Chapter 10

Epilogue: Historical Notes on Functional

Analysis

Differentiae et summae sibi reciprocae sunt, hoc est summa

differentiarum seriei est seriei terminus, et differentia
summarum ! seriei est ipse seriei terminus,
! quorum illud ita
enuntio: dx aequ. x; hoc ita: d x aequ. x.
G. Leibniz

10.1 Integral Calculus

In a concise description of mathematical methods, Henri Lebesgue underlined the

importance of definitions and axioms (see [47]).
When a mathematician foresees, more or less clearly, a proposition, instead of having
recourse to experiment like the physicist, he seeks a logical proof. For him, logical
verification replaces experimental verification. In short, he does not seek to discover new
materials but tries to become aware of the richness that he already unconsciously possesses,
which is built in the definitions and axioms. Herein lies the supreme importance of these
definitions and axioms, which are indeed subjected logically only to the condition that they
be compatible, but which could lead only to a purely formal science, void of meaning, if
they had no relationship to reality.

Leibniz conceived integration as the reciprocal of differentiation:

" "
dx = d x = x.

The computation of the integral of f is reduced to the search for its primitive,
solution of the differential equation
F ′ = f.
The textbooks by Cauchy, in particular the Analyse algébrique (1821) (see [7])
and the Résumé des leçons données à l’Ecole Royale Polytechnique sur le calcul
infinitésimal (1823), opened a new area in analysis. Cauchy was the first to consider
the problem of existence of primitives:

M. Willem, Functional Analysis: Fundamentals and Applications, Cornerstones, 193

DOI 10.1007/978-1-4614-7004-5 10, © Springer Science+Business Media, LLC 2013
194 10 Epilogue: Historical Notes on Functional Analysis

In integral calculus, it seemed necessary to me to demonstrate in general the existence of

integrals or primitive functions before giving their various properties. In order to reach this,
it was necessary to establish the notion of integral between two given limits or definite
integral.

Cauchy defines and proves the existence of the integral of continuous functions:

According to the preceding lecture, if one divides X − x0 into infinitesimal elements x1 −

x0 , x2 − x1 · · · X − xn−1 , the sum

S = (x1 − x0 ) f (x0 ) + (x1 − x2 ) f (x1 ) + · · · + (X − xn−1 ) f (xn−1 )

will converge to a limit given by the definite integral

" X
f (x)dx.
x0

So Cauchy proved the existence of primitives of continuous functions using integral

calculus.
Though every continuous function has a primitive, Weierstrass proved in 1872
the existence of continuous nowhere differentiable functions. In a short note [44],
Lebesgue proved the existence of primitives of continuous functions without using
integral calculus. His proof is clearly functional-analytic.
In 1881 ([37]), Camille Jordan defined the functional space of functions of
bounded variation, which he called functions of limited oscillation. His goal was
to linearize Dirichlet’s condition for the convergence of Fourier series:

Let x1 , . . . , xn be a series of values of x between 0 and ε, and y1 , . . . , yn the corresponding

values of f (x). The points x1 , y1 ; . . . ; xn , yn will form a polygon.
Consider the differences
y2 − y1 , y3 − y2 , . . . , yn − yn−1 .
We will call the sum of the positive terms of this sequence the positive oscillation of the
polygon; negative oscillation is the sum of the negative terms; total oscillation is the sum
of those two partial oscillations in absolute value.
Let us vary the polygon; two cases may occur:
1◦ The polygon may be chosen so that its oscillations exceed every limit.
2◦ For every chosen polygon, its positive and negative oscillations will be less than some
fixed limits Pε and Nε . We will say in that case that F(x) is a function of limited
oscillation in the interval from 0 to ε; Pε will be its positive oscillation; Nε its negative
oscillation; Pε + Nε its total oscillation.
This case will necessarily occur if F(x) is the difference of two finite functions f (x) − ϕ(x),
because it is clear that the positive oscillation of the polygon will be = f (ε) − f (0), and its
<
negative oscillation = ϕ(ε) − ϕ(0).
<
The converse is easy to prove. Indeed, it is easy to verify that
1◦ The oscillation of a function from 0 to ε is equal to the sum of its oscillations from 0 to
x and from x to ε, x being any quantity between 0 and ε.
10.1 Integral Calculus 195

2◦ We have that F(x) = F(0) + Px − Nx , Px and Nx denoting the positive and the negative
oscillations from 0 to x. But F(0) + Px and Nx are finite functions nondecreasing from 0
to ε.
Hence Dirichlet’s proof is applicable, without modification, to every function of bounded
oscillation from x = 0 to x = ε, ε being any finite quantity.
The functions of limited oscillations constitute a well-defined class, whose study could be
of some interest.

Functions of bounded variation will play a fundamental role in the following

domains:
(a) Convergence of Fourier series;
(b) Rectification of curves;
(c) Integration;
(d) Duality.
Let u : [0, 1] → R be a continuous function. The length of the graph of u is
defined by
⎧
⎪
⎪ k
⎨'(
⎪ ) * +1/2
L(u) = sup ⎪
⎪ (a j+1 − a j )2 + u(a j+1 ) − u(a j ) 2 :
⎪
⎩ j=0
,
k ∈ N, 0 = a0 < a1 < . . . < ak+1 = 1 .

In 1887, in Volume III of the first edition of his Cours d’Analyse at the École
Polytechnique, Jordan proved that L(u) is finite if and only if u is of bounded
variation. The case of surfaces is much more delicate (see Sect. 10.3).

In 1894 ([80]), Stieltjes defined a deep generalization of the integral associated

to an increasing function ϕ:
More generally, let us consider the sum
- . - . - .
f (ξ1 ) ϕ(x1 ) − ϕ(x0 ) + f (ξ2 ) ϕ(x2 ) − ϕ(x1 ) + . . . + f (ξn ) ϕ(xn ) − ϕ(xn−1 ) . (A)
It will still have a limit, which we shall denote by
" b
f (u)dϕ(u).
a

1
We will have only to consider some very simple cases like f (u) = uk , f (u) = z+u , and there
is no interest in giving to the function f (u) its full generality. Thus it will suffice, as an
example, to suppose the function f (u) continuous, and then the proof presents no difficulty,
and we have no need to develop it, since it is done as in the ordinary case of a definite
integral.

It is easy to extend Stieltjes’s definition to every function ϕ of bounded variation.

Stieltjes breaks the reciprocity between integral and derivative.
In 1903 ([32]), J. Hadamard characterized the continuous linear functionals on
C([a, b]):
196 10 Epilogue: Historical Notes on Functional Analysis

It is easy to reach this, following Weierstrass and Kirchhoff, and introducing a function
F(x), with a finite number of maxima and minima and such that
" +∞
F(x)dx = 1;
−∞
2
e.g., F(x) = √1π e−x .
Starting then from the well-known identity
" a
lim µ f (x)F[µ(x − x0 )]dx = f (x0 ), a < x0 < b,
µ=±∞ b

and assuming (as the authors quoted before) the operation U to be continuous (in the sense
of Bourlet), it will suffice to define

U[µFµ(x − x0 )] = Φ(x0 , µ)
to show that our operation could be represented as
" b
U[ f (x)] = lim f (x)Φ(x, µ)dx.
µ=±∞ a

In 1909 ([61]), F. Riesz discovered a representation depending on only one

function:

In the present note, we shall develop a new analytic expression of the linear operation,
containing only one generating function.
Given the linear operation A[ f (x)], we can determine a function of bounded variation α(x)
such that for every continuous function f (x), we have
" 1
A[ f (x)] = f (x)dα(x).
0

Riesz’s theorem asserts that every continuous linear functional on C([0, 1]) is
representable by a Stieltjes integral.

10.2 Measure and Integral

Les notions introduites sont exigées par la solution d’un

problème, et, en vertu de la seule présence parmi les notions
antérieures, elles posent à leur tour de nouveaux problèmes.
Jean Cavaillès

In 1898, Emile Borel defined the measure of sets in his Leçons sur la théorie des
fonctions:

The procedure that we have employed actually amounts to this: we have recognized that a
definition of measure could be useful only if it had certain fundamental properties: we have
10.2 Measure and Integral 197

stated these properties a priori, and we have used them to define the class of sets that we
consider measurable.
Those essential properties that we summarize here, since we shall use them, are the
following: The measure of a sum of a denumerable infinity of sets is equal to the sum
of their measures; the measure of the difference of two sets is equal to the difference
of their measures; the measure is never negative; every set with a nonzero measure is
not denumerable. It is mainly this last property that we shall use. Besides, it is explicitly
understood that we speak of measures only for those sets that we called measurable.
Of course, when we speak of the sum of several sets, we assume that every pair them have
no common points, and when we speak of their difference, we assume that one set contains
all the points of the other.

Following Lebesgue,
The descriptive definition of measure stated by M. Borel is without doubt the first clear
example of the use of actual infinity in mathematics.

However, Borel does not prove the existence of the measure!

The Lebesgue integral first appeared on the 29 April 1901. In the note [42],
Lebesgue proved the existence of the Borel measure as a restriction of the Lebesgue
measure.
In the introduction of his thesis [43], Lebesgue stated his program:
In this work, I try to give definitions as general and precise as possible of some of the
numbers considered in Analysis: definite integral, length of a curve, area of a surface.

He formulated the problem of the measure of sets:

We intend to assign to every bounded set a positive or zero number called its measure and
satisfying the following conditions:
1. There exist sets with nonzero measure.
2. Two equal sets have equal measures.
3. The measure of the sum of a finite number or of a countable infinity of sets, without
common points, is the sum of the measures of those sets.
We will solve this problem of measure only for the sets that we will call measurable.

In his Leçons sur l’intégration et la recherche des fonctions primitives of 1904,

see [45], Lebesgue formulated the problem of integration.

We intend to assign to every bounded function f (x) defined on a finite interval (a, b),
!b
positive, negative, or zero, a finite number a f (x)dx, which we call the integral of f (x)
in (a, b) and which satisfies the following conditions:
1. For every a, b, h, we have
" b " b+h
f (x)dx = f (x − h)dx.
a a+h

2. For every a, b, c, we have

" b " c " a
f (x)dx + f (x)dx + f (x)dx = 0.
a b c
198 10 Epilogue: Historical Notes on Functional Analysis

3. " " "

b b b
[ f (x) + ϕ(x)]dx = f (x)dx + ϕ(x)dx.
a a a

4. If we have f ! 0 and b > a, we also have

" b
f (x)dx ! 0.
a

5. We have " 1
1 × dx = 1.
0

6. If fn (x) increases and converges to f (x), then the integral of fn (x) converges to the
integral of f (x).
Formulating the six conditions of the integration problem, we define the integral. This
definition belongs to the class of those that could be called descriptive; in those definitions,
we state the characteristic properties of the object we want to define. In the constructive
definitions, we state which operations are to be done in order to obtain the object we
want to define. Constructive definitions are more often used in Analysis; however, we use
sometimes descriptive definitions; the definition of the integral, following Riemann, is
constructive; the definition of primitive functions is descriptive.

In 1906, in his thesis [23], Maurice Fréchet tried to extend the fundamental
notions of analysis to abstract sets.

In this Mémoire we will use an absolutely general point of view that encompass these
different cases.
To this end, we shall say that a functional operation U is defined on a set E of elements of
every kind (numbers, curves, points, etc.) when to every element A of E there corresponds
a determined numerical value of U : U(A). The search for properties of those operations
constitutes the object of the Functional Calculus.

Fréchet defined distance which he called, in French, écart:

We can associate to every pair of elements A, B a number (A, B) ≥ 0, which we will call the
distance of the two elements and which satisfies the following properties: (a) The distance
(A, B) is zero only if A and B are identical. (b) If A, B, C are three arbitrary elements, we
always have (A, B) ≤ (A, C) + (C, B).

In [24], Fréchet defined additive families of sets and additive functions of sets:

An additive family of sets is a collection of sets such that:

1. If E1 , E2 are two sets of this family, the set E1 − E2 of elements of E1 , if they exist and
that are not in E2 , belongs also to the family.
2. If E1 , E2 , . . . is a denumerable sequence of sets of this family, their sum, i.e., the set
E1 + E2 + · · · of elements belonging at least to one set of the sequence, belongs also to
the family.
10.2 Measure and Integral 199

A set function f (E) defined on an additive family of sets F is additive on F if E1 , E2 , . . .

being a denumerable sequence of sets of F and disjoint, i.e., without pairwise common
elements, we have
f (E1 + E2 + . . .) = f (E1 ) + f (E2 ) + · · · .
When the sequence is infinite, the second member has obviously to converge regardless of
the order of the terms. Hence the series in the second member has to converge absolutely.

Fréchet defined the integral without using topology. Additive functions of sets
will be called measures.
In [12], Daniell chose a different method. He introduced a space L of elementary
functions and an elementary integral
"
L → R : u +→ u dµ

satisfying the axioms of linearity, positivity, and monotone convergence.

The two axiomatics are equivalent if to Daniell’s axioms we add Stone’s axiom
(1948):
for every u ∈ L, min(u, 1) ∈ L,

or the axiom
for every u, v ∈ L, uv ∈ L.

The choice of primitive notions and axioms is rather arbitrary. There are no
absolutely undefinable notions or unprovable propositions.
The axiomatization of integration by Fréchet opened the way to the axiomatiza-
tion of probability by Kolmogorov in 1933. The unification of measure, integral, and
probability was one the greatest scientific achievements of the twentieth century.
In his thesis [5], Banach defined the complete normed spaces:

There exists an operation, called norm (we shall denote it by the symbol ||X||), defined in the
field E, having as an image the set of real numbers and satisfying the following conditions:
||X|| ≥ 0,
||X|| = 0 if and only if X = θ,
||a · X|| = |a| · ||X||,
||X + Y|| ≤ ||X|| + ||Y||.
If 1. {Xn } is a sequence of elements of E, 2. r→∞
lim ||Xr − X p || = 0, there exists an element X
p→∞
such that
lim ||X − Xn || = 0.
n→∞

Banach emphasized the efficiency of the axiomatic method:

The present work intends to prove theorems valid for different functional fields, which I
will specify in the sequel. However, in order not to be forced to prove them individually for
200 10 Epilogue: Historical Notes on Functional Analysis

every particular field, a tedious task, I chose a different way: I consider in some general way
sets of elements with some axiomatic properties, I deduce theorems, and I prove afterward
that the axioms are valid for every specific functional field.

The fundamental book of Banach ([6]), Théorie des opérations linéaires, was
published in 1932. Banach deduces Riesz’s representation theorem from the Hahn–
Banach theorem.
The original proof of the Hahn–Banach theorem holds in every real vector space.
Let F : X → R be a positively homogeneous convex function and let f : Z → R
be a linear function such that f ≤ F on the subspace Z of X. By the well-ordering
theorem, the set X \ Z can be so ordered that each nonempty subset has a least
element. It follows then, from Lemma 4.1.3, / by transfinite induction, that there
exists g : X → R such that g ≤ F on X and g//Z = f .
Let us recall the principle of transfinite induction (see [72]). Let B be a subset of
a well-ordered set A such that

{y ∈ A : y < x} ⊂ B ⇒ x ∈ B.

Then B = A.
In set theory, the well-ordering theorem is equivalent to the axiom of choice and
to Zorn’s lemma. In 1905, Vitali proved the existence of a subset of the real line that
is not Lebesgue measurable. His proof depends on the axiom of choice.

10.3 Differential Calculus

L’activité des mathématiciens est une activité expérimentale.

Jean Cavaillès

Whereas the integral calculus transforms itself into an axiomatic theory, the
differential calculus fits into the general theory of distributions.
The fundamental notions are
– Weak solutions;
– Weak derivatives;
– Functions of bounded variation;
– Distributions.
In [60], Poincaré defined the notion of weak solution of a boundary value
problem:

Let u be a function satisfying the following conditions:

du
+ h u = ϕ, (3)
dn
10.3 Differential Calculus 201

∆u + f = 0. (4)
Now let v be an arbitrary function, which I assume only continuous, together with a first-
order derivative. We shall have
" 0 1 "
du dv
v −u dω = (v∆ u − u∆ v)dτ,
dn dn

so that
" " " " 0 1
dv
v f dτ + u∆ v dτ + vϕ dω = u hv+ dω. (5)
dn
Condition (5) is thus a consequence of condition (3).
Conversely, if condition (5) is satisfied for every function v, condition (3) will be also
satisfied, provided that u and du
dn are finite, well-defined, and continuous functions.
But it can happen that in some cases, we are unaware that du dn is a well-defined and
continuous function; we cannot assert then that condition (5) entails condition (3), and it
is even possible that condition (3) is meaningless.

Poincaré named condition (5) a modified condition and asserted (p. 121),
It is obviously equivalent to condition (3) from the physical point of view.

This Mémoire of Poincaré contains (p. 70) the first example of an integral
inequality between a function and its derivatives:

Let V be an arbitrary function of x, y, z; define:

" " ⎡⎢0 12 0 12 0 12 ⎤

2 ⎢⎢⎢ dV dV dV ⎥⎥⎥⎥
A= V dτ, B= ⎢⎣ + + ⎥ dτ.
dx dy dz ⎦

I will write to shorten:

" ' 0 12
dV
B= dτ.
dx

I assume first that V satisfies the condition:

"
V dτ = 0

and I intend to estimate the lower limit of the quotient AB .

The maximum principle is stated on p. 92. Poincaré’s principle appears in [59]

for the formal construction of the eigenvalues and eigenfunctions of the Laplacian.
In [60], Poincaré proved the existence of eigenvalues (for Dirichlet boundary
conditions) using the theory of meromorphic functions (see [50]).
Let us recall that we denote by L(u) the length of the graph of the continuous
function u : [0, 1] → R. Following Jordan, L(u) < ∞ if and only if u is of
bounded variation. It follows then from a theorem due to Lebesgue that u is almost
everywhere differentiable on [0, 1]. In [82], Tonelli proved a theorem equivalent to
202 10 Epilogue: Historical Notes on Functional Analysis

" 1 8
L(u) = 1 + (u′ (x))2 dx ⇐⇒ u ∈ W 1,1 (]0, 1[).
0

A counterexample due to Schwarz, published in 1882 in the Cours d’Analyse of

Hermite, shows that it is not possible to extend the definition of length due to Jordan
to surfaces. Let z = u(x, y) be a nonparametric surface, with u continuous on [0, 1] ×
[0, 1]. Let Ω =]0, 1[×]0, 1[ and define, on X = C(Ω), the distance

d(u, v) = max{|u(x, y) − v(x, y)| : (x, y) ∈ Ω}.

The space of quasilinear functions on Ω is defined by

Y = {u ∈ X : there exists a triangulation

// τ of Ω
such that, for every T ∈ τ, u/T is affine}.

The graph of u ∈ Y consists of triangles. The sum of the areas of those triangles is
called the elementary area of the graph of u and is denoted by B(u).
The Lebesgue area of the graph of u is defined by
9 ,
A(u) = inf lim B(un) : (un ) ⊂ Y and d(un , u) → 0, n→∞ . (∗)
n→∞

In [83] (see also [53]), Tonelli stated two theorems equivalent to

A(u) < ∞ ⇐⇒ ||Du||Ω < ∞,

:
" 0 12 0 12
∂u ∂u
A(u) = 1+ + dx dy ⇐⇒ u ∈ W 1,1 (Ω).
Ω ∂x ∂y

Lebesgue area is a lower semicontinuous function on X. It extends the elementary

area: for every u ∈ Y, A(u) = B(u).
In [25], Fréchet observed that Lebesgue’s definition allows one to extend lower
semicontinuous functions. Let Y be a dense subset of a metric space X and let B :
Y → [0, +∞] be an l.s.c. function. The function A defined by (∗) is an l.s.c. extension
of B on X such that for every l.s.c. extension C of B on X and for every u ∈ X,
C(u) ≤ A(u).
In [48], Leray defined the weak derivatives of L2 functions, and called them
quasi-dérivées.
In [75], announced in [74] and translated in [78], Sobolev defined the distribu-
tions of finite order on RN , which he called fonctionnelles. (A distribution f on RN
is of order k if for every sequence (un ) ⊂ D(RN ) such that the supports of un are
contained in some compact set and such that sup ||∂α un ||∞ → 0, n → ∞, we have
|α|≤|≤k
10.4 Comments 203

⟨ f, un ⟩ → 0, n → ∞.) Sobolev defined the derivative of a fonctionnelle by duality

and associated a fonctionnelle to every locally integrable function on RN .
Without reference to his theory of fonctionnelles, Sobolev defined in [77] the
weak derivatives of integrable functions. Regularization by convolution is due to
Leray for L2 functions (see [48]) and to Sobolev for L p functions (see [77]).
In [69], Laurent Schwartz defined general distributions. In [70], he defined the
tempered distributions and their Fourier transform. The treatise [71] is a masterful
exposition of distribution theory.
Let g : R → R be a function of bounded variation on every bounded interval.
The formula of integration by parts shows that for every u ∈ D(R),
" "
udg=− u′ g dx.
R R

The Stieltjes integral with respect to g is nothing but the derivative of g in the sense
of distributions! Riesz’s representation theorem asserts that every continuous linear
functional on C([0, 1]) is the derivative in the sense of distributions of a function of
bounded variation.

10.4 Comments

Some general historical references are [15, 19, 29]. We recommend also [46] on
Jordan, [52] on Hadamard, [81] on Fréchet, and [38] on Banach.
References

1. Adams, R., Fournier, J.: Sobolev Spaces, 2nd edn. Elsevier, Oxford (2003)
2. Alberti, G.: Some remarks about a notion of rearrangement. Ann. Scuola Norm. Sup. Pisa Cl.
Sci. 29, 457–472 (2000)
3. Baernstein, A., II, Taylor, B.A.: Spherical rearrangements, subharmonic functions, and
∗-functions in n-space. Duke Math. J. 43, 245–268 (1976)
4. Baker, J.A.: Integration over spheres and the divergence theorem for balls. Am. Math. Mon.
104, 36–47 (1997)
5. Banach, S.: Sur les opérations dans les ensembles abstraits et leur application aux équations
intégrales. Fund. Math. 3, 133–181 (1922)
6. Banach, S.: Théorie des opérations linéaires. Monografje matematyczne, Varsovie (1932)
7. Bradley, R., Sandifer, E.: Cauchy’s Cours d’analyse. An Annotated Translation. Springer, New
York (2009)
8. Brezis, H.: Analyse fonctionnelle, théorie et applications. Masson, Paris (1983)
9. Brezis, H., Browder, F.: Partial differential equations in the 20th century. Adv. Math. 135,
76–144 (1998)
10. Brezis, H., Lieb, E.: A relation between pointwise convergence of functions and convergence
of functionals. Proc. Am. Math. Soc. 88, 486–490 (1983)
11. Choquet, G.: Theory of capacities. Annales de l’Institut Fourier 5, 131–295 (1953)
12. Daniell, P.: A general form of integral. Ann. Math. 19, 279–294 (1918)
13. Degiovanni, M., Magrone, P.: Linking solutions for quasilinear equations at critical growth
involving the “1-Laplace” operator. Calc. Var. Part. Differ. Equat. 36, 591–609 (2009)
14. De Giorgi, E.: Definizione ed espressione analitica del perimetro di un insieme. Atti Accad.
Naz. Lincei Rend. Cl. Sci. Fis. Mat. Natur. (8) 14, 390–393 (1953)
15. De Giorgi, E.: Riflessioni su matematica e sapienza. Accademia Pontaniana, Naples (1996)
16. De Giorgi, E.: Semicontinuity Theorems in the Calculus of Variations. Accademia Pontaniana,
Naples (2008)
17. de la Vallée Poussin, C.: Sur l’intégrale de Lebesgue. Trans. Am. Math. Soc. 16, 435–501
(1915)
18. Deny, J., Lions, J.L.: Les espaces du type de Beppo Levi. Annales de l’Institut Fourier 5,
305–370 (1954)
19. Dugac, P.: Histoire de l’analyse. Vuibert, Paris (2003)
20. Ekeland, I.: On the variational principle. J. Math. Anal. Appl. 47, 324–353 (1974)
21. Ekeland, I.: Nonconvex minimization problems. Bull. Am. Math. Soc. 1, 443–474 (1979)
22. Favard, J.: Cours d’analyse de l’Ecole Polytechnique, tome, vol. I. Gauthier-Villars, Paris
(1960)
23. Fréchet, M.: Sur quelques points du Calcul Fonctionnel. Rend. Circ. Mat. Palermo 22, 1–74
(1906)

M. Willem, Functional Analysis: Fundamentals and Applications, Cornerstones, 205

DOI 10.1007/978-1-4614-7004-5, © Springer Science+Business Media, LLC 2013
206 References

24. Fréchet, M.: Sur l’intégrale d’une fonctionnelle étendue à un ensemble abstrait. Bull. Soc.
Math. de France 43, 248–265 (1915)
25. Fréchet, M.: Sur le prolongement des fonctions semi-continues et sur l’aire des surfaces
courbes. Fund. Math. 7, 210–224 (1925)
26. Gagliardo, E.: Caratterizzazioni delle tracce sulla frontiera relative ad alcune classi di funzioni
in n variabili. Rend. Sem. Mat. Univ. Padova 27, 284–305 (1957)
27. Gagliardo, E.: Proprietà di alcune classi di funzioni in più variabili. Ric. Mat. 7, 102–137
(1958)
28. Garnir, H.G., De Wilde, M., Schmets, J.: Analyse Fonctionnelle. Birkhäuser, Basel, vol. I
(1968), vol. II (1972), vol. III (1973)
29. Giusti, E.: Ipotesi sulla natura degli oggetti matematici. Bollati-Boringhieri, Torino (2000)
30. Golse, F., Laszlo, Y., Viterbo, C.: Analyse Réelle et Complexe. Ecole Polytechnique, Palaiseau
(2010)
31. Guiraldenq, P.: Émile Borel, 1871–1956. L’espace et le temps d’une vie sur deux siècles.
Librairie Albert Blanchard, Paris (1999)
32. Hadamard, J.: Sur les opérations fonctionnelles. C. R. Acad. Sci. Paris 136, 351–354 (1903)
33. Hajłasz, P.: Note on Meyers–Serrin’s theorem. Exposition. Math. 11, 377–379 (1993)
34. Hanner, O.: On the uniform convexity of L p and ℓ p . Arkiv för Mathematik 3, 239–244 (1955)
35. James, R.C.: Orthogonality and linear functionals in normed linear spaces. Trans. Am. Math.
Soc. 61, 265–292 (1947)
36. Jensen, J.L.: Sur les fonctions convexes et les inégalités entre les valeurs moyennes. Acta Math.
30, 175–193 (1905)
37. Jordan, C.: Sur la série de Fourier. C. R. Acad. Sci. Paris 92, 228–230 (1881)
38. Kaluza, R., Kostant A., Woycyznski, W.: The Life of Stefan Banach. Birkhäuser, Boston (1996)
39. Kahane, J.P.: Naissance et postérité de l’intégrale de Lebesgue. Gazette des Mathématiciens
89, 5–20 (2001)
40. Krantz, S.G., Parks, H.R.: The Geometry of Domains in Space. Birkhäuser, Boston (1999)
41. Krantz, S.G., Parks, H.R.: The Implicit Function Theorem. History, Theory and Applications.
Birkhäuser, Boston (2002)
42. Lebesgue, H.: Sur une généralisation de l’intégrale définie. C. R. Acad. Sci. Paris 132, 1025–
1028 (1901)
43. Lebesgue, H.: Intégrale, longueur, aire. Ann. Mat. Pura Appl. 7, 231–359 (1902)
44. Lebesgue, H.: Remarques sur la définition de l’intégrable. Bull. Sci. Math. 29, 272–275 (1905)
45. Lebesgue, H.: Leçons sur l’intégration et la recherche des fonctions primitives, 2nd edn.
Gauthier-Villars, Paris (1928)
46. Lebesgue, H.: Notices d’histoire des mathématiques. L’Enseignement Mathématique, Genève
(1958)
47. Lebesgue, H.: Measure and the Integral. Holden-Day, San Francisco (1966)
48. Leray, J.: Sur le mouvement d’un liquide visqueux emplissant l’espace. Acta Math. 63, 193–
248 (1934)
49. Mawhin, J.: Analyse, fondements techniques, évolution, 2nd edn. De Boeck, Paris-Bruxelles
(1997)
50. Mawhin, J.: Henri Poincaré et les équations aux dérivées partielles de la physique
mathématique. In: L’héritage scientifique de Poincaré, pp. 278–301. Belin, Paris (2006)
51. Maz’ya, V.: Sobolev Spaces with Applications to Elliptic Partial Differential Equations.
Springer, Berlin (2011)
52. Maz’ya, V., Shaposhnikova, T.: Jacques Hadamard, a Universal Mathematician. AMS, Provi-
dence (1998)
53. Meyer, Y.: Comment mesurer les surfaces? Gaz. Math. 109, 23–36 (2006)
54. Milnor, J.: Topology from the Differentiable Viewpoint. The University Press of Virginia,
Charlottesville (1965)
55. Naumann, J.: Remarks on the Prehistory of Sobolev Spaces, prépublication (2002)
56. Nirenberg, L.: On elliptic partial differential equations. Ann. Scuola Norm. Sup. Pisa Cl. Sci.
13, 116–162 (1959)
References 207

57. Pier, J.P.: Histoire de l’intégration. Masson, Paris (1996); Mathématiques entre savoir et
connaissance. Vuibert, Paris (2006)
58. Pietsch, A.: Ein elementarer Beweis des Darstellungssatzes für Distributionen. Math. Nachr.
22, 47–50 (1960)
59. Poincaré, H.: Sur les équations aux dérivées partielles de la physique mathématique. Am. J.
Math. 12, 211–294 (1890)
60. Poincaré, H.: Sur les équations de la physique mathématique. Rendiconti del Circolo Matem-
atico di Palermo 8, 57–155 (1894)
61. Riesz, F.: Sur les opérations fonctionnelles linéaires. C. R. Acad. Sci. Paris 149, 974–977
(1909)
62. Riesz, F., Nagy, B.S.: Leçons d’analyse fonctionnelle, 3rd edn. Gauthier-Villars, Paris (1955)
63. Riesz, M.: Sur les ensembles compacts de fonctions sommables. Acta Szeged Sect. Math. 6,
136–142 (1933)
64. Roselli, P., Willem, M.: A convexity inequality. Am. Math. Mon. 109, 64–70 (2002)
65. Roselli, P., Willem, M.: The Lebesgue integral immediately after calculus. Travaux Mathé-
matiques 13, 61–70 (2002)
66. Royden, H.: Aspects of constructive analysis. Contemp. Math. 39, 57–64 (1983)
67. Saks, S.: Theory of the Integral. Warsaw–Lvov (1937)
68. Sard, A.: The measure of the critical values of differentiable maps. Bull. Am. Math. Soc. 48,
883–890 (1942)
69. Schwartz, L.: Généralisation de la notion de fonction, de dérivation, de transformation de
Fourier et applications mathématiques et physiques. Annales de l’Univ. de Grenoble 21, 57–74
(1945)
70. Schwartz, L.: Théorie des distributions et transformation de Fourier. Annales de l’Univ. de
Grenoble 23, 7–24 (1947)
71. Schwartz, L.: Théorie des distributions. Hermann, Paris (1966)
72. Sierpinski, W.: Leçons sur les nombres transfinis. Gauthier-Villars, Paris (1928)
73. Smets, D., Willem, M.: Partial symmetry and asymptotic behavior for some elliptic variational
problems. Calc. Var. Part. Differ. Equat. 18, 57–75 (2003)
74. Sobolev, S.L.: Le problème de Cauchy dans l’espace des fonctionnelles. Comptes Rendus de
l’Académie des Sciences de l’U.R.S.S. 3, 291–294 (1935)
75. Sobolev, S.L.: Méthode nouvelle à résoudre le problème de Cauchy pour les équations linéaires
hyperboliques normales. Recueil Mathématique 43, 39–71 (1936)
76. Sobolev, S.L.: Sur quelques évaluations concernant les familles de fonctions ayant des dérivées
à carré intégrable. Comptes Rendus de l’Académie des Sciences de l’U.R.S.S. I(X), 279–282
(1936)
77. Sobolev, S.L.: Sur un théorème de l’analyse fonctionnelle. Comptes Rendus de l’Académie des
Sciences de l’U.R.S.S. XX, 5–9 (1938)
78. Sobolev, S.L.: Some Applications of Functional Analysis in Mathematical Physics. American
Mathematical Society, Providence (1991)
79. Steiner, J.: Einfache Beweise der isoperimetrischen Hauptsätze. J. Reine Angew. Math. 18,
281–296 (1838)
80. Stieltjes, T.: Recherches sur les fractions continues. Annales Fac. Sci. Toulouse 8, 1–122 (1894)
81. Taylor, A.E.: A study of Maurice Fréchet. Arch. Hist. Exact Sci. 27, 233–235 (1982); 34, 279–
280 (1985); 37, 25–76 (1987)
82. Tonelli, L.: Sulla rettificazione delle curve. Atti R. Accad. delle Sci. di Torino 63, 783–800
(1908)
83. Tonelli, L.: Sur la quadrature des surfaces. C. R. Acad. Sci. Paris 182, 1198–1200 (1926)
84. Van Schaftingen, J.: Explicit approximation of the symmetric rearrangement by polarizations.
Arch. Math. 93, 181–190 (2009)
85. Van Schaftingen, J., Willem, M.: Symmetry of solutions of semilinear elliptic problems. J. Eur.
Math. Soc. 10, 439–456 (2008)
86. Vitali, G.: Sul problema della misura dei gruppi di punti di una retta. Bologna (1905)
87. Wolontis, V.: Properties of conformal invariants. Am. J. Math. 74, 587–606 (1952)