et se balancent ou tournent les bras étendus, le
chanteur hurle à tue-tête : « Le monde n’est qu’une tromperie,
l’homme passe son temps à désirer, il attend toute sa vie la
réalisation de vœux qu’il lui est aussi difficile d’obtenir que de saisir
la lune elle-même, laquelle il revoit cependant chaque mois. »
Vous voyez que les moralistes ne manquent pas à Tcharkalik, au
seuil du Gobi. Le chanteur passe pour être l’auteur de ces couplets,
et nous lui proposons de nous accompagner ainsi que sa guitare,
faite de deux planches de peuplier bien lisse. Un moraliste fera bien
dans notre troupe, surtout celui-ci. Il a couru le monde, il a vu
Yarkand et je ne sais combien d’années cherché l’or en tous
endroits. Il ne paraît pas avoir fait fortune et ses déboires lui auront
inspiré cette chanson résignée. Il passe pour un brave homme, et à
propos de la fête de Mahomet il a encore remporté le prix de la lutte
aux jeux Olympiques.
Quoique Khotanli, il est l’ami intime d’un certain Abdoullah Ousta,
maître dans l’art de travailler le fer, qui est de Lob. Autrefois Tokta —
c’est le nom du chanteur — a rendu un service considérable au vieil
Abdoullah. Celui-ci se serait égaré en poursuivant des chameaux
sauvages, il n’aurait pu rejoindre ses compagnons, et Tokta serait
survenu fort à propos pour secourir le chasseur, fatigué et mourant
de faim. Depuis ce jour une amitié solide lie ces deux hommes.
Nous avons commandé du fer, des clous, des piquets à
Abdoullah Ousta et nous espérons l’enrôler. D’après Tokta, personne
ne connaît mieux la montagne que le vieux maître, qui est encore
très vigoureux, quoique sa barbe soit grisonnante.
S’il consent à partir avec nous, sa décision en entraînera
beaucoup d’autres.
Voilà de bonnes paroles ; on nous fait bonne figure, on promet
tout, mais attendons la fin.
Tokta, avant de nous quitter, assure qu’on nous aidera si les
barbes blanches des Lobis ne s’y opposent pas. Les Khotanlis nous
seraient acquis.
Rachmed prétend qu’on doit croire Tokta :
« J’en suis sûr, dit-il, il nous accompagnera, car il est Saïa.
— Qu’est cela, « Saïa » ?
— Un homme comme nous, qui ne peut rester en place, par la
faute de sa mère.
— Explique-toi.
— Oui, voilà ce qui m’est arrivé, ce qui a dû arriver à Tokta. Nos
mères étant grosses de nous ont voyagé à dos de chameau dans le
désert, elles ont promené un regard tout autour d’elles en cherchant
à voir au delà de l’horizon et elles ont fait de nous des « Saïa », des
coureurs de grands chemins, voulant toujours voir au delà de
l’horizon. Et voilà pourquoi nous allons encore marcher vers le sud,
et Allah seul peut dire quand et où nous nous arrêterons. Et nous
ferons bien de partir, car la route me paraît longue et ces maudits
chameaux ne vont guère vite : qu’Allah nous aide ! »
Là-dessus Rachmed me reproche de l’avoir pris à mon service
alors qu’il avait à peine de la barbe, de lui avoir fait pousser plus de
cheveux blancs qu’il n’en a de noirs, et, par de trop longues
absences, fait « rater » plusieurs mariages avantageux. Puis, comme
il est mobile, il passe à un autre ordre d’idées, fait une farce à son
voisin et l’accable de ces injures que les Ousbegs profèrent sans
méchanceté.
Rachmed a raison : il est temps de partir, mais tout n’est pas
encore prêt ; il faut que le Doungane se décide à nous accompagner,
et alors on partagera les charges, on les préparera selon la force
des bêtes. Au moins quarante ânes et dix hommes nous sont
nécessaires pour soulager un peu nos bêtes et les nourrir aussi bien
que les hommes durant un mois. Les Khontalis nous ont presque
promis la moitié ; mais les Lobis fourniront-ils l’autre moitié ? Se
mettre en marche en étant bien prêt à tout événement est chose
difficile. Nous nous en apercevons une fois de plus, et Rachmed ne
laisse pas de manifester confidentiellement quelques craintes au
sujet du Doungane et des Lobis.
Nous organisons le retour de nos trois Sibériens. Ils retourneront
à Kouldja avec nos collections, nos lettres, et le consul russe les
expédiera à Paris par la Russie. Nous leurs donnons des chameaux
pour transporter les ballots à Kourla, où ils achèteront un arba, car
leur intention est de revenir par la route impériale d’Ouroumtsi en
contournant les Monts Célestes. On les munit de provisions et de
munitions. Nous aurions bien voulu en garder au moins un avec
nous ; mais l’un, Borodine, était marié ; l’autre, Maltzeff, avait fait
cette route afin de rassembler une petite somme destinée à célébrer
ses noces : une fiancée l’attendait à la maison. Quant à notre
préparateur, Kouznetzoff, que nous avions engagé à Tioumen, il ne
nous aurait pas été aussi utile que n’importe lequel des deux autres,
étant plus jeune et impropre aux durs travaux. Mais, comme
préparateur naturaliste, il a toujours fait preuve de la plus grande
conscience et de beaucoup de bonne volonté. Tout ce qu’il fait est
bien fait ; il a du soin, de l’ordre, de la patience. Nous ne saurions
trop le louer et le remercier. Il est prêt pour prendre part à une
exploration quelconque.
Nous prions la municipalité de nous fournir, à un prix qu’on
débattra, des hommes et des ânes qui porteront une partie de nos
provisions jusqu’aux environs du Kizil Sou en suivant la route de
Bogalik. Cette demande est faite le 12 ; on nous apportera la
réponse le lendemain après avoir tenu conseil.
Le 13 novembre, dans la matinée, nous voyons une troupe
s’approcher de notre camp. Presque tout le village est là. Khotanlis
et Lobis sont présents. Ils s’arrêtent au bord de l’aire qui figure notre
domaine momentané, et un grand gaillard à barbiche menue que
nous n’avons pas encore vu prend la parole et s’explique avec
Rachmed qui reçoit. Les Dounganes s’efforcent de comprendre.
L’orateur, nous dit-on, est le chef le plus considérable des Lobis. En
peu de mots, il expose que « l’on ne nous donnera ni hommes, ni
ânes, parce que le froid est trop rigoureux dans la montagne ; que la
parcourir en cette saison, c’est y chercher la mort », etc.
Rachmed insiste doucement ; il rappelle « le bien que nous avons
fait au pays, l’argent que nous lui laissons, les prix élevés que nous
avons payé chaque chose dans le but d’aider aux pauvres vendeurs.
Et les promesses qu’on nous faisait hier encore. Comment advient-il
que l’on ne veuille plus les tenir ? Avons-nous dit que nous ne
payerions pas les services qu’on nous offrait ? L’accord semblait
régner entre nous : d’où vient ce changement ? » etc.
Entre temps nous apprenons que des ordres secrets seraient
venus de Kourla. Les chefs lobis auraient reçu défense de nous
aider, et, comme ils ont demandé l’aide des Chinois contre les
Khotanlis, ils seraient décidés à obéir et à obstructionner…
Le chef lobi devient arrogant et il s’écrie : « Par Jupiter ! si tu veux
des ânes, tu les payeras deux fois leur valeur, et moi je ne t’en
vendrai point. Quant à des hommes pour vous servir, il n’en sortira
pas un du pays. Nous ne vous devons rien, nous ne vous payons
pas l’impôt, nous le payons aux Chinois. Non, nous ne vous devons
rien. Vous ne nous faites pas peur, nous avons le nombre ; nous
sommes des braves, vous ne nous faites pas peur… »
Comme il disait ces mots, Rachmed, qui voyait, qui sentait la
nécessité d’agir, emploie des arguments ad hominem, et il rosse ce
grand orateur. Les siens veulent le défendre, nous les repoussons
en les menaçant de nos armes et nous gardons à notre disposition
le chef des rebelles. Nous annonçons que nous ne le lâcherons que
contre les dix-huit ânes et les cinq hommes qui constituent le
contingent que les Lobis doivent fournir.
Les Khotanlis interviennent alors, ils servent de médiateurs entre
les deux partis, implorent pour le chef, de qui la tête est très basse,
et nous demandent de la patience ; ils promettent de tout arranger.
On entend des clameurs de femmes sur les toits et dans la
brousse, les chiens aboient, les ânes braient ; c’est un bruit
d’émeute.
Cependant le chef en notre pouvoir est consolé avec une tasse
de thé et du sucre. Timour l’engage à revenir à de bons sentiments,
attendu qu’il a tout à gagner à nous obéir et que nous ne lui
rendrons certainement pas la liberté avant que nous soyons assurés
de son concours.
Le chef fait demander l’un des siens et lui donne des ordres :
« Qu’on leur donne ce qu’ils réclament. » Ce messager retourne à
l’assemblée, tenue à distance, devant le palais d’un chef ayant une
femme de Lob, quoiqu’il soit originaire de Khotan. Et immédiatement
des ambassadeurs viennent nous trouver. Ils demandent la libération
du roi. Mais nous refusons, il nous faut des garanties. Ils s’éloignent,
tiennent de nouveau conseil et reviennent en chœur. Les barbes
blanches jurent qu’on nous donnera autant d’ânes, de guides, de
chasseurs que nous désirons. Mais ils ne dépasseront pas le pays
des Kalmouks du Tsaïdam.
« Nous ne pouvons pas vous montrer les ânes, disent-ils, le
temps de les rassembler nous ayant manqué, mais voici les cinq
Lobis qui vous accompagneront. » On les fait sortir de la foule, on
les met sur une ligne et l’on nous prie de les examiner. Et ce sont
des affirmations par la barbe, par Jupiter : tous les dieux sont
invoqués. La foule approuve, gesticule, élève la voix, et tout autour
de nous ce ne sont que gens souriant, agitant les bras avec des
gestes suppliants, montrant des dents affables, et renforçant à
propos par des exclamations les raisons de celui qui parle.
« Laissez aller le chef, disent-ils : c’est un brave homme, il n’a
pas de mauvaises intentions. »
Le grand chef, rendu à la liberté, ne tarde pas à venir nous faire
ses adieux, et, le nez légèrement enflé, il renouvelle les promesses
déjà faites, et jure qu’il a donné des ordres et qu’ils seront exécutés.
Après de longues politesses, il monte à cheval et part.
Le 16 novembre au soir, les charges sont préparées, nous
sommes parés, comme disent les marins. Nous emportons même
sept cents petites bottes de foin afin de soutenir les forces de nos
chevaux, condamnés à mourir les premiers.
Nous avons tenu compte des probabilités, des certitudes de
mort, dans nos calculs, pour établir le nombre des rations à
emporter ; il est proportionné au nombre des bêtes de somme dont
nous disposons pour le transport, mais les charges « doivent »
diminuer en même temps que les bêtes mourront, de telle façon que
les survivantes n’aient pas une surcharge au moment où leurs forces
seront moindres. L’expérience nous permet de fixer à peu près à
l’avance ce qu’il faut pour nourrir les quatorze hommes de notre
armée régulière durant cinq et, à la rigueur, six mois.
La vue de ces sacs pleins, de ces coffres bourrés inspire
confiance à Rachmed. « Qu’Allah nous aide, dit-il, et tout ira bien. »
Pourtant nous n’irions pas bien loin, au dire des indigènes, car
des chameaux ne pourraient passer par l’Altyn Tagh en suivant la
route de l’Anglais Carey. Et selon le « Petit Homme », Prjevalski
aurait été du même avis. Il nous tarde d’aller voir les obstacles, aussi
le départ est fixé irrévocablement au 17 novembre. En avant !
Le 17 novembre, le chargement des bêtes s’opère avec un
brouhaha de parlement le jour d’une interpellation. Toute la
population est présente. Il y a les femmes, les amis, les enfants, les
parents des partants et les curieux : c’est dire qu’il ne manque
personne. Les hommes, au premier rang, regardent, bavardent ; les
femmes plus loin jacassent ; quelques fillettes hardies se glissent
parmi les petits garçons. Ce monde n’est pas attiré que par le
spectacle du départ. Il est là aussi pour la même raison que la nuée
de moineaux qui s’est assemblée sur les saules près du camp. Les
moineaux pépient gaiement parce qu’ils savent que dans un instant
ils s’abattront sur le camp abandonné et picoreront les grains d’orge
qu’ils voient bien. Les badauds en feront autant, et, s’ils n’étaient
contenus par la crainte, ils se précipiteraient sur les boîtes vides, ils
s’arracheraient les chiffons de toile ; déjà ils se disputent des riens
qu’ils ont pu ramasser. Un enfant à pu saisir une boîte à conserves ;
il veut la porter à la maison, et il fuit à toutes jambes, poursuivi par
ses camarades.
Enfin la caravane est prête et nous partons. Le soleil luit. Les
chefs, à cheval, nous accompagnent. Ils iront avec nous jusqu’au
camp, à quelques kilomètres de Tcharkalik, la première étape étant
toujours très courte. Celle-ci finit au seuil du désert, de l’autre côté
de la petite rivière qui fait l’oasis et où nous boirons encore une fois
de bonne eau. Une bonne eau est pour nous autres la plus
délicieuse des boissons. En avons-nous bu de l’eau saumâtre !
Quarante minutes de cheval suffisent pour sortir de l’oasis et
arriver au désert qui guette le voyageur. En quittant la selle pour
nous installer sur le feutre où les chefs nous ont offert le « coup de
l’étrier », nous jetons un regard au Gobi : il nous sourit avec des
mirages de beaux lacs. Nous savons à quoi nous en tenir, nous
savons ce que valent de telles promesses. Au sud-est on devine les
montagnes dans la brume. Elles nous attendent.
Avant le coucher du soleil, les anciens nous font leurs adieux. Le
chef rossé est du nombre, et il n’est pas le moins cordial ; nous lui
faisons un beau cadeau. Les autres reçoivent aussi des souvenirs.
« Qu’Allah vous accorde un bon voyage ! disent-ils ; qu’il fasse
que votre santé soit toujours bonne et que vous rentriez sains et
saufs auprès des vôtres qui habitent si loin de nous !
« Nous sommes pauvres et nous n’avons pu vous être aussi
agréables que notre cœur le souhaitait. Veuillez nous pardonner.
Qu’Allah vous protège ! Qu’Allah vous protège ! »
Nous leur serrons les mains, nous les remercions. Nous
regrettons qu’il y ait eu un petit malentendu ; ils n’avaient jamais vu
de gens de notre race et ils étaient défiants. Nous espérons qu’ils
recevront désormais les nôtres à cœur ouvert, qu’ils ne garderont
pas de nous un mauvais souvenir, et qu’ils nous considéreront
comme des amis.
« Oui, nous sommes amis, nous sommes amis, répètent-ils en
nous serrant les mains. Qu’Allah vous protège ! »
Après quoi ils échangent des recommandations avec les
chasseurs et les chercheurs d’or, qui sont décidés à nous suivre.
« Veille sur mon père ; aie soin de mon bétail ; fais prendre patience
à ma femme : donne-lui du blé à crédit, je te payerai au retour.
Porte-toi bien ! Qu’Allah vous protège ! » etc. Puis ils s’embrassent,
ceux du même sang sur la bouche ; les autres pressent la main de
leurs aînés qui leur déposent un baiser sur le front. Une barbe
blanche récite ensuite une fatiha à haute voix, et, la prière terminée,
tous portent les mains à la barbe en criant : « Allah est grand ! Allah
est grand ! » ; alors les uns s’en vont, les autres restent et vaquent
immédiatement à leurs occupations.
La femme de Timour, petite brune alerte, est restée près de son
mari. Elle coud des sacs agilement, tandis que son petit garçon, de
quatre ans environ, tout de peau de mouton habillé, figure sale, nez
épaté, et roulant les petits yeux noirs et vifs de son père, s’amuse à
frapper contre les coffres en chantant : « Il n’y a de Dieu qu’Allah ! »
Puis, le soleil se couchant, nos trois Russes se décident à quitter
leurs compagnons de route. Après échange d’embrassades et de
souhaits, ils retournent à notre camp du matin, où ils ont laissé leurs
bagages à l’abandon.
Nous espérons que les lettres qu’ils emportent seront en Europe
dans trois mois environ. On s’endort après avoir bavardé de l’avenir.
Tous nous sommes tombés d’accord que jusqu’à ce jour nous avons
pleinement réussi dans tout ce que nous avons entrepris.
18 novembre. — Le minimum de la nuit n’a été que de − 9
degrés, mais ces neuf degrés suffisent pour geler la rivière et nous
allons lui emprunter sa glace. Aujourd’hui nous ne trouverons pas
d’eau potable au camp du soir, nous emportons des sacs de
glaçons. Dorénavant nous n’aurons pas d’autre boisson.
Nous sommes dans le désert pétré et nu. A notre droite, une
masse sombre se dessine mal sous la gaze d’un léger brouillard, et
le vieil Abdoullah dit : « C’est l’Altyn Tagh », les montagnes d’Or qui
ne se sont pas encore montrées depuis que nous sommes auprès
d’elles. Elles semblent hautes, mais on ne distingue aucun détail ;
aucune cime n’est visible. « De l’autre côté, ajoute Abdoullah Ousta,
commence le pays des vents de glace. Vous aurez froid, très froid
dans ce pays-là. »
Notre troupe est silencieuse. Nos hommes ne bavardent pas
gaiement comme d’habitude ; chacun fouette machinalement son
cheval, le regard fixe. Les lendemains de séparation sont toujours
semblables, surtout s’ils coïncident avec un départ vers l’inconnu :
on n’est pas encore en selle, ni au physique ni au moral, d’où des
rêveries.
Nous nous rapprochons des tertres de sables semés à notre
gauche, l’avant-garde du Gobi ; c’est là que nous camperons, paraît-
il. La steppe est aride et nous la quittons.
Soudain voilà nos ânes, nos moutons, — car nous emmenons un
troupeau de moutons, vivres qui se transportent eux-mêmes, — ils
sont chassés par de souples marcheurs vêtus de bure blanche, et
au soleil ce spectacle est un joli Guillaumet. Nous passons du sable
à des takirs de fine argile, puis nous retournons au sable, et
péniblement nous gravissons et descendons les monticules formés
par des émiettements de la montagne et des balayures de la plaine.
Abdoullah Ousta s’arrête, descend de cheval et dit : « Je vais
chercher par ici. » Dans le fond des vasques de sable apparaissent
à la surface comme des moisissures. C’est du sel qui indique le
voisinage de l’humidité, et, plus loin, le vieux guide tend le doigt vers
un petit trou : « On creusera là. » En effet le niveau de l’eau est à
une faible profondeur. Les ânes déchargés, les âniers saisissent
leurs pioches, et une fontaine est créée ; un trou se remplit d’eau
salpêtrée. On donne à boire aux bêtes, on les rationne.
Nous préparons un peu de thé, que nous buvons en attendant la
glace chargée sur les chameaux. Il n’est pas très bon, mais nous
refaisons l’apprentissage du désert. Je l’ai souvent observé : chaque
fois qu’on reprend le large, il y a des malades dans la caravane.
Aujourd’hui quatre ou cinq déclarent être brisés, et cependant l’étape
a été courte, et on l’a faite par un temps superbe. C’est ce qu’on
pourrait appeler le mal de mer du désert, comparable au malaise
qu’éprouvent certains marins pendant les premiers jours de
traversée.
Cette place s’appelle Yandachkak ; on y trouve beaucoup de
ioulgoun (tamarix) ; aussi notre campement bien illuminé me rappelle
certain campement de l’Oust-Ourt où le saksaoul abondait.
Un chant s’élève. C’est Tokta, notre poète, qui gratte son allah-
rabôb. Sa voix est très pure. Le chant est d’une grande tristesse ; il
est charmant dans ce paysage, il semble inspiré par le sable, par le
trou où l’on puise une eau salée, par la stérilité de la terre. C’est d’un
homme qui s’avoue vaincu par la nature ; c’est une vraie plainte de
captif se demandant s’il pourra s’échapper de la solitude menaçante
où il est pris. Les Israélites devaient chanter leurs psaumes sur un
air semblable lorsqu’ils se reposaient de leurs travaux d’esclaves, le
soir, sur les quais de Babylone, ou bien lorsqu’ils s’exposaient à la
brise, accroupis sur le toit des maisons à Samarcande, du temps de
Salmanazar.
19 novembre. — Au réveil, la première nouvelle est que les
chameaux manquent. Les hommes partent dans toutes les
directions. Habitués à boire copieusement chaque jour, ils sont sans
doute retournés à la rivière près de laquelle nous campions la veille.
Le sable, retenu par des tamarix, forme un petit pic ; je grimpe en
haut afin de voir si nos chameaux sont retrouvés. Je ne tarde pas à
voir dans le désert des cavaliers qui les ramènent. Sauf eux, rien. A
ma droite, des vagues de sable, des touffes de ioulgoun ; à ma
gauche, les montagnes jaillissent de la brume ; au-dessous de moi
s’étend la plaine nue, pierreuse. Deux fois mon œil circule autour de
l’horizon sans voir trace de vie. Il y a vraiment de quoi s’étonner des
feux du camp, du bruit des voix, du gargarisme des chameaux.
Pourquoi des êtres ici ? On ne peut que passer en de tels endroits,
et la seule habitation qui convienne, le seul abri qu’il faut dresser est
une légère tente de toile. On l’abat, et l’on se sauve plus loin.
Ayant marché pendant six heures presque droit sur l’est, nous
nous arrêtons dans une vallée où bruit le Djahan Saï, qui porte aussi
le nom de Kountchi Kan, un grand chef de Lob. Il serait venu
autrefois du Tsaïdam avec des troupeaux. Ayant découvert cette
rivière en chassant, il la trouva belle et vint en habiter les bords avec
sa famille. « Cela est arrivé il y a des années, des années », dit
Abdoullah Ousta.
Nous allons camper à Tchoukour Saï. En chemin nous
rencontrons des saksaouls ; nos hommes s’empressent d’en
emporter quelques fagots.
Ils savent que nul bois dans ces régions ne produit plus de
chaleur que le saksaoul. Ces arbustes avaient leurs graines, mais
mauvaises, malheureusement. Ils sont en état de décrépitude et
disparaissent, ne pouvant plus se propager.
Notre camp est dans le désert, au delà du Tchoukour Saï, gorge
profonde où l’on ne trouve plus une goutte d’eau. Demain nous
séjournerons à cette place. Nous envoyons les bêtes paître dans la
montagne près de l’eau : des chasseurs les accompagnent avec des
vivres ; ils ne reviendront que le lendemain soir. Il est indispensable
d’entreprendre le passage du Koum Davane et du Tach Davane
avec des bêtes bien portantes.
La journée du 21 novembre est consacrée au repos ; la nuit n’a
pas été froide, − 2 degrés avec une très légère brise nord-ouest.
Dans la journée + 10 degrés, température très agréable, due à un air
moins sec.
Superbe journée, employée à des réparations, à des nettoyages
divers. Tout le monde est gai, sauf le chamelier doungane, qui a
posé son bivouac à distance du nôtre. Il boude.
Son serviteur Niaz nous annonce que l’humeur du maître est plus
insupportable que jamais. Le Doungane se plaint d’avoir été trompé,
et il répète sans cesse : « On m’a donné de belles paroles, où
allons-nous ? Je le vois clairement, la route est mauvaise. Qui
pourrait dire où nous allons ? Est-ce là un chemin de marchands ?
Ah oui ! on m’a mis dans un sac ! »
« Oui, dit Niaz, je ne puis plus vivre à ses côtés. Il va comme un
chien à qui l’on a mis la corde au cou, mais c’est un chien méchant,
il me montre sans cesse les dents. » Aussi ce pauvre garçon se
plaît-il auprès du feu de nos hommes, où il est toujours accueilli par
une tasse de thé.
Demain la journée sera fatigante. Une gorge étroite nous attend
où les chameaux ne pourront peut-être pas passer.
Le 22 novembre, à trois quarts d’heure du camp, après une petite
passe, la première mais non pas la dernière après Tcharkalik, nous
descendons de plus de cent mètres dans un cañon. Il est dirigé vers
le sud et aboutit au pied du Koum Davane, la Passe de Sable.
Les traces de bêtes fauves sont nombreuses : loups, renards,
gazelles, errent dans ces solitudes. Une troupe de beaux animaux
aux cornes recourbées nous regardent du haut des crêtes, lorsque
nous descendons de cheval. Ils se proposaient sans doute d’aller
boire à la source, dont les abords sont piétinés et où les traces
fraîches sont nombreuses. Notre vue leur donne à réfléchir, ils vont
d’un pas lent. Henri d’Orléans les tire et voilà une superbe
dégringolade de toute la bande : elle fuit hardiment vers le côté
opposé de la gorge et en gravit les pentes avec une vélocité
prodigieuse. Notre tireur les poursuit si loin que, la nuit venue, il
manque à l’appel. On court à sa recherche, car on craint un accident
et finalement on le retrouve non loin du camp, arrêté sur une plate-
forme de rochers où il a glissé. Il lui est impossible d’en descendre,
impossible de retourner en arrière.
Avec des cordes on le tire d’affaire et il rentre au camp, très
content d’avoir vu des koukou-iamane (Pseudo ovis Burhell), mais
regrettant bien de n’avoir pu retrouver la bête qu’il avait blessée.
Voilà comment nous faisons connaissance avec la faune
particulière au Tibet. C’est le commencement des chasses, des gens
perdus et retrouvés, mais c’est une occasion de constater que le
voyage lie vite les hommes, car des gens qui partagent depuis peu
notre fortune ont montré véritablement beaucoup de bonne volonté ;
il n’a pas été nécessaire de leur ordonner de parcourir la montagne
après une journée de fatigue. Ils étaient inquiets et ils sont partis tout
de suite à la recherche de Henri d’Orléans. En quelques jours ils
sont devenus « nôtres ».
C’est une joie pour moi de voir ces aventuriers assis sur le feutre,
buvant le thé, dans l’attitude d’hommes après un acte d’énergie. Les
cous nerveux laissent un peu pencher la tête, les poitrines nues se
montrent par la pelisse entr’ouverte, les torses solides sont posés
noblement sur les reins, les mains rudes tiennent les genoux. La
sueur sèche sur les fronts, les figures sont joyeuses. C’est le
commencement de la route, ils ne sont pas encore fatigués.
Je les remercie de ce qu’ils ont fait pour un de leurs maîtres, et
ils ne se répandent pas en protestations. Cela est de bon augure,
leur silence marquant qu’ils n’ont pas de pensées à déguiser.
Près de notre camp se voient les traces d’hommes et d’ânes.
Nous questionnons à ce sujet Abdoullah Ousta.
« Un parti de quatorze hommes, dit-il, est allé à la chasse du côté
de Bokalik depuis un mois environ. Dans le nombre se trouve deux
de mes fils.
— Le Kizil Sou est-il de ce côté ?
— Oui.
— Y es-tu allé ?
— Non. »
Décidément, lorsqu’on parle du Kizil Sou, on ne peut obtenir
aucun renseignement. Je remarque une gêne chez Abdoullah

Ebook Direct and Inverse Scattering For The Matrix Schrodinger Equation 1St Edition Tuncay Aktosun Ricardo Weder Online PDF All Chapter

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ebook Direct and Inverse Scattering For The Matrix Schrodinger Equation 1St Edition Tuncay Aktosun Ricardo Weder Online PDF All Chapter

Uploaded by

Copyright:

Available Formats

Direct and Inverse Scattering for the

Matrix Schrödinger Equation 1st

Chess Explained The c3 Sicilian 1st Edition Sam Collins

Starting Out The c3 Sicilian 1st Edition John Emms

The Paper Issue 83 1st Edition Origamiusa

Structural Equation Modeling for Health and Medicine

Dilations Linear Matrix Inequalities the Matrix Cube

Direct Selling For Dummies, 2nd Edition Belinda

The American Revolution 1774 83 2nd Edition Daniel

Random Matrix Methods for Machine Learning 1st Edition

Direct and Inverse

More information about this series at http://www.springer.com/series/34

Direct and Inverse Scattering

ISSN 0066-5452 ISSN 2196-968X (electronic)

Mathematics Subject Classification: 34A55, 34L25, 34L40, 47A40, 81Uxx, 81U40

© Springer Nature Switzerland AG 2021

In this monograph, we present a complete and self-contained presentation of the

Arlington, TX, USA Tuncay Aktosun

The content of this monograph is based on research supported in part by CONACYT

2 The Matrix Schrödinger Equation and the Characterization

3.15 The Boundary Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237

4 Direct Scattering II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261

5 Inverse Scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339

6 Some Explicit Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485

Appendix A Mathematical Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545

A.10 Hardy Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564

List of Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 623

© Springer Nature Switzerland AG 2021 1

boundary condition, it is impossible to have the uniqueness in the inverse problem

There are also other relevant applications of non-Dirichlet boundary conditions

1.2 Notes on the Bibliography

of the Borg–Marchenko theorem. In their monograph [219] Pöschel and Trubowitz

2.1 Outline of the Chapter

© Springer Nature Switzerland AG 2021 19

2.2 The Matrix Schrödinger Equation on the Half Line

− ψ  (k, x) +V (x) ψ (k, x) = k2 ψ (k, x), x ∈ R+ . (2.2.1)

V (x) = V (x)† , x ∈ R+ , (2.2.2)

eral self-adjoint boundary condition at x = 0. There are various equivalent formu-

Proposition 2.2.1. Given n × n matrices A1 , B1 , A2 , B2 , assume that

A†1 A1 + B†1 B1 > 0. (2.2.7)

Then, 1 = 2 if and only if there exists an invertible matrix T such that

D1 (Z1 , Z2 ) := −B†1 Z1 + A†1 Z2 , Z1 , Z2 ∈ Cn , (2.2.11)

D2 (Z1 , Z2 ) := −B†2 Z1 + A†2 Z2 , Z1 , Z2 ∈ Cn . (2.2.12)

Ker [D1 ] = Ker [D2 ], (2.2.13)

W, −B†1 Z1 + A†1 Z2  = 0, Z1 , Z2 ∈ Cn , (2.2.14)

where ·, · is the standard scalar product in Cn . In (2.2.14) choosing Z1 = −B1 Z

(A†1 A1 + B†1 B1 )W, Z = 0, Z ∈ Cn . (2.2.15)

Choosing Z as Z = (A†1 A1 + B†1 B1 )W, from (2.2.15) we get

(A†1 A1 + B†1 B1 )W, (A†1 A1 + B†1 B1 )W  = 0. (2.2.16)

E U = −B†2 Z1 + A†2 Z2 . (2.2.18)

Let us confirm that E U is well defined, i.e., it is independent of the particular

Equating the right-hand sides of (2.2.17) and (2.2.19) we obtain

which indicates that

(Z1 − Z̃1 , Z2 − Z̃2 ) ∈ Ker[D2 ]. (2.2.21)

Using (2.2.21) in (2.2.11) and (2.2.12) we get

− B†2 Z1 + A†2 Z2 = −B†2 Z̃1 + A†2 Z̃2 . (2.2.22)

E A†1 = A†2 . (2.2.25)

From (2.2.24) and (2.2.25) we see that (2.2.10) holds with T = E † .

f (k, x) = eikx [I + o(1)], x → +∞, (2.2.26)

J(k) := f (−k∗ , 0)† B − f  (−k∗ , 0)† A, k ∈ R, (2.2.27)

where the asterisk denotes complex conjugation. The redundant appearance of

Ψ (k, x) := f (−k, x) + f (k, x) S(k), k ∈ R. (2.2.29)

In Proposition 3.2.12 it is shown that Ψ (k, x) is a solution to (2.2.1) and it satis-

− ψ (k, x) +V (x) ψ (k, x) = k2 ψ (k, x), x ∈ R+ . (2.2.1)

W, −B†1 Z1 + A†1 Z2 = 0, Z1 , Z2 ∈ Cn , (2.2.14)

where ·, · is the standard scalar product in Cn . In (2.2.14) choosing Z1 = −B1 Z

(A†1 A1 + B†1 B1 )W, Z = 0, Z ∈ Cn . (2.2.15)

(A†1 A1 + B†1 B1 )W, (A†1 A1 + B†1 B1 )W = 0. (2.2.16)

J(k) := f (−k∗ , 0)† B − f (−k∗ , 0)† A, k ∈ R, (2.2.27)

− B† Y(0) + A† Y (0) = 0, (2.3.4)

−Y (x) + Q(x)Y (x) = k2 Y (x), x ∈ R \ {0},

ψ1 (0) = ψ2 (0), ψ1 (0) + ψ2 (0) = c ψ1 (0),

where we recall that the superscript T denotes the matrix transpose.

−Y (x) + Q(x)Y (x) = k2 Y (x), x ∈ R \ {0},

− B†1 Y (0+ ) − B†2 Y (0− ) + A†1 Y (0+ ) − A†2 Y (0− ) = 0. (2.4.5)

Y (0+ ) = Y (0− ), Y (0+ ) −Y (0− ) = Λ Y (0), (2.4.7)

− B†Ψ (k, 0) + A†Ψ (k, 0) = 0, k ∈ R. (2.5.9)