Textbook Generalized Locally Toeplitz Sequences Theory and Applications Volume Ii Carlo Garoni Ebook All Chapter PDF

Generalized Locally Toeplitz Sequences
Theory and Applications Volume II

Carlo Garoni
Visit to download the full and correct content document:
https://textbookfull.com/product/generalized-locally-toeplitz-sequences-theory-and-ap
plications-volume-ii-carlo-garoni/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Generalized Intuitionistic Multiplicative Fuzzy

Calculus Theory and Applications Shan Yu
https://textbookfull.com/product/generalized-intuitionistic-
multiplicative-fuzzy-calculus-theory-and-applications-shan-yu/
Combinatorics and Number Theory of Counting Sequences

Discrete Mathematics and Its Applications 1st Edition
Istvan Mezo
https://textbookfull.com/product/combinatorics-and-number-theory-
of-counting-sequences-discrete-mathematics-and-its-
applications-1st-edition-istvan-mezo/
Structural Dynamics Fundamentals and Advanced

Applications, Volume II: Volume II 1st Edition Alvar M.
Kabe
https://textbookfull.com/product/structural-dynamics-
fundamentals-and-advanced-applications-volume-ii-volume-ii-1st-
edition-alvar-m-kabe/
Map Projections: Theory and Applications Frederick

Pearson Ii
https://textbookfull.com/product/map-projections-theory-and-
applications-frederick-pearson-ii/
Recurrent Sequences Key Results Applications and
Problems Dorin Andrica
https://textbookfull.com/product/recurrent-sequences-key-results-
applications-and-problems-dorin-andrica/
Legal Theory and the Social Sciences Volume II

Maksymilian Del Mar (Editor)
https://textbookfull.com/product/legal-theory-and-the-social-
sciences-volume-ii-maksymilian-del-mar-editor/
Existence Theory for Generalized Newtonian Fluids 1st

Edition Breit
https://textbookfull.com/product/existence-theory-for-
generalized-newtonian-fluids-1st-edition-breit/
Generalized Fractional Calculus: New Advancements and

Applications George A. Anastassiou
https://textbookfull.com/product/generalized-fractional-calculus-
new-advancements-and-applications-george-a-anastassiou/
Computation of Generalized Matrix Inverses and

Applications 1st Edition Ivan Stanimirovi■
https://textbookfull.com/product/computation-of-generalized-
matrix-inverses-and-applications-1st-edition-ivan-stanimirovic/
Carlo Garoni · Stefano Serra-Capizzano
Generalized
Locally Toeplitz
Sequences:
Theory and
Applications
Volume II
Generalized Locally Toeplitz Sequences:
Theory and Applications
Carlo Garoni Stefano Serra-Capizzano
•
Generalized Locally Toeplitz

Sequences: Theory
and Applications
Volume II
123
Carlo Garoni Stefano Serra-Capizzano
Department of Science Department of Science
and High Technology and High Technology
University of Insubria University of Insubria
Como, Italy Como, Italy
ISBN 978-3-030-02232-7 ISBN 978-3-030-02233-4 (eBook)

https://doi.org/10.1007/978-3-030-02233-4
This book has been realized with the financial support of the Italian INdAM (Istituto Nazionale di Alta
Matematica) and the European “Marie-Curie Actions” Programme through the Grant PCOFUND-GA-
2012-600198.
Library of Congress Control Number: 2018958367
Mathematics Subject Classification (2010): 15B05, 65N06, 65N25, 65N30, 65N35, 47B06, 35P20,
15A18, 15A60, 15A69
© Springer Nature Switzerland AG 2018

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
Sequences of matrices with increasing size naturally arise in several contexts and
especially in the discretization of continuous problems, such as integral and dif-
ferential equations. The theory of generalized locally Toeplitz (GLT) sequences was
developed in order to compute/analyze the asymptotic spectral distribution of these
sequences of matrices, which in many cases turn out to be GLT sequences.
In the first volume [22], we presented the theory of univariate/unilevel GLT
sequences, which arise in the discretization of unidimensional integral and differ-
ential equations; this is the reason why the first volume addressed only unidi-
mensional applications. In this second volume, we present the theory of
multivariate/multilevel GLT sequences, which arise in the discretization of multi-
dimensional integral and differential equations. The focus here is accordingly on
multidimensional applications, especially partial differential equations (PDEs).
It is important to emphasize that the extension from the univariate case addressed
in [22] to the multivariate case addressed here, despite being fundamental for the
applications as it allows one to face concrete PDEs, is essentially a technical matter
whose purpose is to illustrate the appropriate generalization of ideas already pre-
sented in [22]. The fact that all the main “GLT ideas” have been covered in [22]
makes it an essential prerequisite to this book. In particular, apart from (almost)
obvious adaptations, several “multivariate proofs” are the same as their corre-
sponding “univariate versions” from [22]. We have therefore been tempted to omit
them here so as to shorten the book, but ultimately we did not opt for this solution
in order to help the reader gain familiarity with the multivariate language (espe-
cially the multi-index notation).
The book is conceptually divided into two parts. The first part (Chaps. 1–5)
covers the theory of multilevel GLT sequences, which is finally summarized in
Chap. 6. The second part (Chap. 7) is devoted to PDE applications.
The book is intended for use as a text for graduate or advanced undergraduate
courses. It should also be useful as a reference for researchers working in the fields
of linear and multilinear algebra, numerical analysis, and matrix analysis. Given its
analytic spirit, it could also be of interest to analysts, particularly those working in
the fields of measure and operator theory.
v
vi Preface
As already pointed out, the first volume [22] is an essential prerequisite to this
second volume. It also provides detailed motivations to the theory of GLT
sequences [22, pp. 1–3] which will not be repeated here for the sake of conciseness.
In addition to [22], a basic knowledge of multidimensional integro-differential
calculus (partial derivatives, multiple integrals, etc.) is required.
Assuming the reader possesses the necessary prerequisites, most of which, if not
already addressed in [22], will be tackled in Chap. 2, there exists a way of reading
this book that allows one to omit essentially all the mathematical details/
technicalities without losing the core. This is probably “the best way of reading” for
those who love practice more than theory, but it is also advisable for theorists, who
can recover the missing details afterward. It consists in reading carefully the
summary of the theory in Chap. 6 and the applications in Chap. 7.
To conclude, we wish to express our gratitude to Bruno Iannazzo, Carla Manni,
and Hendrik Speleers, who awakened the interest in the theory of GLT sequences
and ultimately inspired the writing of this book. We also wish to thank all
of our colleagues who have worked in the field of “Toeplitz matrices and spectral
distributions” and contributed to laying the foundations of the theory of GLT
sequences. We mention in particular Bernhard Beckermann, Albrecht Böttcher,
Fabio Di Benedetto, Marco Donatelli, Leonid Golinskii, Sergei Grudsky,
Arno Kuijlaars, Maya Neytcheva, Debora Sesana, Bernd Silbermann, Paolo Tilli,
Eugene Tyrtyshnikov, and Nickolai Zamarashkin. Finally, special thanks go to
those researchers who, possibly attracted by the first volume [22], decided to
enter the research field of GLT sequences. We mention in particular
Giovanni Barbarino from Scuola Normale Superiore (Pisa, Italy), Davide Bianchi
and Isabella Furci from University of Insubria (Como, Italy), Ali Dorostkar and
Sven-Erik Ekström from Uppsala University (Uppsala, Sweden), Mariarosa Mazza
and Ahmed Ratnani from the Max Planck Institute for Plasma Physics
(Munich, Germany). Several of their contributions will certainly appear in a future
edition of both volumes I and II.
Based on their research experience, the authors propose a reference textbook in

two volumes on the theory of generalized locally Toeplitz sequences and their
applications. The first volume focuses on the univariate version of the theory and
the related applications in the unidimensional setting, while this second volume,
which addresses the multivariate case, is mainly devoted to concrete PDE
applications.
Como, Italy Carlo Garoni

August 2018 Stefano Serra-Capizzano
Contents
1 Notes to the Reader . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2 Mathematical Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1 Notation and Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1.1 General Notation and Terminology . . . . . . . . . . . . . . . . . . 3
2.1.2 Multi-index Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.3 Multilevel Matrix-Sequences . . . . . . . . . . . . . . . . . . . . . . 12
2.2 Multivariate Trigonometric Polynomials . . . . . . . . . . . . . . . . . . . . 12
2.3 Multivariate Riemann-Integrable Functions . . . . . . . . . . . . . . . . . . 16
2.4 Matrix Norms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5 Tensor Products and Direct Sums . . . . . . . . . . . . . . . . . . . . . . . . 18
2.6 Singular Value and Eigenvalue Distribution of a Sequence
of Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 24
2.6.1 The Notion of Singular Value and Eigenvalue
Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 24
2.6.2 Clustering and Attraction . . . . . . . . . . . . . . . . . . . ...... 26
2.6.3 Zero-Distributed Sequences . . . . . . . . . . . . . . . . . ...... 27
2.6.4 Sparsely Unbounded and Sparsely Vanishing
Sequences of Matrices . . . . . . . . . . . . . . . . . . . . . ...... 28
2.6.5 Spectral Distribution of Sequences of Perturbed
Hermitian Matrices . . . . . . . . . . . . . . . . . . . . . . . ...... 30
2.7 Approximating Classes of Sequences . . . . . . . . . . . . . . . . ...... 31
2.7.1 Definition of a.c.s. and a.c.s. Topology . . . . . . . . . ...... 31
2.7.2 The a.c.s. Tools for Computing Singular Value
and Spectral Distributions . . . . . . . . . . . . . . . . . . . . . . . . 33
2.7.3 The a.c.s. Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.7.4 Some Criteria to Identify a.c.s. . . . . . . . . . . . . . . . . . . . . . 35
2.7.5 An Extension of the Concept of a.c.s. . . . . . . . . . . . . . . . . 36
vii
viii Contents
3 Multilevel Toeplitz Sequences . . . . . . . . . . . . . . . . . . . . . ......... 39

3.1 Multilevel Toeplitz Matrices and Multilevel Toeplitz
Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.2 Basic Properties of Multilevel Toeplitz Matrices . . . . . . . . . . . . . . 42
3.3 Schatten p-Norms of Multilevel Toeplitz Matrices . . . . . . . . . . . . 46
3.4 Multilevel Circulant Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.5 Singular Value and Spectral Distribution of Multilevel
Toeplitz Sequences: An a.c.s.-Based Proof . . . . . . . . . ......... 55
3.6 Extreme Eigenvalues of Hermitian Multilevel Toeplitz
Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ......... 58
4 Multilevel Locally Toeplitz Sequences . . . . . . . . . . . . . . . . . . . . . . . . 61
4.1 Multilevel LT Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.1.1 Definition of Multilevel LT Operator . . . . . . . . . . . . . . . . 62
4.1.2 Properties of the Multilevel LT Operator . . . . . . . . . . . . . . 67
4.2 Definition of Multilevel LT and sLT Sequences . . . . . . . . . . . . . . 72
4.3 Fundamental Examples of Multilevel LT Sequences . . . . . . . . . . . 72
4.3.1 Zero-Distributed Sequences . . . . . . . . . . . . . . . . . . . . . . . 73
4.3.2 Sequences of Multilevel Diagonal Sampling Matrices . . . . 73
4.3.3 Multilevel Toeplitz Sequences . . . . . . . . . . . . . . . . . . . . . 79
4.4 Singular Value and Spectral Distribution of a Finite Sum
of Multilevel LT Sequences . . . . . . . . . . . . . . . . . . . . . . . . .... 82
4.5 Algebraic Properties of Multilevel LT Sequences . . . . . . . . . .... 85
4.6 Characterizations of Multilevel LT Sequences . . . . . . . . . . . .... 86
5 Multilevel Generalized Locally Toeplitz Sequences . . . . . . . . . . .... 91
5.1 Equivalent Definitions of Multilevel GLT Sequences . . . . . . .... 91
5.2 Singular Value and Spectral Distribution of Multilevel GLT
Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.3 Approximation Results for Multilevel GLT Sequences . . . . . . . . . 95
5.3.1 Characterizations of Multilevel GLT Sequences . . . . . . . . . 99
5.3.2 Sequences of Multilevel Diagonal Sampling Matrices . . . . 100
5.4 The Multilevel GLT Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
5.5 Algebraic-Topological Definitions of Multilevel GLT
Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
6 Summary of the Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
7 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
7.1 Auxiliary Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
7.1.1 Multilevel GLT Preconditioning . . . . . . . . . . . . . . . . . . . . 121
7.1.2 Multilevel Arrow-Shaped Sampling Matrices . . . . . . . . . . . 122
7.2 Applications to PDE Discretizations: An Introduction . . . . . . . . . . 124
7.3 FD Discretization of Convection-Diffusion-Reaction PDEs . . . . . . 127
7.4 FE Discretization of Convection-Diffusion-Reaction PDEs . . . . . . 136
Contents ix
7.5 B-Spline IgA Collocation Discretization

of Convection-Diffusion-Reaction PDEs . . . . . . . . . . . . . . . . . . . . 146
7.6 Galerkin B-Spline IgA Discretization
of Convection-Diffusion-Reaction PDEs . . . . . . . . . . . . . . . . . . . . 160
7.7 Galerkin B-Spline IgA Discretization of Second-Order
Eigenvalue Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
8 Future Developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
About the Authors
Dr. Carlo Garoni graduated in Mathematics at the University of Insubria (Italy) in

2011 and received his Ph.D. in Mathematics at the same university in 2015. He has
pursued research at the Universities of Insubria and Rome “Tor Vergata,” and he
now has a Marie-Curie postdoctoral position at the USI University of Lugano
(Switzerland). He has published about 25 research papers in different areas of
Mathematics, most of which are connected with the theory of GLT sequences and
its applications.
Prof. Stefano Serra-Capizzano is a Full Professor of Numerical Analysis, Deputy

Rector of the University of Insubria (Italy), and a long-term Visiting Professor at
Uppsala University (Sweden). He has authored over 200 research papers in different
areas of Mathematics, with more than 90 collaborators all over the world, and he
has recently won a Prodi Chair Professorship in Nonlinear Analysis at Würzburg
University (Germany). He is the founder of the Ph.D. program “Mathematics of
Computation” at the University of Insubria’s Department of Science and High
Technology.
xi
Chapter 1
Notes to the Reader
The present book covers the multivariate version of the theory of Generalized Locally
Toeplitz (GLT) sequences, also known as the theory of multilevel GLT sequences. In
addition, the book presents some emblematic (multidimensional) applications of this
theory in the context of the numerical discretization of Partial Differential Equations
(PDEs).
The generalization of the theory of GLT sequences from the univariate case
addressed in [22] to the multivariate case addressed here is essentially a matter
of technicalities, which results in the technical nature of the present volume. We
therefore recommend that, before going into this book, the reader give a reading to
[22, pp. 1–3] in order to call to mind the motivations behind the theory of (unilevel
and multilevel) GLT sequences, which will not be repeated here for the sake of con-
ciseness. When reading [22, pp. 1–3] in a multidimensional perspective, the GLT
sequences and the Differential Equations (DEs) mentioned therein should be under-
stood as multilevel GLT sequences and PDEs, respectively.
After going through [22, pp. 1–3], we encourage the reader to try reading this
book according to the scheme suggested in the preface, which consists in reading
Chaps. 6 and 7 first, and then coming back to fill the gaps (if necessary or wanted).
When reading the present book, it is advisable that the reader have at hand the first
volume [22], for at least two reasons. First, [22] is cited many times throughout the
book. Secondly, several “multivariate proofs” from Chaps. 2–5 are essentially the
same as their corresponding “univariate versions” from [22], and we recommend that
the reader compare them with each other so as to learn the way in which the multilevel
language (especially, the multi-index notation) allows one to transfer many results
from the univariate to the multivariate case. Roughly speaking, this transfer process
is carried out through a sort of “automatic procedure” consisting in turning some
letters (n, i, j, x, θ, etc.) in boldface (n, i, j , x, θ , etc.). Finally, we remark that, as
highlighted in the preface, the first volume [22] is an essential prerequisite to this
second volume. In addition to [22], the other necessary prerequisite for reading this
book is a basic knowledge of multidimensional integro-differential calculus (partial
derivatives, multiple integrals, etc.).
© Springer Nature Switzerland AG 2018 1

C. Garoni and S. Serra-Capizzano, Generalized Locally Toeplitz Sequences:
Theory and Applications, https://doi.org/10.1007/978-3-030-02233-4_1
Chapter 2
Mathematical Background
This chapter collects the necessary preliminaries to develop the multivariate version
of the theory of GLT sequences. The reader is supposed to be familiar with the univari-
ate version of the theory [22] and to possess a basic knowledge of multidimensional
integro-differential calculus (partial derivatives, multiple integrals, etc.).
2.1 Notation and Terminology
For the reader’s convenience, we report in this section some of the most common
notations and terminologies that will be used throughout the book. Special attention
is devoted to the multi-index notation and the notion of multilevel matrix-sequences.
Together with the index at the end, this section can be used as a reference whenever
an unknown notation/terminology is encountered.
2.1.1 General Notation and Terminology
• The cardinality of a set S is denoted by #S.

• If S is a subset of a topological space, the closure of S is denoted by S.
• A permutation σ of the set {1, 2, . . . , n} is denoted by [σ (1), σ (2), . . . , σ (n)].
• Rm×n (resp., Cm×n ) is the space of real (resp., complex) m × n matrices.
• Om and Im denote, respectively, the m × m zero matrix and the m × m identity
matrix. Sometimes, when the size m can be inferred from the context, O and I
are used instead of Om and Im . The symbol O is also used to indicate rectangular
zero matrices whose sizes are clear from the context.
• If x is a vector and X is a matrix, x T and x∗ (resp., X T and X ∗ ) are the transpose
and the conjugate transpose of x (resp., X ).
• If x is a vector with m components x1 , . . . , xm , diag(x) is the diagonal matrix
whose diagonal entries are x1 , . . . , xm .
© Springer Nature Switzerland AG 2018 3
C. Garoni and S. Serra-Capizzano, Generalized Locally Toeplitz Sequences:
Theory and Applications, https://doi.org/10.1007/978-3-030-02233-4_2
4 2 Mathematical Background
• If x, y are vectors with m components, x · y denotes their scalar product.

• We use the abbreviations HPD, HPSD, SPD, SPSD for “Hermitian Positive Def-
inite”, “Hermitian Positive SemiDefinite”, “Symmetric Positive Definite”, “Sym-
metric Positive SemiDefinite”.
• If X, Y ∈ Cm×m , the notation X ≥ Y (resp., X > Y ) means that X, Y are Hermitian
and X − Y is HPSD (resp., HPD).
• If X, Y ∈ Cm×m , we denote by X ◦ Y the componentwise (or Hadamard) product
of X and Y : (X ◦ Y )i j = xi j yi j , i, j = 1, . . . , m.
• If X ∈ Cm×m , we denote by X † the Moore–Penrose pseudoinverse of X . For more
on the Moore–Penrose pseudoinverse, see [22, Sect. 2.4.2].
• If X ∈ Cm×m , we denote by Λ(X ) the spectrum of X .
• If X ∈ Cm×m , the singular values and eigenvalues of X are denoted by σ j (X ), j =
1, . . . , m, and λ j (X ), j = 1, . . . , m, respectively. The maximum and minimum
singular values are also denoted by σmax (X ) and σmin (X ). If the eigenvalues are
real, their maximum and minimum are also denoted by λmax (X ) and λmin (X ).
• If 1 ≤ p ≤ ∞, the symbol | · | p denotes both the p-norm of vectors and the asso-
ciated operator norm for matrices:
m 1/ p
|xi | p
, if 1 ≤ p < ∞,
|x| p = i=1 x ∈ Cm ,
maxi=1,...,m |xi |, if p = ∞,
|X x| p
|X | p = maxm , X ∈ Cm×m .
x∈C |x| p
x=0
The 2-norm | · |2 is also known as the spectral (or Euclidean) norm and it will be
preferably denoted by · . For more on p-norms, see [22, Sect. 2.4.1].
• Given X ∈ Cm×m and 1 ≤ p ≤ ∞, X p denotes the Schatten p-norm of X ,
which is defined as the p-norm of the vector (σ1 (X ), . . . , σm (X )) formed by the
singular values of X . The Schatten 1-norm is also known under the names of trace-
norm and nuclear norm. For more on Schatten p-norms, see [22, Sect. 2.4.3].
• (X ) and (X ) are, respectively, the real and the imaginary part of the square
matrix X :
X + X∗ X − X∗
(X ) = , (X ) = ,
2 2i
where i is the imaginary unit (i2 = −1). Note that (X ), (X ) are Hermitian and
X = (X ) + i (X ) for all square matrices X .
• If X ∈ Cm×m is diagonalizable and f : Λ(X ) → C, we denote by f (X ) the matrix
obtained by applying the function f to X . For more on matrix functions, see [22,
Sect. 2.4.6].
• We use the abbreviations FDs, FEs, IgA for “Finite Differences”, “Finite Ele-
ments”, “Isogeometric Analysis”.
• Given two sequences {ζn }n and {ξn }n , with ζn ≥ 0 and ξn > 0 for all n, the notation
ζn = O(ξn ) means that there exists a constant C, independent of n, such that
ζn ≤ Cξn for all n; and the notation ζn = o(ξn ) means that ζn /ξn → 0 as n → ∞.
2.1 Notation and Terminology 5
• Cc (C) (resp.,Cc (R)) is the space of complex-valued continuous functions defined

on C (resp., R) with bounded support. Moreover, for m ∈ N ∪ {∞}, Ccm (R) =
Cc (R) ∩ C m (R), where C m (R) is the space of functions F : R → C such that the
real and imaginary parts (F), (F) are of class C m over R in the classical sense.
• If wi : Di → C, i = 1, . . . , d, we define the tensor-product function w1 ⊗ · · · ⊗
wd : D1 × · · · × Dd → C as follows: for every (ξ1 , . . . , ξd ) ∈ D1 × · · · × Dd ,
(w1 ⊗ · · · ⊗ wd )(ξ1 , . . . , ξd ) = w1 (ξ1 ) · · · wd (ξd ).
• If f : D → E and g : E → F are arbitrary functions, the composite function

g ◦ f is preferably denoted by g( f ).
• If g : D → C, we set g ∞ = supξ ∈D |g(ξ )|. If we need/want to specify the
domain D, we write g ∞,D instead of g ∞ . Clearly, g ∞ < ∞ if and only
if g is bounded over its domain.
• If g : D → C is continuous over D, with D ⊆ Ck for some k, we denote by ωg (·)
the modulus of continuity of g,
ωg (δ) = sup |g(x) − g(y)|, δ > 0.

x,y∈D
|x−y|∞ ≤δ
If we need/want to specify D, we will say that ωg (·) is the modulus of continuity

of g over D.
• χ E is the characteristic (or indicator) function of the set E,

1, if ξ ∈ E,
χ E (ξ ) =
0, otherwise.
• μk denotes the Lebesgue measure in Rk . Throughout this book, unless otherwise

stated, all the terminology coming from measure theory (such as “measurable set”,
“measurable function”, “almost everywhere (a.e.)”, etc.) is always referred to the
Lebesgue measure.
• If E 1 , . . . , E d ⊆ R are measurable sets and f : E 1 × · · · × E d → C, we say that
f is d-separable if there exist d measurable functions f i : E i → C, i = 1, . . . , d,
such that f = f 1 ⊗ · · · ⊗ f d . In this case, f 1 ⊗ · · · ⊗ f d is called a factorization
of f . Note that any d-separable function is measurable. Throughout this book,
“separable function” is an abbreviation of “d-separable function”.
• If f : D ⊆ Rk → C is measurable, we denote by ER( f ) its essential range. For
more on the essential range, see [22, Sect. 2.2.1].
• If D is any measurable subset of some Rk , we set
M D = { f : D → C : f is measurable},

L (D) = f ∈ M D :
p
|f| < ∞ ,
p
1 ≤ p < ∞,
D

L ∞ (D) = f ∈ M D : ess sup D | f | < ∞ .
If D is the special domain [0, 1]d × [−π, π ]d , we preferably use the notation Md
instead of M D :
Md = {κ : [0, 1]d × [−π, π ]d → C : κ is measurable}.
If f ∈ L p (D) and the domain D is clear from the context, we write f Lp instead
of f L p (D) to indicate the L p -norm of f , which is defined as

( D | f | p )1/ p , if 1 ≤ p < ∞,
f Lp =
ess sup D | f |, if p = ∞.
For more on L p spaces, see [22, Sect. 2.2.2].

• If D ⊆ Rk is a measurable set with 0 < μk (D) < ∞, we denote by dmeasure the
pseudometric on M D defined in [22, Eq. (2.14)], which induces on M D the topol-
ogy τmeasure of convergence in measure. For more details on this topic, see [22,
Sect. 2.3.2].
• If f ∈ L 1 ([−π, π ]d ), the Fourier coefficients of f are denoted by f k and are
defined as follows:

1
fk = f (θ) e−ik·θ dθ , k ∈ Zd . (2.1)
(2π )d [−π,π]d
• We use a notation borrowed from probability theory to indicate sets. For example,
if f, g : D ⊆ Rk → C, then
{ f = 1} = {x ∈ D : f (x) = 1},
{0 ≤ f ≤ 1, g > 2} = {x ∈ D : 0 ≤ f (x) ≤ 1, g(x) > 2},
μk { f > 0, g < 0} is the measure of the set {x ∈ D : f (x) > 0, g(x) < 0},
χ{ f =0} is the characteristic function of the set where f vanishes,
...
• A functional φ is any function defined on some vector space (such as, for example,
Cc (C) or Cc (R)) and taking values in C.
• If K is either R or C and g : D ⊂ Rk → K is a measurable function defined on a
set D with 0 < μk (D) < ∞, we denote by φg the functional

1
φg : Cc (K) → C, φg (F) = F(g(x))dx. (2.2)
μk (D) D
• A sequence of matrices is a sequence of the form {An }n , where n varies in some

infinite subset of N and An is a square matrix of size dn such that dn → ∞ as
n → ∞. Throughout this book, unless otherwise specified, the size of the nth
matrix of a sequence of matrices is always assumed to be dn .
2.1.2 Multi-index Notation
Throughout this book, we will systematically use the multi-index notation. When
discretizing a linear PDE over a d-dimensional domain ⊂ Rd by means of a
linear numerical method, the actual computation of the numerical solution reduces to
solving a linear system whose coefficient matrix usually possesses a d-level structure
(see Example 2.5 below). As we shall see later on, especially in Chap. 7, the multi-
index notation is a powerful tool that allows one to give a compact expression of this
matrix by treating the dimensionality parameter d as any other parameter involved
in the considered discretization. In this way, the dependence of the matrix structure
on d is highlighted and a compact presentation is made possible.
A multi-index i of size d, also called a d-index, is simply a (row) vector in Zd ;
its components are denoted by i 1 , . . . , i d .
• 0, 1, 2, . . . are the vectors of all zeros, all ones, all twos, . . . (their size will be
clear from the context).
• For any d-index m, we set N (m) = dj=1 m j and we write m → ∞ to indicate
that min(m) → ∞. The notation N (α) = dj=1 α j will be actually used for any
vector α with d components and not only for d-indices.
• Let {an }n∈Nd be a family of numbers parameterized by a d-index n. The limit of
an as n → ∞ is defined, as in the case of a traditional sequence {an }n∈N , in the
following way: lim n→∞ an = a if and only if for every neighborhood U of a there
exists N such that an ∈ U for n ≥ N. Moreover, we define
lim sup an = lim sup am , lim inf an = lim inf am .

n→∞ n→∞ m≥n n→∞ n→∞ m≥n
• If h, k are d-indices, h ≤ k means that h r ≤ kr for all r = 1, . . . , d, while h ≤ k

means that h r > kr for at least one r ∈ {1, . . . , d}.
• If h, k are d-indices such that h ≤ k, the multi-index range h, . . . , k (or, more pre-
cisely, the d-index range h, . . . , k) is the set of cardinality N (k − h + 1) given by
{ j ∈ Zd : h ≤ j ≤ k}. We assume for this set the standard lexicographic ordering:

... [ ( j1 , . . . , jd ) ] jd =h d ,...,kd jd−1 =h d−1 ,...,kd−1
... . (2.3)
j1 =h 1 ,...,k1
For instance, in the case d = 2 the ordering is
(h 1 , h 2 ), (h 1 , h 2 + 1), . . . , (h 1 , k2 ),
(h 1 + 1, h 2 ), (h 1 + 1, h 2 + 1), . . . , (h 1 + 1, k2 ),
. . . . . . . . . , (k1 , h 2 ), (k1 , h 2 + 1), . . . , (k1 , k2 ).
• When a d-index j varies over a d-index range h, . . . , k (this is often written as

j = h, . . . , k), it is understood that j varies from h to k following the specific
ordering (2.3). For instance, if m ∈ Nd and we write x = [x i ]m
i=1 , then x is a vector
of size N (m) whose components x i , i = 1, . . . , m, are ordered in accordance with

(2.3): the first component is x1 = x(1,...,1,1) , the second component is x(1,...,1,2) , and
so on until the last component, which is x m = x(m 1 ,...,m d ) . Similarly, if
X = [x i j ]m
i, j =1 , (2.4)
then X is an N (m) × N (m) matrix whose components are indexed by a pair of

d-indices i, j , both varying from 1 to m according to the lexicographic ordering
(2.3).
• If h, k are d-indices such that h ≤ k, the notation kj =h indicates the summation
over all j in h, . . . , k.
• If i, j are d-indices, i j means that i precedes (or equals) j in the lexicographic
ordering (which is a total ordering on Zd ). Moreover, we define

i, if i j ,
i∧ j= (2.5)
j , if i j .
Note that i ∧ j is the minimum among i and j with respect to the lexicographic
ordering. In the case where i and j are 1-indices (i.e., normal scalar indices), it is
clear that i ∧ j = min(i, j).
• Operations involving d-indices that have no meaning in the vector space Zd
must always be interpreted in the componentwise sense. For instance, n p =
(n 1 p1 , . . . , n d pd ), αi/ j = (αi 1 /j1 , . . . , αi d /jd ) for all α ∈ C, i 2 = (i 12 , . . . , i d2 ),
max(i, j ) = (max(i 1 , j1 ), . . . , max(i d , jd )), i mod m = (i 1 mod m 1 , . . . , i d mod
m d ), etc.
• When a multi-index appears as subscript or superscript, we sometimes suppress
the brackets to simplify the notation. For instance, the component of the vector
x = [x i ]m i=1 corresponding to the d-index i is denoted by x i or x i 1 ,...,i d , and we
often avoid the heavy notation x(i1 ,...,id ) .
We provide below a few examples to help the reader become familiar with the multi-
index notation.
Example 2.1 Let h = 1 = (1, 1) and k = (4, 2). The multi-index range h, . . . , k
consists of N (k − h + 1) = N (k) = 8 elements which are sorted according to the
lexicographic ordering (2.3) as follows:
(1, 1), (1, 2), (2, 1), (2, 2), (3, 1), (3, 2), (4, 1), (4, 2).
k
Note that j =h j 2 = (60, 20).
Example 2.2 Let a : [0, 1]2 → C and n ∈ N2 . Set
i n
x= a .
n i=1
Then, x is the vector of size N (n) given by
⎡ ⎤
a( n11 , n12 )
⎢ 1 2 ⎥
⎢ a( n 1 , n 2 ) ⎥
⎢ ⎥
⎢ .. ⎥
⎢ . ⎥
⎢ ⎥
⎢ ⎥
⎢ a( n 1 , 1) ⎥
1
⎢ ⎥
⎢ 2 1 ⎥
⎢ a( n , n ) ⎥
⎢ 1 2 ⎥
⎢ 2 2 ⎥
⎢ a( , ) ⎥ ⎡ ⎤
⎢ n1 n2 ⎥ ⎡ ⎤ i
a( n11 , n12 )
⎢ .. ⎥
⎢ . ⎥ x1 ⎢ i ⎥
⎢ ⎥ ⎢x ⎥ ⎢ a( 1 , 2 ) ⎥
⎢ ⎥ ⎢ 2⎥ ⎢ n1 n2 ⎥
x=⎢
⎢ a( n 1 , 1) ⎥ = ⎢
⎥ xi 1 = ⎢ ⎥,
2
. ⎥, ⎢ .. ⎥ i 1 = 1, . . . , n 1 .
⎢ . ⎥ ⎣ .. ⎦ ⎢ . ⎥
⎢ .. ⎥ ⎣ ⎦
⎢ ⎥ xn 1
⎢ .. ⎥ i
a( n11 , 1)
⎢ ⎥
⎢ . ⎥
⎢ .. ⎥
⎢ ⎥
⎢ . ⎥
⎢ ⎥
⎢ a(1, 1 ) ⎥
⎢ n2 ⎥
⎢ ⎥
⎢ ⎥
⎢ a(1, n22 ) ⎥
⎢ ⎥
⎢ .. ⎥
⎣ . ⎦
a(1, 1)
n n
Moreover, x 2
= i=1 |x i |2 = i=1 |a(i/n)|2 .
Example 2.3 Consider the matrix

⎡ ⎤
4 4 0 0
⎢4 4 0 0⎥
A=⎢
⎣0
⎥. (2.6)
0 1 1⎦
0 0 1 1
Instead of indexing the entries of A in the standard way, i.e., by means of two
traditional scalar indices i, j = 1, . . . , 4, we can decide to index the entries of A by
means of two 2-indices (or bi-indices) i, j = 1, . . . , 2 = (1, 1), (1, 2), (2, 1), (2, 2).
This is possible because both the ranges 1, . . . , 4 and 1, . . . , 2 have 4 elements, and
4 is the size of A. The two writings A = [ai j ]i,4 j=1 and A = [a i j ]2i, j =1 correspond to
the two different indicizations. We have
a(1,1),(1,1) = 4, a(1,1),(1,2) = 4, a(1,1),(2,1) = 0, a(1,1),(2,2) = 0,

a(1,2),(1,1) = 4, a(1,2),(1,2) = 4, a(1,2),(2,1) = 0, a(1,2),(2,2) = 0,
a(2,1),(1,1) = 0, a(2,1),(1,2) = 0, a(2,1),(2,1) = 1, a(2,1),(2,2) = 1,
a(2,2),(1,1) = 0, a(2,2),(1,2) = 0, a(2,2),(2,1) = 1, a(2,2),(2,2) = 1.
The indicization of the entries of A with the bi-indices i, j reflects the fact that we
are thinking of A as a block matrix partitioned into 4 blocks as indicated in (2.6): for
all i, j = 1, . . . , 2, the entry ai j is the (i 2 , j2 ) entry of the (i 1 , j1 ) block of A. For
example,
a(2,1),(2,2) = entry (1, 2) in the (2, 2) block of A = 1.
We can therefore write

⎧
⎪
⎪ 4, if i 1 = j1 = 1,
⎨
0, if i 1 = 1 and j1 = 2,
ai j =
⎪
⎪ 0, if i 1 = 2 and j1 = 1,
⎩
1, if i 1 = j1 = 2.
Note that a i j depends only on the first components of the bi-indices i and j .
Example 2.4 (tensor products) Let X ∈ Cm 1 ×m 2 and Y ∈ C1 ×2 , and define the
block matrix
⎡ ⎤
x11 Y x12 Y · · · x1m 2 Y
⎢ x21 Y x22 Y · · · x2m 2 Y ⎥
⎢ ⎥ m ×m
Z = [xi j Y ] i=1,...,m 1 = ⎢ . .. .. ⎥ ∈ C 1 1 2 2 . (2.7)
j=1,...,m 2 ⎣ .. . . ⎦
xm 1 1 Y xm 1 2 Y · · · xm 1 m 2 Y
Using the identities
r = r/ss + r mod s, r/s = (r + 1)/s − 1,
which are satisfied for all integers r ≥ 0 and s ≥ 1, for every i = 1, . . . , m 1 1 and
j = 1, . . . , m 2 2 we can write
i = (i/1 − 1)1 + ((i − 1) mod 1 ) + 1,

j = ( j/2 − 1)2 + (( j − 1) mod 2 ) + 1.
The (i, j) entry of the matrix Z is then given by
z i j = xi/1 , j/2 y((i−1) mod 1 )+1,(( j−1) mod 2 )+1 .
It is clear that this expression is rather complicated. Now, suppose we decide to index
the entries of Z by two bi-indices i, j such that i = 1, . . . , n and j = 1, . . . , k, where
n = (m 1 , 1 ) and k = (m 2 , 2 ). This indicization, which is possible because
#{1, . . . , n} = N (n) = m 1 1 = number of rows of Z ,

#{1, . . . , k} = N (k) = m 2 2 = number of columns of Z ,
reflects the fact that we are thinking of Z as an m 1 × m 2 block matrix in which each
of the m 1 m 2 blocks is of size 1 × 2 . Actually, this is the natural way of thinking in
view of the block structure of Z ; see (2.7). With such indicization, for all i = 1, . . . , n
and j = 1, . . . , k we have
z i j = entry (i 2 , j2 ) in the (i 1 , j1 ) block of Z = xi1 j1 yi2 j2 .
We then see that z i j has a much simpler expression than z i j . To conclude, we remark
that the matrix Z is the so-called tensor (Kronecker) product of X and Y , and it is
usually denoted by X ⊗ Y ; we shall come back to tensor products in Sect. 2.5.
Example 2.5 (multilevel matrices) In many cases, it is convenient to partition matri-

ces into blocks, which are partitioned into smaller blocks, which are partitioned into
smaller blocks, and so on until a certain nesting level d is reached. Such matrices are
called multilevel matrices. More precisely, following Tyrtyshnikov [41, Sect. 6], we
say that a square matrix A of size N is a d-level matrix with level orders n 1 , n 2 , . . . , n d
if N = n 1 n 2 · · · n d and A is partitioned into n 21 square blocks of size N /n 1 , each of
which is partitioned into n 22 square blocks of size N /(n 1 n 2 ), each of which is parti-
tioned into n 23 square blocks of size N /(n 1 n 2 n 3 ), and so on until the last n 2d square
blocks of size N /(n 1 n 2 · · · n d ) = 1, which are scalars. In formulas,
A = [Ai1 j1 ]in11, j1 =1 , Ai1 j1 ∈ Cñ 1 ×ñ 1 , ñ 1 = N

n1
;
Ai1 j1 = [Ai1 j1 ; i2 j2 ]in22, j2 =1 , Ai1 j1 ; i2 j2 ∈ Cñ 2 ×ñ 2 , ñ 2 = N
n1 n2
;
..
.
Ai1 j1 ; ...; id−1 jd−1 = [Ai1 j1 ; ...; id jd ]indd, jd =1 , Ai1 j1 ; ...; id jd ∈ Cñ d ×ñ d , ñ d = N
n 1 ···n d
= 1.
The level orders n 1 , n 2 , . . . , n d and the order N are also referred to as partial orders
and total order, respectively. Indexing the entries of a d-level matrix A by two tradi-
tional scalar indices i, j = 1, . . . , N is a nightmare. On the contrary, A admits a natu-
ral indicization by means of two d-indices i, j = 1, . . . , n, where n = (n 1 , . . . , n d ).
Indeed, A can be written in the form (2.4) as follows:
A = [A i j ]ni, j =1 ,
where A i j = Ai1 j1 ; ...; id jd for i, j = 1, . . . , n. We remark that, as we shall see in

Chap. 7, when a linear PDE over a d-dimensional hyperrectangle is discretized by
a linear numerical method, the resulting discretization matrix is normally a d-level
matrix with level orders n 1 , n 2 , . . . , n d and total order N (n) = n 1 n 2 · · · n d , where
n i is the discretization parameter in the ith direction.
Example 2.6 (matrix computations with multi-indices) Let n ∈ Nd and let
A = [a i j ]ni, j =1 , B = [bi j ]ni, j =1 , x = [x j ]nj=1 , y = [y j ]nj=1 .

It is not difficult to see that the following properties hold.

• A∗ = [a j i ]ni, j =1 .
• α A + β B = [αai j + βbi j ]ni, j =1 for all α, β ∈ C.

• AB = [ nk=1 a i k bk j ]ni, j =1 .
n
• Ax = [ j =1 a i j x j ]ni=1 .

• x∗ Ay = ni, j =1 a i j x i y j .
These properties show that the usual matrix computation rules remain formally the
same when passing from standard scalar indices to multi-indices. We invite the reader
to prove them as an exercise.
2.1.3 Multilevel Matrix-Sequences
We recall from Sect. 2.1.1 that a sequence of matrices is a sequence of the form
{An }n , where n varies in some infinite subset of N and An is a square matrix of size
dn → ∞. A d-level matrix-sequence is a special sequence of matrices of the form
{A n }n , where:
• n varies in some infinite subset of N;
• n = n(n) is a d-index with positive components which depends on n and satisfies
n → ∞ as n → ∞;
• A n is a square matrix of size N (n).
Recall from Sect. 2.1.2 that n → ∞ means min(n) → ∞.
The name “d-level matrix-sequence” is due to the fact that, in practical applica-
tions, especially in the context of PDE discretizations, each matrix A n of a d-level
matrix-sequence {A n }n is normally a d-level matrix with level orders (n 1 , . . . , n d ) =
n; see Example 2.5. Throughout this book, we often use the abbreviation “matrix-
sequence” for both “d-level matrix-sequence” and “multilevel matrix-sequence”.
2.2 Multivariate Trigonometric Polynomials
A d-variate trigonometric polynomial is a finite linear combination of the d-variate

Fourier frequencies
eik·θ = ei(k1 θ1 +...+kd θd ) , k ∈ Zd ,
that is, a function of the form

N
f (θ) = f k eik·θ , f −N , . . . , f N ∈ C, N ∈ Nd . (2.8)
k=−N
2.2 Multivariate Trigonometric Polynomials 13
Note that, as a consequence of the orthogonality relations

i·θ −ik·θ (2π )d , if k = ,
e e dθ = (2.9)
[−π,π]d 0, if k = ,
the numbers f −N , . . . , f N appearing in (2.8) are the (only possible nonzero) Fourier
coefficients of f according to the definition (2.1). Note also that a 1-variate (or
univariate) trigonometric polynomial is just a trigonometric polynomial in the clas-
sical sense. In what follows, we say that a d-variate trigonometric polynomial is
d-separable (or simply separable) if it is a d-separable function from Rd to C accord-
ing to the definition in Sect. 2.1.1.
Lemma 2.1 Let f : Rd → C be a separable d-variate trigonometric polynomial

and let f = f 1 ⊗ · · · ⊗ f d be a factorization of f . If f is not identically 0, then
f 1 , . . . , f d are (univariate) trigonometric polynomials.
Proof Since f 2 , . . . , f d are not identically 0, there exists (ϑ2 , . . . , ϑd ) such that
f 2 (ϑ2 ) · · · f d (ϑd ) = 0. The definition of d-variate trigonometric polynomials implies
that θ1 → f (θ1 , ϑ2 , . . . , ϑd ) = f 1 (θ1 ) f 2 (ϑ2 ) · · · f d (ϑd ) is a (univariate) trigonomet-
ric polynomial, and this means that f 1 is a trigonometric polynomial. With the same
argument, one can show that f 2 , . . . , f d are trigonometric polynomials as well.
Corollary 2.1 Let f : Rd → C be a separable d-variate trigonometric polynomial.

Then, there exist (univariate) trigonometric polynomials f 1 , . . . , f d : R → C such
that f = f 1 ⊗ · · · ⊗ f d .
The next lemma shows that the set of zeros of every non-trivial d-variate trigono-
metric polynomial has zero measure.
Lemma 2.2 Let f : Rd → C be a d-variate trigonometric polynomial with at least

one nonzero Fourier coefficient. Then μd { f = 0} = 0.
Proof The proof proceeds by induction on d. For d = 1, the result has already been
proved in [22, solution of Exercise 6.2, pp. 286–287]. Supposed > 1 and assume
N
that the lemma holds for dimensions up to d − 1. Let f (θ) = k=−N f k eik·θ and
set Z = { f = 0}. By Fubini’s theorem,

μd (Z ) = χ Z (θ1 , . . . , θd )dθ1 . . . dθd = dθ2 . . . dθd χ Z (θ1 , . . . , θd )dθ1
R Rd−1 R
d

= dθ2 . . . dθd dθ1 , (2.10)
Rd−1 Z θ2 ,...,θd
where, for each fixed (θ2 , . . . , θd ) ∈ Rd−1 , the set Z θ2 ,...,θd is defined by
Z θ2 ,...,θd = {θ1 ∈ R : f (θ1 , θ2 , . . . , θd ) = 0}.

Write

N1
f (θ ) = pk1 (θ2 , . . . , θd ) eik1 θ1 , (2.11)
k1 =−N1
where p−N1 , . . . , p N1 are (d − 1)-variate trigonometric polynomials given by
(N2
,...,Nd )
pk1 (θ2 , . . . , θd ) = f k ei(k2 ,...,kd )·(θ2 ,...,θd ) , k1 = −N1 , . . . , N1 .
(k2 ,...,kd )=−(N2 ,...,Nd )
Let
A = {(θ2 , . . . , θd ) ∈ Rd−1 : pk1 (θ2 , . . . , θd ) = 0 for all k1 = −N1 , . . . , N1 }.
Since not all the Fourier coefficients f k are equal to 0, at least one of the polyno-
mials pk1 has at least a nonzero Fourier coefficient. Thus, by induction hypothesis,
μd−1 (A) = 0. Moreover, by (2.11),
• if (θ2 , . . . , θd ) ∈ A then Z θ2 ,...,θd = R;
• if (θ2 , . . . , θd ) ∈ Ac = Rd−1 \A then there exists an index k1 ∈ {−N1 , . . . , N1 }
such that pk1 (θ2 , . . . , θd ) = 0, and so μ1 (Z θ2 ,...,θd ) = 0 by induction hypothesis.
Going back to (2.10), we see that

μd (Z ) = dθ2 . . . dθd dθ1 + dθ2 . . . dθd dθ1 = 0,
A Z θ2 ,...,θd Ac Z θ2 ,...,θd
and the proof is complete.

The next two lemmas are the multivariate versions of [22, Lemmas 2.7 and 2.8].
They show how it is possible to approximate an L 1 (resp., a measurable) function
by means of standard (resp., weighted) multivariate trigonometric polynomials. In
what follows, for any D ⊆ Rk we denote by Cc (D) the space of continuous func-
tions f : D → C such that support the supp( f ) = {x ∈ D : f (x) = 0} = { f = 0}
is compact. We recall that, if D is measurable (so that it makes sense to talk about
L p (D)), then the space Cc (D) is dense in L p (D) for all 1 ≤ p < ∞ [22, p. 13].
Lemma 2.3 Let f ∈ L 1 ([−π, π ]d ). Then, there exists a sequence of d-variate
trigonometric polynomials { pm }m such that pm ∞ ≤ ess sup[−π,π]d | f | for all m
and pm → f a.e. and in L 1 ([−π, π ]d ).
Proof The proof follows the same pattern as [22, proof of Lemma 2.7]. It suffices to
show that, for each ε > 0, there exists a d-variate trigonometric polynomial pε such
that
pε ∞ ≤ ess sup[−π,π]d | f |, f − pε L 1 ≤ ε.
Indeed, this shows the existence of a sequence of d-variate trigonometric polynomials

{ pm }m such that pm ∞ ≤ ess sup[−π,π]d | f | for all m and pm → f in L 1 ([−π, π ]d ).
2.2 Multivariate Trigonometric Polynomials 15
Recalling that the L 1 convergence of a sequence implies the a.e. convergence of

an appropriate subsequence, passing to an appropriate subsequence of { pm }m (if
necessary), we may assume that pm → f a.e.
Let ε > 0. By [22, Theorem 2.2] in combination with the dominated conver-
gence theorem and the density of Cc ((−π, π )d ) in L 1 ((−π, π )d ), there exists
f ε ∈ Cc ((−π, π )d ) such that
fε ∞ ≤ ess sup[−π,π]d | f |, f − fε L1 < ε. (2.12)
The function f ε is continuous on [−π, π ]d and satisfies f ε (θ ) = 0 for every

θ ∈ ∂([−π, π ]d ). Thus, by the multivariate version of Fejér’s theorem, which is
proved essentially in the same way as the classical (univariate) Fejér theorem [28,
Theorem 3.1], there exists a d-variate trigonometric polynomial pε such that
pε ∞ ≤ fε ∞, f ε − pε ∞ < ε. (2.13)
By combining (2.12) and (2.13), we arrive at
pε ∞ ≤ ess sup[−π,π]d | f |, f − pε L1 ≤ ε(1 + (2π )d ),
which proves the thesis.
Lemma 2.4 Let κ : [0, 1]d × [−π, π ]d → C be a measurable function. Then, there
exists a sequence {κm }m such that κm : [0, 1]d × [−π, π ]d → C is a function of the
form

Nm
κm (x, θ ) = a (m)
j (x) e
i j ·θ
, a (m)
j ∈ C ∞ ([0, 1]d ), N m ∈ Nd , (2.14)
j =−N m
and κm → κ a.e.
Proof The proof is essentially the same as [22, proof of Lemma 2.8]. The function
κ̃m = κχ{|κ|≤1/m} belongs to L ∞ ([0, 1]d × [−π, π ]d ) and converges to κ in measure.
Indeed, κ̃m → κ pointwise over [0, 1]d × [−π, π ]d , and the pointwise (a.e.) conver-
gence on a set of finite measure implies the convergence in measure [22, Lemma 2.4].
By [22, Lemma 2.2], the space generated by the trigonometric monomials

e2πi·x ei j ·θ = ei(2π1 x1 +...+2πd xd + j1 θ1 +...+ jd θd ) : , j ∈ Zd
is dense in L 1 ([0, 1]d × [−π, π ]d ), so we can choose a function κm belonging to this

space such that κm − κ̃m L 1 ≤ 1/m. Note that κm is a function of the form (2.14).
Moreover, for each ε > 0, using Chebyshev’s inequality [22, Eq. (2.4)] we obtain
μ2d {|κm − κ| > ε} ≤ μ2d ({|κm − κ̃m | > ε/2} ∪ {|κ̃m − κ| > ε/2})
≤ μ2d {|κm − κ̃m | > ε/2} + μ2d {|κ̃m − κ| > ε/2}
κm − κ̃m L 1
≤ + μ2d {|κ̃m − κ| > ε/2},
(ε/2)
which converges to 0 as m → ∞. Hence, κm → κ in measure. Since the convergence

in measure on a set of finite measure implies the existence of a subsequence that
converges a.e. [22, Lemma 2.4], passing to a subsequence of {κm }m (if necessary)
we may assume that κm → κ a.e.
2.3 Multivariate Riemann-Integrable Functions
A function a : [0, 1]d → C is said to be Riemann-integrable if its real and imaginary

parts (a), (a) : [0, 1]d → R are Riemann-integrable in the classical sense. Recall
that any Riemann-integrable function is bounded by definition. We report below a
list of properties possessed by Riemann-integrable functions that will be used in this
book, either explicitly or implicitly.
• If α, β ∈ C and a, b : [0, 1]d → C are Riemann-integrable, then αa + βb is
Riemann-integrable.
• If a, b : [0, 1]d → C are Riemann-integrable, then ab is Riemann-integrable.
• If a : [0, 1]d → C is Riemann-integrable and F : C → C is continuous, then
F(a) : [0, 1]d → C is Riemann-integrable.
• If a : [0, 1]d → C is Riemann-integrable, then a belongs to L ∞ ([0, 1]d ) and its
Lebesgue and Riemann integrals over [0, 1]d coincide.
• If a : [0, 1]d → C is bounded, then a is Riemann-integrable if and only if a is
continuous a.e.
Note that the last two properties imply the first three. The proof of the second-to-
last property can be found in [30, pp. 73–74], while the last property is Lebesgue’s
characterization theorem of Riemann-integrable functions [30, p. 104]. Note that
the proofs in [30] are made for the case d = 1 only, but the generalization to the
case d > 1 is straightforward. A further property of Riemann-integrable functions
that will be used in this book is stated and proved in the next lemma, which is the
multivariate version of [22, Lemma 2.9].
Lemma 2.5 Let a : [0, 1]d → R be Riemann-integrable. For each n ∈ Nd , consider

the partition of (0, 1]d given by the d-dimensional hyperrectangles

i −1 i i1 − 1 i1 id − 1 id
I i,n = , = , × ··· × , , i = 1, . . . , n,
n n n1 n1 nd nd
and let
2.3 Multivariate Riemann-Integrable Functions 17

a i,n ∈ inf a(x), sup a(x) , i = 1, . . . , n.
x∈I i,n x∈I i,n
Then

n
a i,n χ Ii,n → a a.e. in [0, 1]d (2.15)
i=1
and
1
n
lim a i,n = a(x)dx. (2.16)
n→∞ N (n) i=1 [0,1]d
Proof The proof is essentially the same as [22, proof of Lemma 2.9]. Fix ε > 0 and let
x ∈ (0, 1]d be a continuity point of a. Then, there is a δ > 0 such that |a(y) − a(x)| ≤
ε whenever y ∈ [0, 1]d and |y − x|∞ ≤ δ. Take n ≥ (1/δ)1 and call I k,n the unique
hyperrectangle of the partition (0, 1]d = ni=1 I i,n containing x. For y ∈ I k,n , we
have y ∈ [0, 1]d and |y − x|∞ ≤ δ, hence |a(y) − a(x)| ≤ ε. It follows that
! n !
! !
! a χ (x) − a(x) ! = |ak,n − a(x)|
! i,n I i,n !
i=1
≤ max a(x) − inf a(y), sup a(y) − a(x) ≤ ε.

y∈I k,n y∈I k,n

As a consequence, ni=1 a i,n χ Ii,n (x) → a(x) whenever x is a continuity point of a in
(0, 1]d . This implies (2.15), because a is Riemann-integrable and hence continuous
a.e. in [0, 1]d . Since
! n ! "
! ! 1
n n
! a i,n χ Ii,n !! ≤ a < ∞, a i,n = a i,n χ Ii,n ,
! ∞
N (n) i=1
i=1 [0,1]d i=1
(2.16) follows from (2.15) and from the dominated convergence theorem.
2.4 Matrix Norms
For the reader’s convenience, we report from [22, Sects. 2.4.1 and 2.4.3] several
matrix-norm inequalities that we shall use throughout the book. First, we recall the
expressions of the p-norms for p = 1, ∞:

m
m
|X |1 = max |xi j |, |X |∞ = max |xi j |, X ∈ Cm×m .
j=1,...,m i=1,...,m
i=1 j=1
An important bound for X in terms of |X |1 and |X |∞ (and hence in terms of the

components of X ) is the following:
#
X ≤ |X |1 |X |∞ ≤ max(|X |1 , |X |∞ ), X ∈ Cm×m . (2.17)
Given 1 ≤ p, q ≤ ∞, we say that p, q are conjugate exponents if 1p + q1 = 1 (it is

1
understood that ∞ = 0). The following Hölder-type inequality holds for the Schatten
norms:
XY 1 ≤ X p Y q, X, Y ∈ Cm×m . (2.18)
We will also need the following trace-norm inequalities:
X 1 ≤ rank(X ) X ≤ m X , X ∈ Cm×m , (2.19)

m
X 1 ≤ |xi j |, X ∈ Cm×m . (2.20)
i, j=1
2.5 Tensor Products and Direct Sums
If X, Y are matrices of any dimension, say X ∈ Cm 1 ×m 2 and Y ∈ C1 ×2 , the tensor
(Kronecker) product of X and Y is the m 1 1 × m 2 2 matrix defined by
⎡ ⎤
x11 Y · · · x1m 2 Y
⎢ .. ⎥ ,
X ⊗ Y = xi j Y i=1,...,m 1 = ⎣ ... . ⎦
j=1,...,m 2
xm 1 1 Y · · · xm 1 m 2 Y
and the direct sum of X and Y is the (m 1 + 1 ) × (m 2 + 2 ) matrix defined by

X O
X ⊕ Y = diag(X, Y ) = .
O Y
Tensor products and direct sums possess a lot of nice algebraic properties.
(i) Associativity: for all matrices X , Y , Z ,
(X ⊗ Y ) ⊗ Z = X ⊗ (Y ⊗ Z ),
(X ⊕ Y ) ⊕ Z = X ⊕ (Y ⊕ Z ).
We can therefore omit parentheses in expressions like X 1 ⊗ X 2 ⊗ · · · ⊗ X d or

X1 ⊕ X2 ⊕ · · · ⊕ Xd .
2.5 Tensor Products and Direct Sums 19
(ii) If X 1 , X 2 can be multiplied and Y1 , Y2 can be multiplied, then
(X 1 ⊗ Y1 )(X 2 ⊗ Y2 ) = (X 1 X 2 ) ⊗ (Y1 Y2 ),
(X 1 ⊕ Y1 )(X 2 ⊕ Y2 ) = (X 1 X 2 ) ⊕ (Y1 Y2 ).
(iii) For all matrices X , Y ,
(X ⊗ Y )∗ = X ∗ ⊗ Y ∗ , (X ⊗ Y )T = X T ⊗ Y T
(X ⊕ Y )∗ = X ∗ ⊕ Y ∗ , (X ⊕ Y )T = X T ⊕ Y T .
(iv) Bilinearity (of tensor products): for each fixed matrix X , the application
Y → X ⊗ Y
is linear on C1 ×2 for all 1 , 2 ∈ N; for each fixed matrix Y , the application
X → X ⊗ Y
is linear on Cm 1 ×m 2 for all m 1 , m 2 ∈ N.

From (i)–(iv), a lot of other interesting properties follow. For example, if X, Y are
invertible, then X ⊗ Y is invertible, with inverse X −1 ⊗ Y −1 . If X, Y are normal
(resp., Hermitian, symmetric, unitary) then X ⊗ Y is also normal (resp., Hermitian,
symmetric, unitary). If X ∈ Cm×m and Y ∈ C× , the eigenvalues and singular values
of X ⊗ Y are
{λi (X )λ j (Y ) : i = 1, . . . , m, j = 1, . . . , }, (2.21)

{σi (X )σ j (Y ) : i = 1, . . . , m, j = 1, . . . , }; (2.22)
and the eigenvalues and singular values of X ⊕ Y are
{λi (X ) : i = 1, . . . , m} ∪ {λ j (Y ) : j = 1, . . . , }, (2.23)

{σi (X ) : i = 1, . . . , m} ∪ {σ j (Y ) : j = 1, . . . , }; (2.24)
see [22, Exercise 2.5]. In particular, for all X ∈ Cm×m , Y ∈ C× , and 1 ≤ p ≤ ∞,
we have
X ⊗Y p = X p Y p, (2.25)

! ! p p 1/ p
( X p + Y p ) , if 1 ≤ p < ∞,
X ⊕Y = ! ( X p , Y p )! p =
p
max( X ∞ , Y ∞ ), if p = ∞,
(2.26)
and
rank(X ⊗ Y ) = rank(X )rank(Y ), (2.27)

rank(X ⊕ Y ) = rank(X ) + rank(Y ). (2.28)
In addition to the properties considered so far, we need to highlight two further

properties of tensor products, which are very important for the “multidimensional
purposes” of this book. The first one is the multi-index formula for tensor products:
if we have d matrices X k ∈ Cm k ×m k , k = 1, . . . , d, then
(X 1 ⊗ X 2 ⊗ · · · ⊗ X d )i j = (X 1 )i1 j1 (X 2 )i2 j2 · · · (X d )id jd , i, j = 1, . . . , m,

(2.29)
where m = (m 1 , m 2 , . . . , m d ). Note that (2.29) can be rewritten in the form (2.4) as
follows:
m
X 1 ⊗ X 2 ⊗ · · · ⊗ X d = (X 1 )i1 j1 (X 2 )i2 j2 · · · (X d )id jd i, j =1 . (2.30)
Note also that X 1 ⊗ X 2 ⊗ · · · ⊗ X d is one of the most eminent example of a d-level

matrix with level orders m 1 , m 2 , . . . , m d and total order N (m); see Example 2.5 for
the corresponding definitions. Equation (2.29) is of fundamental importance and,
indeed, it motivates the introduction of multi-indices to index the entries of a matrix
formed by a sum of tensor products. To better understand the importance of (2.29),
try to write the (i, j) entry of X 1 ⊗ X 2 ⊗ · · · ⊗ X d as a function of two scalar indices
i, j = 1, . . . , N (m); see also Example 2.4.
The second property is a natural upper bound for the rank of the difference of two
tensor products formed by d factors. More precisely, suppose we have 2d matrices
X 1 , . . . , X d , Y1 , . . . , Yd , with X i , Yi ∈ Cm i ×m i for all i = 1, . . . , d; then,
d
rank(X i − Yi )
rank(X 1 ⊗ · · · ⊗ X d − Y1 ⊗ · · · ⊗ Yd ) ≤ N (m) , (2.31)
i=1
mi
where m = (m 1 , . . . , m d ). This is true because
rank(X 1 ⊗ · · · ⊗ X d − Y1 ⊗ · · · ⊗ Yd )
d "
= rank Y1 ⊗ · · · ⊗ Yi−1 ⊗ (X i − Yi ) ⊗ X i+1 ⊗ · · · ⊗ X d
i=1

d
≤ rank(Y1 ⊗ · · · ⊗ Yi−1 ⊗ (X i − Yi ) ⊗ X i+1 ⊗ · · · ⊗ X d )
i=1

d
= rank(Y1 ⊗ · · · ⊗ Yi−1 )rank(X i − Yi )rank(X i+1 ⊗ · · · ⊗ X d )
i=1
Another random document with
no related content on Scribd:
Bathe the eye frequently with warm milk and water, and apply,
every night at bedtime, a warm white-bread poultice.
No medicine is required; but if the child be gross, keep him for a
few days from meat, and let him live on bread and milk and
farinaceous puddings.
243. If a child have large bowels, what would you recommend as
likely to reduce their size?
It ought to be borne in mind that the bowels of a child are larger in
proportion than those of an adult. But, if they be actually larger than
they ought to be, let them be well rubbed for a quarter of an hour at a
time night and morning, with soap liniment, and then apply a broad
flannel belt. “A broad flannel belt worn night and day, firm but not
tight, is very serviceable.”[244] The child ought to be prevented from
drinking as much as he has been in the habit of doing; let him be
encouraged to exercise himself well in the open air; and let strict
regard be paid to his diet.
244. What are the best aperients for a child?
If it be actually necessary to give him opening medicine, one or
two teaspoonfuls of syrup of senna, repeated if necessary, in four
hours, will generally answer the purpose; or, for a change, one or two
teaspoonfuls of castor oil may be substituted. Lenitive electuary
(compound confection of senna) is another excellent aperient for the
young, it being mild in its operation, and pleasant to take; a child
fancying it is nothing more than jam, and which it much resembles
both in appearance and in taste. The dose is half or one teaspoonful
early in the morning occasionally. Senna is an admirable aperient for
a child, and is a safe one, which is more than can be said of many
others. It is worthy of note that “the taste of senna may be concealed
by sweetening the infusion,[245] adding milk, and drinking as ordinary
tea, which, when thus prepared, it much resembles.”[246] Honey, too,
is a nice aperient for a child—a teaspoonful ought to be given either
by itself, or spread on a slice of bread.
Some mothers are in the habit of giving their children jalap
gingerbread. I do not approve of it, as jalap is a drastic griping
purgative; besides, jalap is very nasty to take—nothing will make it
palatable.
Fluid magnesia—solution of the bicarbonate of magnesia—is a
good aperient for a child; and, as it has very little taste, is readily
given, more especially if made palatable by the addition either of a
little syrup or of brown sugar. The advantages which it has over the
old solid form are, that it is colorless and nearly tasteless, and never
forms concretions in the bowels, as the solid magnesia, if persevered
in for any length of time, sometimes does. A child two or three years
old may take one or two tablespoonfuls of the fluid, either by itself or
in his food, repeating it every four hours until the bowels be opened.
When the child is old enough to drink the draught off immediately,
the addition of one or two teaspoonfuls of lemon-juice, to each dose
of the fluid magnesia, makes a pleasant effervescing draught, and
increases its efficacy as an aperient.
Bran-bread[247] and treacle will frequently open the bowels; and as
treacle is wholesome, it may be substituted for butter when the
bowels are inclined to be costive. A roasted apple, eaten with raw
sugar, is another excellent mild aperient for a child. Milk gruel—that
is to say, milk thickened with oatmeal—forms an excellent food for
him, and often keeps his bowels regular, and thus (which is a very
important consideration) supersedes the necessity of giving him an
aperient. An orange (taking care he does not eat the peel or the pulp),
or a fig after dinner, or a few Muscatel raisins, will frequently
regulate the bowels.
Stewed prunes is another admirable remedy for the costiveness of
a child. The manner of stewing them is as follows: Put a pound of
prunes in a brown jar, add two tablespoonfuls of raw sugar, then
cover the prunes and the sugar with cold water; place them in the
oven, and let them stew for four hours. A child should every morning
eat half a dozen or a dozen of them, until the bowels be relieved,
taking care that he does not swallow the stones.
A suppository is a mild and ready way of opening the bowels of a
child. When he is two or three years old and upwards, a candle
suppository is better than a soap suppository. The way of preparing
it is as follows: Cut a piece of dip-tallow candle—the length of three
inches—and insert it as you would a clyster pipe, about two inches up
the fundament, allowing the remaining inch to be in sight, and there
let the suppository remain until the bowels be opened.
245. What are the most frequent causes of Protrusion of the lower
bowel?
The too common and reprehensible practice of a parent
administering frequent aperients, especially calomel and jalap, to her
child. Another cause, is allowing him to remain for a quarter of an
hour or more at a time on his chair; this induces him to strain, and to
force the gut down.
246. What are the remedies?
If the protrusion of the bowel have been brought on by the abuse
of aperients, abstain for the future from giving them; but if medicine
be absolutely required, give the mildest—such as either syrup of
senna or castor oil—and the less of those the better.
If the external application of a purgative will have the desired
effect, it will, in such cases, be better than the internal
administration of aperients. Dr. Merriman’s Purgative Liniment[248]
is a good one for the purpose. Let the bowels be well rubbed every
night and morning, for five minutes at a time, with the liniment.
A wet compress to the bowels will frequently open them, and will
thus do away with the necessity of giving an aperient—a most
important consideration. Fold a napkin in six thicknesses, soak it in
cold water, and apply it to the bowels, over which put either a thin
covering or sheet of gutta-percha, or a piece of oiled silk; keep it in its
place with a broad flannel roller, and let it remain on the bowels for
three or four hours, or until they be opened.
Try what diet will do, as opening the bowels by a regulated diet is
far preferable to the giving of aperients. Let him have either bran-
bread, Robinson’s Patent Groats made into gruel with new milk, or
Du Barry’s Arabica Revalenta, or a slice of Huntley and Palmer’s
lump gingerbread. Let him eat stewed prunes, stewed rhubarb,
roasted apples, strawberries, raspberries, the inside of grapes and
gooseberries, figs, etc. Give him early every morning a draught of
cold water.
Let me, again, urge you not to give aperients in these cases, or in
any case, unless you are absolutely compelled. By following my
advice you will save yourself an immense deal of trouble, and your
child a long catalogue of misery. Again, I say, look well into the
matter, and whenever it be practicable, avoid purgatives.
Now, with regard to the best manner of returning the bowel, lay
the child upon the bed on his face and bowels, with his hips a little
raised; then smear lard on the forefinger of your right hand (taking
care that the nail be cut close) and gently with your forefinger press
the bowel into its proper place.
Remember, if the above methods be observed, you cannot do the
slightest injury to the bowel, and the sooner it be returned the better
it will be for the child; for, if the bowel be allowed to remain long
down, it may slough or mortify, and death may ensue. The nurse,
every time he has a motion, must see that the bowel does not come
down, and if it does, she ought instantly to return it. Moreover, the
nurse should be careful not to allow the child to remain on his chair
more than two or three minutes at a time.
Another excellent remedy for the protrusion of the lower bowel is
to use every morning a cold salt and water sitz-bath. There need not
be more than a depth of three inches of water in the bath; a small
handful of table salt should be dissolved in the water; a dash of warm
water in the winter time must be added to take off the extreme chill,
and the child ought not to be allowed to sit in the bath for more than
one minute, or while the mother can count a hundred, taking care
the while to throw either a square of flannel or a small shawl over his
shoulders. The sitz-bath ought to be continued for months, or until
the complaint be removed. I cannot speak in too high praise of these
baths.
247. Do you advise me, every spring and fall, to give my child
brimstone to purify and sweeten his blood, and as a preventive
medicine?
Certainly not: if you wish to take away his appetite and to weaken
and depress him, give him brimstone! Brimstone is not a remedy fit
for a child’s stomach. The principal use and value of brimstone is as
an external application in itch, and as an internal remedy, mixed
with other laxatives, in piles—piles being a complaint of adults. In
olden times poor unfortunate children were dosed every spring and
fall with brimstone and treacle, to sweeten their blood! Fortunately
for the present race, there is not so much of that folly practiced, but
still there is room for improvement.
To dose a healthy child with physic is the grossest absurdity. No,
the less physic a delicate child has the better it will be for him, but
physic to a healthy child is downright poison!—and brimstone of all
medicines! It is both weakening and depressing to the system, and by
opening the pores of the skin and by relaxing the bowels, is likely to
give cold, and thus to make a healthy a sickly child. Sweeten his
blood! It is more likely to weaken his blood, and thus to make his
blood impure! Blood is not made pure by drugs, but by Nature’s
medicine: by exercise, by pure air, by wholesome diet, by sleep in a
well-ventilated apartment, by regular and thorough ablution.
Brimstone a preventive medicine? Preventive medicine—and
brimstone especially in the guise of a preventive medicine—is “a
mockery, a delusion, and a snare.”
248. When a child is delicate, and his body, without any
assignable cause is gradually wasting away, and the stomach
rejects all food that is taken, what plan can be adopted likely to
support his strength, and thus, probably, be the means of saving his
life?
I have seen, in such a case, great benefit to arise from half a
teacupful of either strong mutton broth, or of strong beef-tea, used as
an enema, every four hours.[249]
It should be administered slowly, in order that it may remain in
the bowel. If the child be sinking, either a dessertspoonful of brandy,
or half a wineglassful of port wine ought to be added to each enema.
The above plan ought only to be adopted if there be no diarrhœa. If
there be diarrhœa, an enema must not be used. Then, provided there
be great wasting away, and extreme exhaustion, and other remedies
having failed, it would be advisable to give, by the mouth, raw beef
of the finest quality, which ought to be taken from the hip-bone, and
should be shredded very fine. All fat and skin must be carefully
removed. One or two teaspoonfuls (according to the age of the child)
ought to be given every four hours. The giving of raw meat to
children in exhaustive diseases, such as excessive long-standing
diarrhœa, was introduced into practice by a Russian physician, a
Professor Weisse, of St. Petersburg. It certainly is, in these cases, a
most valuable remedy, and has frequently been the means of
snatching such patients from the jaws of death. Children usually take
raw meat with avidity and with a relish.
249. If a child be naturally delicate, what plan would you
recommend to strengthen him?
I should advise strict attention to the rules above mentioned, and
change of air—more especially, if it be possible, to the coast. Change
of air, sometimes, upon a delicate child, acts like magic, and may
restore him to health when all other means have failed. If a girl be
delicate, “carry her off to the farm, there to undergo the discipline of
new milk, brown bread, early hours, no lessons, and romps in the
hay-field.”[250] This advice is, of course, equally applicable for a
delicate boy, as delicate boys and delicate girls ought to be treated
alike. Unfortunately, in these very enlightened days! there is too
great a distinction made in the respective management and
treatment of boys and girls.
The best medicines for a delicate child will be the wine of iron and
cod-liver oil. Give them combined in the manner I shall advise when
speaking of the treatment of Rickets.
In diseases of long standing, and that resist the usual remedies,
there is nothing like change of air. Hippocrates, the father of
medicine, says:
“In longis morbis solum mutare.”
(In tedious diseases to change the place of residence.)
A child who, in the winter, is always catching cold, whose life

during half of the year is one continued catarrh, who is in
consequence likely, if he grow up at all, to grow up a confirmed
invalid, ought, during the winter months, to seek another clime; and
if the parents can afford the expense, they should, at the beginning of
October, cause him to bend his steps to the south of Europe—
Mentone being as good a place as they could probably fix upon.
250. Do you approve of sea bathing for a delicate young child?
No: he is frequently so frightened by it that the alarm would do
him more harm than the bathing would do him good. The better plan
would be to have him every morning well sponged, especially his
back and loins, with sea water; and to have him as much as possible
carried on the beach, in order that he may inhale the sea breezes.
When he be older, and is not frightened at being dipped, sea
bathing will be very beneficial to him. If bathing is to do good, either
to an adult or to a child, it must be anticipated with pleasure, and
neither with dread nor with distaste.
251. What is the best method of administering medicine to a
child?
If he be old enough, appeal to his reason; for, if a mother endeavor
to deceive her child, and he detect her, he will for the future suspect
her.
If he be too young to be reasoned with, then, if he will not take his
medicine, he must be compelled. Lay him across your knees, let both
his hands and his nose be tightly held, and then, by means of the
patent medicine-spoon, or, if that be not at hand, by either a tea or a
dessertspoon, pour the medicine down his throat, and he will be
obliged to swallow it.
It may be said that this is a cruel procedure; but it is the only way
to compel an unruly child to take physic, and is much less cruel than
running the risk of his dying from the medicine not having been
administered.[251]
252. Ought a sick child to be roused from his sleep to give him
physic, when it is time for him to take it?
On no account, as sleep, being a natural restorative, must not be
interfered with. A mother cannot be too particular in administering
the medicine, at stated periods, while he is awake.
253. Have you any remarks to make on the management of a
sick-room, and have you any directions to give on the nursing of a
sick child?
In sickness select a large and lofty room; if in the town, the back of
the house will be preferable—in order to keep the patient free from
noise and bustle—as a sick-chamber cannot be kept too quiet. Be
sure that there be a chimney in the room—as there ought to be in
every room in the house—and that it be not stopped, as it will help to
carry off the impure air of the apartment. Keep the chamber well
ventilated, by, from time to time, opening the window. The air of the
apartment cannot be too pure; therefore, let the evacuations from the
bowels be instantly removed, either to a distant part of the house, or
to an out-house, or to the cellar, as it might be necessary to keep
them for the medical man’s inspection.
Let there be a frequent change of linen, as in sickness it is even
more necessary than in health, more especially if the complaint be
fever. In an attack of fever clean sheets ought, every other day, to be
put on the bed; clean body-linen every day. A frequent change of
linen in sickness is most refreshing.
If the complaint be fever, a fire in the grate will not be necessary.
Should it be a case either of inflammation of the lungs or of the chest,
a small fire in the winter time is desirable, keeping the temperature
of the room as nearly as possible at 60° Fahrenheit. Bear in mind
that a large fire in a sick-room cannot be too strongly condemned;
for if there be fever—and there are scarcely any complaints without—
a large fire only increases it. Small fires, in cases either of
inflammation of the lungs or of the chest, in the winter time,
encourage ventilation of the apartment, and thus carry off impure
air. If it be summer time of course fires would be improper. A
thermometer is an indispensable requisite in a sick-room.
In fever, free and thorough ventilation is of vital importance, more
especially in scarlet fever; then a patient cannot have too much air; in
scarlet fever, for the first few days the windows, be it winter or
summer, must to the widest extent be opened. The fear of the patient
catching cold by doing so is one of the numerous prejudices and
baseless fears that haunt the nursery, and the sooner it is exploded
the better it will be for human life. The valances and bed-curtains
ought to be removed, and there should be as little furniture in the
room as possible.
If it be a case of measles, it will be necessary to adopt a different
course; then the windows ought not to be opened, but the door must
from time to time be left ajar. In a case of measles, if it be winter
time, a small fire in the room will be necessary. In inflammation of
the lungs or of the chest, the windows should not be opened, but the
door ought occasionally to be left unfastened, in order to change the
air and to make it pure. Remember, then, that ventilation, either by
open window or by open door, is most necessary in all diseases.
Ventilation is one of the best friends a doctor has.
In fever, do not load the bed with clothes; in the summer a sheet is
sufficient, in the winter a sheet and a blanket.
In fever, do not be afraid of allowing the patient plenty either of
cold water or of cold toast and water; Nature will tell him when he
has had enough. In measles, let the chill be taken off the toast and
water.
In croup, have always ready a plentiful supply of hot water, in case
a warm bath might be required.
In child-crowing, have always in the sick-room a supply of cold
water, ready at a moments notice to dash upon the face.
In fever, do not let the little patient lie on the lap; he will rest more
comfortably on a horse-hair mattress in his crib or cot. If he have
pain in the bowels, the lap is most agreeable to him: the warmth of
the body, either of the mother or of the nurse, soothes him; besides,
if he be on the lap, he can be turned on his stomach and on his
bowels, which often affords him great relief and comfort. If he be
much emaciated, when he is nursed, place a pillow upon the lap and
let him lie upon it.
In head affections, darken the room with a green calico blind;
keep the chamber more than usually quiet; let what little talking is
necessary be carried on in whispers, but the less of that the better;
and in head affections, never allow smelling-salts to be applied to the
nose, as they only increase the flow of blood to the head, and
consequently do harm.
It is often a good sign when a child, who is seriously ill, suddenly
becomes cross. It is then he begins to feel his weakness, and to give
vent to his feelings. “Children are almost always cross when
recovering from an illness, however patient they may have been
during its severest moments, and the phenomenon is not by any
means confined to children.”[252]
A sick child must not be stuffed with much food at a time. He will
take either a tablespoonful of new milk or a tablespoonful of chicken-
broth every half hour, with greater advantage than a teacupful of
either the one or the other every four hours, which large quantity
would very probably be rejected from his stomach, and may cause
the unfortunately treated child to die of starvation!
If a sick child be peevish, attract his attention either by a toy or by
an ornament; if he be cross, win him over to good humor by love,
affection, and caresses, but let it be done gently and without noise.
Do not let visitors see him; they will only excite, distract, and irritate
him, and help to consume the oxygen of the atmosphere, and thus
rob the air of its exhilarating health-giving qualities and purity; a
sick-room, therefore, is not a proper place either for visitors or for
gossips.
In selecting a sick-nurse, let her be gentle, patient, cheerful, quiet,
and kind, but firm withal; she ought to be neither old nor young; if
she be old, she is often garrulous and prejudiced, and thinks too
much of her trouble; if she be young, she is frequently thoughtless
and noisy; therefore choose a middle-aged woman. Do not let there
be in the sick-room more than, besides the mother, one efficient
nurse; a great number can be of no service—they will only be in each
other’s way, and will distract the patient.
Let stillness, especially if the head be the part affected, reign in a
sick-room. Creaking shoes[253] and rustling silk dresses ought not to
be worn in sick-chambers—they are quite out of place there. If the
child be asleep, or if he be dozing, perfect stillness must be enjoined
—not even a whisper should be heard:
“In the sick-room be calm,
Move gently and with care,
Lest any jar or sudden noise
Come sharply unaware.
You cannot tell the harm,

The mischief it may bring,
To wake the sick one suddenly,
Besides the suffering.
The broken sleep excites

Fresh pain, increased distress;
The quiet slumber undisturb’d
Soothes pain and restlessness.
Sleep is the gift of God:

Oh! bear these words at heart,—
‘He giveth his beloved sleep,’
And gently do thy part.”[254]
If there be other children, let them be removed to a distant part of

the house; or, if the disease be of an infectious nature, let them be
sent away from home altogether.
In all illnesses—and bear in mind the following is most important
advice—a child must be encouraged to try and make water, whether
he ask or not, at least four times during the twenty-four hours; and at
any other time, if be expresses the slightest inclination to do so. I
have known a little fellow to hold his water, to his great detriment,
for twelve hours, because either the mother had in her trouble
forgotten to inquire, or the child himself was either too ill or too
indolent to make the attempt.
See that the medical man’s directions are, to the very letter, carried
out. Do not fancy that you know better than he does, otherwise you
have no business to employ him. Let him, then, have your implicit
confidence and your exact obedience. What you may consider to be a
trifling matter, may frequently be of the utmost importance, and may
sometimes decide whether the case shall either end in life or death!
Lice.—It is not very poetical, as many of the grim facts of everyday
life are not, but, unlike a great deal of poetry, it is unfortunately too
true that after a severe and dangerous illness, especially after a bad
attack of fever, a child’s head frequently becomes infested with
vermin—with lice! It therefore behooves a mother herself to
thoroughly examine, by means of a fine-tooth comb,[255] her child’s
head, in order to satisfy her mind that there be no vermin there. As
soon as he be well enough, he ought to resume his regular ablutions
—that is to say, that he must go again regularly into his tub, and have
his head every morning thoroughly washed with soap and water. A
mother ought to be particular in seeing that the nurse washes the
hairbrush at least once every week; if she does not do so, the dirty
brush which had, during the illness, been used, might contain the
“nits,”—the eggs of the lice,—and would thus propagate the vermin,
as they will, when on the head of the child, soon hatch. If there be
already lice on the head, in addition to the regular washing every
morning with the soap and water, and after the head has been
thoroughly dried, let the hair be well and plentifully dressed with
camphorated oil—the oil being allowed to remain on until the next
washing on the following morning. Lice cannot live in oil (more
especially if, as in camphorated oil, camphor be dissolved in it), and
as the camphorated oil will not, in the slightest degree, injure the
hair, it is the best application that can be used. But as soon as the
vermin have disappeared, let the oil be discontinued, as the natural
oil of the hair is, at other times, the only oil that is required on the
head.
The “nit”—the egg of the louse—might be distinguished from scurf
(although to the naked eye it is very much like it in appearance) by
the former fastening firmly on one of the hairs as a barnacle would
on a rock, and by it not being readily brushed off as scurf would,
which latter (scurf) is always loose.
254. My child, in the summer time, is much tormented with fleas:
what are the best remedies?
A small muslin bag, filled with camphor, placed in the cot or bed,
will drive fleas away. Each flea-bite should, from time to time, be
dressed by means of a camel’s-hair brush, with a drop or two of spirit
of camphor, an ounce bottle of which ought, for the purpose, to be
procured from a chemist. Camphor is also an excellent remedy to
prevent bugs from biting. Bugs and fleas have a horror of camphor;
and well they might, for it is death to them!
There is a famous remedy for the destruction of fleas
manufactured in France, entitled “La Poudre Insecticide,” which,
although perfectly harmless to the human economy, is utterly
destructive to fleas. Bugs are best destroyed by oil of turpentine; the
places they do love to congregate in should be well saturated, by
means of a brush, with the oil of turpentine. A few dressings will
effectually destroy both them and their young ones.
255. Suppose a child to have had an attack either of inflammation
of the lungs or of bronchitis, and to be much predisposed to a
return: what precautions would you take to prevent either the one
or the other for the future?
I would recommend him to wear fine flannel instead of lawn
shirts; to wear good lamb’s-wool stockings above the knees, and
good, strong, dry shoes to his feet; to live, weather permitting, a great
part of every day in the open air; to strengthen his system by good
nourishing food—by an abundance of both milk and meat (the
former especially); to send him, in the autumn, for a couple of
months, to the sea-side; to administer to him, from time to time,
cod-liver oil; in short, to think only of his health, and to let learning,
until he be stronger, be left alone.
I also advise either table salt or bay salt to be added to the water in
which the child is washed with in the morning, in a similar manner
as recommended in answer to the 123d question.
256. Then do you not advise such a child to be confined within
doors?
If any inflammation be present, or if he have but just recovered
from one, it would be improper to send him into the open air, but not
otherwise, as the fresh air would be a likely means of strengthening
the lungs, and thereby of preventing an attack of inflammation for
the future. Besides, the more a child is coddled within doors, the
more likely will he be to catch cold, and to renew the inflammation.
If the weather be cold, yet neither wet nor damp, he ought to be sent
out, but let him be well clothed; and the nurse should have strict
injunctions not to stand about entries, or in any draughts—indeed,
not to stand about at all, but to keep walking about all the time she is
in the open air. Unless you have a trustworthy nurse, it will be well
for you either to accompany her in her walk with your child, or
merely to allow her to walk with him in the garden, as you can then
keep your eye upon both of them.
257. If a child be either chicken-breasted, or if he be narrow-
chested, are there any means of expanding and of strengthening his
chest?
Learning ought to be put out of the question; attention must be
paid to his health alone, or consumption will probably mark him as
its own! Let him live as much as possible in the open air; if it be
country, so much the better. Let him rise early in the morning, and
let him go to bed betimes; and if he be old enough to use the dumb-
bells, or, what is better, an india-rubber chest expander, he should
do so daily. He ought also to be encouraged to use two short sticks,
similar to, but heavier than, a policeman’s staff, and to go, every
morning, through regular exercises with them. As soon as he be old
enough, let him have lessons from a drill-sergeant and from a
dancing-master. Let him be made both to walk and to sit upright,
and let him be kept as much as possible upon a milk diet,[256] and
give him as much as he can eat of fresh meat every day. Cod-liver oil,
a teaspoonful or a dessertspoonful, according to his age, twice a day,
is serviceable in these cases. Stimulants ought to be carefully
avoided. In short, let every means be used to nourish, to strengthen,
and invigorate the system, without at the same time creating fever.
Such a child should be a child of nature; he ought almost to live in
the open air, and throw his books to the winds. Of what use is
learning without health? In such a case as this you cannot have both.
258. If a child be round-shouldered, or if either of his shoulder-
blades have “grown out,” what had better be done?
Many children have either round shoulders, or have their
shoulder-blades grown out, or have their spines twisted, from
growing too fast, from being allowed to slouch in their gait, and from
not having sufficient nourishing food, such as meat and milk, to
support them while the rapid growth of childhood is going on.
If your child be affected as above described, nourish him well on
milk and on farinaceous food, and on meat once a day, but let milk
be his staple diet; he ought, during the twenty-four hours, to take two
or three pints of new milk. He should almost live in the open air, and
must have plenty of play. If you can so contrive it, let him live in the
country. When tired, let him lie, for half an hour, two or three times
daily, flat on his back on the carpet. Let him rest at night on a horse-
hair mattress, and not on a feather bed.
Let him have every morning, if it be summer, a thorough cold
water ablution; if it be winter, let the water be made tepid. Let either
two handfuls of table salt or a handful of bay salt be dissolved in the
water. Let the salt and water stream well over his shoulders and
down his back and loins. Let him be well dried with a moderately
coarse towel, and then let his back be well rubbed, and his shoulders
be thrown back—exercising them, much in the same manner as in
skipping, for five or ten minutes at a time. Skipping, by-the-by, is of
great use in these cases, whether the child be either a boy or a girl—
using, of course, the rope backward, and not forward.
Let books be utterly discarded until his shoulders have become
strong, and thus no longer round, and his shoulder-blades have
become straight. It is a painful sight to see a child stoop like an old
man.
Let him have twice daily a teaspoonful or a dessertspoonful
(according to his age) of cod-liver oil.
When he is old enough, let the drill-sergeant give him regular
lessons, and let the dancing-master be put in requisition. Let him go
through regular gymnastic exercises, provided they are not of a
violent character.
But, bear in mind, let there be in these cases no mechanical
restraints—no shoulder-straps, no abominable stays. Make him
straight by natural means—by making him strong. Mechanical
means would only, by weakening and wasting the muscles, increase
the mischief, and thus the deformity. In this world of ours there is
too much reliance placed on artificial, and too little on natural means
of cure.
259. What are the causes of Bow Legs in a child; and what is the
treatment?
Weakness of constitution, poor and insufficient nourishment, and
putting a child, more especially a fat and heavy one, on his legs too
early.
Treatment.—Nourishing food, such as an abundance of milk, and,
if he be old enough, of meat; iron medicines; cod-liver oil; thorough
ablution, every morning, of the whole body; an abundance of
exercise either on pony, or on donkey, or in carriage, but not, until
his legs be stronger, on foot. If they are much bowed, it will be
necessary to consult an experienced surgeon.
260. If a child, while asleep, “wet his bed,” is there any method of
preventing him from doing so?
Let him be held out just before he himself goes to bed, and again
when the family retires to rest. If, at the time, he be asleep, he will
become so accustomed to it that he will, without awaking, make
water. He ought to be made to lie on his side; for, if he be put on his
back, the urine will rest upon an irritable part of the bladder, and, if
he be inclined to wet his bed, he will not be able to avoid doing so. He
must not be allowed to drink much with his meals, especially with his
supper. Wetting the bed is an infirmity with some children—they
cannot help it. It is, therefore, cruel to scold and chastise them for it.
Occasionally, however, wetting the bed arises from idleness, in which
case, of course, a little wholesome correction might be necessary.
A waterproof cloth,[257] or bed-sheeting, as it is sometimes called—
one yard by three-quarters of a yard—will effectually preserve the
bed from being wetted, and ought always, on these occasions, to be
used.
A mother ought, every morning, to ascertain for herself whether
her child have wet his bed; if he have, and if unfortunately the
waterproof cloth have not been used, the mattress, sheets, and
blankets must be instantly taken to the kitchen fire and be properly
dried. Inattention to the above has frequently caused a child to
suffer, either from a cold, from a fever, or from an inflammation; not
only so, but if they be not dried, he is wallowing in filth and in an
offensive effluvium. If both mother and nurse were more attentive to
their duties—in frequently holding a child out, whether he asks or
not—a child wetting his bed would be the exception, and not, as it
frequently is, the rule. If a child be dirty, you may depend upon it the
right persons to blame are the mother and the nurse, and not the
child!
261. If a child should catch Small-pox, what are the best means to
prevent pitting?
He ought to be desired neither to pick nor to rub the pustules. If he
be too young to attend to these directions, his hands must be secured
in bags (just large enough to hold them), which bags should be
fastened round the wrists. The nails must be cut very close.
Cream smeared by means of a feather, frequently in the day, on the
pustules, affords great comfort and benefit. Tripe-liquor has, for the
same purpose, been strongly recommended. I, myself, in several
cases have tried it, and with the happiest results. It is most soothing,
comforting, and healing to the skin.
262. Can you tell me of any plan to prevent Chilblains, or, if a
child be suffering from them, to cure them?
First, then, the way to prevent them.—Let a child, who is subject
to them, wear, in the winter time, a square piece of wash-leather over
the toes, a pair of warm lamb’s-wool stockings, and good shoes; but,
above all, let him be encouraged to run about the house as much as
possible, especially before going to bed; and on no account allow him
either to warm his feet before the fire, or to bathe them in hot water.
If the feet be cold, and the child be too young to take exercise, then
let them be well rubbed with the warm hand. If adults suffer from
chilblains, I have found friction, night and morning, with horse-hair
flesh-gloves, the best means of preventing them.
Secondly, the way to cure them.—If they be unbroken, let them be
well rubbed, every night and morning, with spirits of turpentine and
camphorated oil,[258] first shaking the bottle, and then let them be
covered with a piece of lint, over which a piece of wash-leather
should be placed. “An excellent chilblain remedy is made by shaking
well together, in a bottle, spirits of turpentine, white vinegar, and the
contents of an egg, in equal proportion. With this the chilblains
should be rubbed gently whenever they are in a state of irritation,
and until the swelling and redness are dissipated.”[259]
If they be broken, let a piece of lint be spread with spermaceti
cerate, and be applied, every morning, to the part, and let a white-
bread poultice be used every night.
263. During the winter time my child’s hands, legs, etc. chap very
much: what ought I to do?
Let a teacupful of bran be tied up in a muslin bag, and be put, over
night, into either a large water-can or jug of rain water;[260] and let
this water from the can or jug be the water he is to be washed with on
the following morning, and every morning until the chaps be cured.
As often as water is withdrawn, either from the water-can or from
the jug, let fresh rain water take its place, in order that the bran may
be constantly soaking in it. The bran in the bag should be renewed
about twice a week.
Take particular care to dry the skin well every time he be washed;
then, after each ablution, as well as every night at bedtime, rub a
piece of deer’s suet over the parts affected: a few dressings will
perform a cure. The deer’s suet may be bought at any of the shops
where venison is sold. Another excellent remedy is glycerin,[261]
which should be smeared, by means of the finger or by a camel’s-hair
brush, on the parts affected, two or three times a day. If the child be
very young, it might be necessary to dilute the glycerin with
rosewater: fill a small bottle one-third with glycerin, and fill up the
remaining two-thirds of the bottle with rosewater—shaking the bottle
every time just before using it. The best soap to use for chapped
hands is the glycerin soap: no other being required.
264. What is the best remedy for Chapped Lips?
Cold-cream (which may be procured of any respectable chemist) is
an excellent application for chapped lips. It ought, by means of the
finger, to be frequently smeared on the parts affected.
265. Have the goodness to inform me of the different varieties of
Worms that infest a child’s bowels?
Principally three—1, The tape-worm; 2, the long round-worm; and
3, the most frequent of all, the common thread or maw-worm. The
tape-worm infests the whole course of the bowels, both small and
large: the long round-worm, principally the small bowels,
occasionally the stomach; it sometimes crawls out of the child’s
mouth, causing alarm to the mother; there is, of course, no danger in
its doing so: the common thread-worm or maw-worm infests the
rectum or fundament.
266. What are the causes of Worms?
The causes of worms are—weak bowels; bad and improper food,
such as unripe, unsound, or uncooked fruit, and much green
vegetables; pork, especially underdone pork;[262] an abundance of
sweets; the neglecting of giving salt in the food.
267. What are the symptoms and the treatment of Worms?
The symptoms of worms are—emaciation; itching and picking of
the nose; a dark mark under the eyes; grating, during sleep, of the
teeth; starting in the sleep; foul breath; furred tongue; uncertain
appetite—sometimes voracious, at other times bad, the little patient
sitting down very hungry to his dinner, and before scarcely tasting a
mouthful, the appetite vanishing; large bowels; colicky pains of the
bowels; slimy motions; itching of the fundament. Tape-worm and
round-worm, more especially the former, are apt, in children, to
produce convulsions. Tape-worm is very weakening to the
constitution, and usually causes great emaciation and general ill
health; the sooner therefore it is expelled from the bowels the better
it will be for the patient.
Many of the obscure diseases of children arise from worms. In all
doubtful cases, therefore, this fact should be borne in mind, in order
that a thorough investigation may be instituted.
With regard to treatment, a medical man ought, of course, to be
consulted. He will soon use means both to dislodge them, and to
prevent a future recurrence of them.
Let me caution a mother never to give her child patent medicines
for the destruction of worms. There is one favorite quack powder,
which is composed principally of large doses of calomel, and which is
quite as likely to destroy the patient as the worms! No, if your child
have worms, put him under the care of a judicious medical man, who
will soon expel them, without, at the same time, injuring health and
constitution!
268. How may Worms be prevented from infesting a child’s
bowels?
Worms generally infest weak bowels; hence, the moment a child
becomes strong worms cease to exist. The reason why a child is so
subject to them is owing to the improper food which is usually given
to him. When he be stuffed with unsound and with unripe fruits,
with much sweets, with rich puddings, and with pastry, and when he
is oftentimes allowed to eat his meat without salt, and to bolt his
food without chewing it, is there any wonder that he should suffer
from worms? The way to prevent them is to avoid such things, and,
at the same time, to give him plenty of salt to his fresh and well-
cooked meat. Salt strengthens and assists digestion, and is absolutely
necessary to the human economy. Salt is emphatically a worm-
destroyer. The truth of this statement may be readily tested by
sprinkling a little salt on the common earthworm. “What a comfort
and real requisite to human life is salt! It enters into the constituents
of the human blood, and to do without it is wholly impossible.”[263]
To do without it is wholly impossible! These are true words. Look
well to it, therefore, ye mothers, and beware of the consequences of
neglecting such advice, and see for yourselves that your children
regularly eat salt with their food. If they neglect eating salt with their
food, they must, of necessity, have worms, and worms that will
eventually injure them and make them miserable.
269. You have a great objection to the frequent administration of
aperient medicines to a child: can you devise any method to prevent
their use?
Although we can scarcely call constipation a disease, yet it
sometimes leads to disease. The frequent giving of aperients only
adds to the stubbornness of the bowels.
I have generally found a draught, early every morning, of cold
pump-water, the eating either of loaf gingerbread or of oatmeal
gingerbread, a variety of animal and vegetable food, ripe sound fruit,
Muscatel raisins, a fig, or an orange after dinner, and, when he be old
enough, coffee and milk instead of tea and milk, to have the desired
effect, more especially if, for a time, aperients be studiously avoided.
270. Have you any remarks to make on Rickets?
Rickets is owing to a want of a sufficient quantity of earthy matter
in the bones; hence the bones bend and twist, and lose their shape,
causing deformity. Rickets generally begins to show itself between
the first and second years of a child’s life. Such children are generally
late in cutting their teeth, and when the teeth do come, they are bad,
deficient of enamel, discolored, and readily decay. A rickety child is
generally stunted in stature; he has a large head, with overhanging
forehead, or what nurses call a watery-head-shaped forehead. The
fontanelles, or openings of the head as they are called, are a long
time in closing. A rickety child is usually talented; his brain seems to
thrive at the expense of his general health. His breast-bone projects
out, and the sides of his chest are flattened—hence he becomes what
is called chicken-breasted or pigeon-breasted; his spine is usually
twisted, so that he is quite awry, and, in a bad case, he is hump-
backed; the ribs, from the twisted spine, on one side bulge out; he is
round-shouldered; the long bones of his body, being soft, bend; he is
bow-legged, knock-kneed, and weak ankled.
Rickets are of various degrees of intensity, the hump-backed being
among the worst. There are many mild forms of rickets; weak ankles,
knocked-knees, bowed-legs, chicken-breasts, being among the latter
number. Many a child, who is not exactly hump-backed, is very
round-shouldered, which latter is also a mild species of rickets.
Show me a child that is rickety, and I can generally prove that it is
owing to poor living, more especially to poor milk. If milk were
always genuine, and if a child had an abundance of it, my belief is
that rickets would be a very rare disease. The importance of genuine
milk is of national importance. We cannot have a race of strong men
and women unless, as children, they have had a good and plentiful
supply of milk. It is utterly impossible. Milk might well be considered
one of the necessaries of a child’s existence.
Genuine fresh milk, then, is one of the grand preventives, as well
as one of the best remedies, for rickets. Many a child would not now
have to swallow quantities of cod-liver oil if previously he had
imbibed quantities of good genuine milk. An insufficient and a poor
supply of milk in childhood sows the seeds of many diseases, and
death often gathers the fruit. Can it be wondered at, when there is so

Textbook Generalized Locally Toeplitz Sequences Theory and Applications Volume Ii Carlo Garoni Ebook All Chapter PDF

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

Textbook Generalized Locally Toeplitz Sequences Theory and Applications Volume Ii Carlo Garoni Ebook All Chapter PDF

Uploaded by

Copyright:

Available Formats

Generalized Locally Toeplitz Sequences

Theory and Applications Volume II

Generalized Intuitionistic Multiplicative Fuzzy

Combinatorics and Number Theory of Counting Sequences

Structural Dynamics Fundamentals and Advanced

Map Projections: Theory and Applications Frederick

Legal Theory and the Social Sciences Volume II

Existence Theory for Generalized Newtonian Fluids 1st

Generalized Fractional Calculus: New Advancements and

Computation of Generalized Matrix Inverses and

Generalized Locally Toeplitz

ISBN 978-3-030-02232-7 ISBN 978-3-030-02233-4 (eBook)

Library of Congress Control Number: 2018958367

© Springer Nature Switzerland AG 2018

Based on their research experience, the authors propose a reference textbook in

Como, Italy Carlo Garoni

1 Notes to the Reader . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

3 Multilevel Toeplitz Sequences . . . . . . . . . . . . . . . . . . . . . ......... 39

7.5 B-Spline IgA Collocation Discretization

Dr. Carlo Garoni graduated in Mathematics at the University of Insubria (Italy) in

Prof. Stefano Serra-Capizzano is a Full Professor of Numerical Analysis, Deputy

© Springer Nature Switzerland AG 2018 1

2.1 Notation and Terminology

2.1.1 General Notation and Terminology

• The cardinality of a set S is denoted by #S.

• If x, y are vectors with m components, x · y denotes their scalar product.

• Cc (C) (resp.,Cc (R)) is the space of complex-valued continuous functions defined

(w1 ⊗ · · · ⊗ wd )(ξ1 , . . . , ξd ) = w1 (ξ1 ) · · · wd (ξd ).

• If f : D → E and g : E → F are arbitrary functions, the composite function

ωg (δ) = sup |g(x) − g(y)|, δ > 0.

If we need/want to specify D, we will say that ωg (·) is the modulus of continuity

• μk denotes the Lebesgue measure in Rk . Throughout this book, unless otherwise

Md = {κ : [0, 1]d × [−π, π ]d → C : κ is measurable}.

For more on L p spaces, see [22, Sect. 2.2.2].

• A sequence of matrices is a sequence of the form {An }n , where n varies in some

2.1.2 Multi-index Notation

lim sup an = lim sup am , lim inf an = lim inf am .

• If h, k are d-indices, h ≤ k means that h r ≤ kr for all r = 1, . . . , d, while h ≤ k

For instance, in the case d = 2 the ordering is

• When a d-index j varies over a d-index range h, . . . , k (this is often written as

of size N (m) whose components x i , i = 1, . . . , m, are ordered in accordance with

then X is an N (m) × N (m) matrix whose components are indexed by a pair of

Example 2.3 Consider the matrix

a(1,1),(1,1) = 4, a(1,1),(1,2) = 4, a(1,1),(2,1) = 0, a(1,1),(2,2) = 0,

We can therefore write

Using the identities

r = r/ss + r mod s, r/s = (r + 1)/s − 1,

i = (i/1  − 1)1 + ((i − 1) mod 1 ) + 1,

The (i, j) entry of the matrix Z is then given by

z i j = xi/1 , j/2  y((i−1) mod 1 )+1,(( j−1) mod 2 )+1 .

#{1, . . . , n} = N (n) = m 1 1 = number of rows of Z ,

z i j = entry (i 2 , j2 ) in the (i 1 , j1 ) block of Z = xi1 j1 yi2 j2 .

Example 2.5 (multilevel matrices) In many cases, it is convenient to partition matri-

A = [Ai1 j1 ]in11, j1 =1 , Ai1 j1 ∈ Cñ 1 ×ñ 1 , ñ 1 = N

where A i j = Ai1 j1 ; ...; id jd for i, j = 1, . . . , n. We remark that, as we shall see in

Example 2.6 (matrix computations with multi-indices) Let n ∈ Nd and let

A = [a i j ]ni, j =1 , B = [bi j ]ni, j =1 , x = [x j ]nj=1 , y = [y j ]nj=1 .

It is not difficult to see that the following properties hold.

2.1.3 Multilevel Matrix-Sequences

2.2 Multivariate Trigonometric Polynomials

A d-variate trigonometric polynomial is a finite linear combination of the d-variate

that is, a function of the form

Note that, as a consequence of the orthogonality relations

• If h, k are d-indices, h ≤ k means that h r ≤ kr for all r = 1, . . . , d, while h ≤ k

r = r/ss + r mod s, r/s = (r + 1)/s − 1,

i = (i/1 − 1)1 + ((i − 1) mod 1 ) + 1,

z i j = xi/1 , j/2 y((i−1) mod 1 )+1,(( j−1) mod 2 )+1 .

#{1, . . . , n} = N (n) = m 1 1 = number of rows of Z ,

and the direct sum of X and Y is the (m 1 + 1 ) × (m 2 + 2 ) matrix defined by

{λi (X )λ j (Y ) : i = 1, . . . , m, j = 1, . . . , }, (2.21)

{λi (X ) : i = 1, . . . , m} ∪ {λ j (Y ) : j = 1, . . . , }, (2.23)