Lagrange and The Calculus of Variations
Lagrange and The Calculus of Variations
Lagrange and The Calculus of Variations
DOI 10.1007/s40329-014-0049-x
Abstract This paper gives a simple presentation in reader can also consult [2–4] for a more advanced episte-
modern language of the theory of calculus of variations as mological analysis, as well as [9] regarding biographical
invented by Euler and Lagrange, as well as an account of elements.
the history of its invention. The discussion will show how it
serves to solve simple optimization problems and how it
has influenced mathematics, physics and related fields up to 2 The first steps
the present day.
In 1754, at the age of eighteen, Lagrange read the article
Keywords Lagrange Euler Calculus of variations ‘‘Une méthode pour trouver des lignes courbes jouissant de
Mechanics Partial differential equations propriétés de maximum ou de minimum’’ [5] by the great
Euler. Inspired by this lesson, he obtained his first original
mathematical result, and dared to communicate it by letter
1 Introduction to Euler, already at the time a leading figure in science. His
letter remained unanswered.
Together with Euler, Lagrange is the inventor of the cal- Lagrange, however, was undeterred, and continued to
culus of variations, a simple and elegant idea that revolu- reflect on Euler’s article. In 1755 he wrote a second letter
tionised the way of solving problems of optimisation, the to Euler in which he described the new method that he had
formulation of classical physics, and had an enormous developed, that is, his own manner for dealing with the
influence on how partial derivatives equations are viewed. problem examined by Euler. That method would be named
Used today almost as much as ordinary differential calcu- by Euler himself in one of his letters: ‘‘calculus of varia-
lus, with all sorts of domains of application, the calculus of tions’’. This time, Euler replied to Lagrange, in terms of
variations forms the basis of the mechanics known as praise:
Lagrangian, without which modern physics could not exist.
Votre solution du problème des isopérimètres ne
Here we will look at how Lagrange was led to his
laisse rien à désirer, et je me réjouis que ce sujet, dont
interest in these problems, discuss the simplest elements
je m’étais presque seul occupé depuis les premières
and principles of his discovery, and finally, show the
tentatives, ait été porté par vous au plus haut degré de
repercussions they have had up to the present day.
perfection. L’importance de la matière m’a excité à
The history of the calculus of variations and Lagrange’s
en tracer, à l’aide de vos lumières, une solution
contribution to it is well documented. A good point of
analytique à laquelle je ne donnerai aucune publicité
departure is the work of Catherine Goldstein [6]. The
jusqu’à ce que vous-même ayez publié la suite de vos
recherches, pour ne vous enlever aucune partie de la
S. Serfaty (&) gloire qui vous est due.
Laboratoire Jacques-Louis Lions, UPMC, 4 place Jussieu,
75005 Paris, France (Your solution to the isoperimetric problem leaves
e-mail: [email protected] nothing to be desired, and I rejoice that this subject,
123
40 Lett Mat Int (2014) 2:39–46
123
Lett Mat Int (2014) 2:39–46 41
The solution is an (inverse) cycloid, that is, a curve n’a pas toute la simplicité qu’on peut désirer dans un
described by a fixed point on a circle that rolls without ouvrage de pure analyse. L’auteur lui-même le fait
slipping on a straight line (Fig. 3). sentir par ces paroles: ‘‘il semble désirable de trouver
Before Euler, the solutions were all of a geometric une méthode indépendante de la géométrie’’
nature, and very ad hoc.
(an original work and one throughout which there
In [5], Euler is the first to propose a systematic treatment
shines a profound knowledge of calculus. However,
of this kind of problem: instead of concerning himself only
as ingenious and fertile as its method is, we must
with the problem of the brachistochrone, he seeks a method
admit that it has not all the simplicity that one might
to find a curve that minimises or maximises any quantity
wish in a work of pure analysis. The author himself
expressed by an integral, and to derive the equation that
seems to feel this, by his words: ‘‘it seems desirable
must be satisfied by the minima. That equation, which has
to find a method that is independent of geometry’’).
become known as the Euler–Lagrange equation, takes for
example the form (in the notation of physics): Lagrange’s accomplishment was that he was able to find
anew the results of Euler while freeing himself from geo-
d oL oL
¼0 ð1Þ metric intuition (displacing the graph of the function), and
dt oq_ oq
replacing it with a ‘‘machinery’’ of operations of calculus.
We will discuss this in more detail equation below. He had seen that Euler’s calculus led to defining a new type
Euler is also the first to transform this question of Ber- of differential calculus, in which the objects are no longer
noulli’s into a domain of mathematics and to formulate the functions of real variables, but functions of functions
‘‘principle of least action’’ (although the concept had (today called functionals). This crucial conceptual leap
already been introduced by Maupertuis). (seeing the functions themselves as variables) is truly due
More precisely, he wrote the problem as that of mini- to Lagrange, and can be seen as one of his fundamental
mising an integral quantity of the form: contributions.
Z The new calculus of Lagrange consists in defining a new
Z x; u; Du; D2 u dx notion of derivative or of differential, this time for an
integral expression on curves, which are none other than
where u(x) represents the curve, and Du and D2u are the the particular case of functions of functions. This new
first and second derivatives of u (he even considers the case differential, he denotes d (and his notion of differential is
of dependencies and of derivatives of arbitrary order). He thus called the ‘‘d-calculus of Lagrange’’), in order to
writes: distinguish it from the ordinary derivative of the differen-
tial calculus of Newton or Leibniz; for example, in his
dZ ¼ Mdx þ Ndu þ PdðDuÞ þ QdðD2 uÞ;
notation the differential of Z is denoted as dZ. Here is what
calculating the variation of Z when one moves the graph of he says in the Essai [7]:
u from u(x) to u(x) ?du(x), he obtains that the relation
Maintenant voici une méthode qui ne demande
N ¼ dP
dx must be satisfied by a minimum. In other words, qu’un usage fort simple des principes du calcul
and since N ¼ oZou , he obtains the equation différentiel et intégral; mais avant tout je dois
oZ o oZ avertir que, comme cette méthode exige que les
¼ ; mêmes quantités varient de deux manières diffé-
oui ox oðDuÞ
rentes, pour ne pas confondre ces variations, j’ai
which is Eq. (1). Below we will see a more precise way to introduit dans mes calculs une nouvelle caractéris-
derive it. tique d. Ainsi d exprimera une différence de Z qui
Even though Euler’s treatment is systematic for the first ne sera pas la même que dZ, mais qui sera cepen-
time, his argumentation for deriving the equation called dant formée par les mêmes règles…
‘‘Euler–Lagrange’’ still remains somewhat geometric in
(Now here is a method that requires only a very
spirit. This leaves Lagrange dissatisfied, and pushes him to
simple use of the principles of differential and
improve the method.
integral calculus, but first of all I must remark that,
Lagrange, who at the time was 19 years old, was greatly
as this method requires that the same quantities vary
influenced by this work of Euler’s. In his article [7], he
in two different ways, so as not to confuse these
described it as:
variations, I introduced in my calculations a new
ouvrage original et qui brille partout d’une profonde character d. Thus d expresses a difference of
science de calcul. Cependant, quelque ingénieuse et Z which is not the same as dZ, but will be however
féconde que soit sa méthode, il faut avouer qu’elle formed by the same rules…).
123
42 Lett Mat Int (2014) 2:39–46
123
Lett Mat Int (2014) 2:39–46 43
123
44 Lett Mat Int (2014) 2:39–46
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where 0 indicates the derivative with respect to the spatial Multiplying by 2gyð1 þ ðy0 Þ2 Þ and simplifying, we
variable x, that is, that the Euler–Lagrange equation obtain
associated with the problem of minimisation is satisfied.
yð1 þ ðy0 Þ2 ¼ const;
It is important to underline that this equation is not
equivalent to the fact that a minimising function exists; the differential equation of an inverted cycloid, generated
rather, it characterises the fact that there exists a critical by a circle of diameter const.
point of F. Euler and Lagrange do not appear to be too We shall conclude with a fundamental example, because
concerned about making this distinction, nor the need to it is the one that led to the whole development of the
underline this point. Méchanique analitique [8]: that of the principle of least
Replacing in our notation x for t, z for q and p for q, _ we action. The action along a path X(t) for a particle in a
find once again the expression (1), the most used in physics. potential V is defined by
In fact, the method proceeds by the function of several Zt1
1 2
variables: in lieu of varying the curve, one varies the FðXðtÞÞ ¼ mjX 0 j ðtÞ VðXðtÞÞ dt Lðz; pÞ
2
‘‘surfaces’’. t0
Lagrange remarks that this method can be applied 1
¼ mj pj2 VðzÞ
whether the ends are fixed or not. Let us now see the 2
application of this result to the examples given previously. where we recognise the integral of the difference between a
For the problem of geodesy, applying the results to kinetic energy and a potential energy. Since oL 0
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi oz ¼ V ðzÞ
Lðz; pÞ ¼ 1 þ j pj2 , since oL op ¼
p
pffiffiffiffiffiffiffiffiffiffi
2
, we find that the and oL
o8 e ¼ mp, the Euler–Lagrange equation is
1þj pj
associated Euler–Lagrange equation is: mX 00 ðtÞ ¼ V 0 ðXðtÞÞ;
u0 which is none other than Newton’s law ~ F ¼ mc~ with ~
F the
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2 ¼ const;
1 þ ðu0 Þ force, m the mass, and ~ c the acceleration. Here we find a
0 first form of the principle of least action, which says that
which implies that u = const. We find that in Euclidean the particles follow the course that minimises the action.
space the curve that gives the shortest path must be a Incidentally, this point is not clear, because we recall that
straight line! the Euler–Lagrange equation does not necessarily charac-
This works in higher dimensions by the area of a graph, terize minimizers; we should say rather that the particles
and we obtain that obey Newton’s law follow the trajectories that are
0 1
critical points of the action.
B ru C
div @qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiA ¼ 0;
2 4.1 The extensions due to Lagrange
1 þ jruj
which is the equation for minimal surfaces, derived for the In his article ‘‘Essai d’une nouvelle methode pour determiner
first time by Lagrange in [7]! It is the equation associated les maxima et les minima des formules integrals indefinies’’
with Plateau’s problem of finding the surface of minimal [7], Lagrange also introduced the method of multipliers—
area that has a given boundary. today called Lagrange multipliers—which makes it possible
In the case of the brachistochrone, reprising the equation to treat problems with constraints. An example of such a
qffiffiffiffiffiffiffiffi problem is the isoperimetric problem: find a curve of given
2
Lðz; pÞ ¼ 1þp oL
2gz, we calculate op ¼
p
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi. Moreover,
2
2gzð1þp Þ length that delimits a domain of maximum area (in the
we can see that the Euler–Lagrange equation implies that plane); the solution to this classic problem is the circle.
dL Another example, a variant of the geodetic problem, is the
LðuðxÞ; u0 ðxÞÞ u0 ðxÞ ðuðxÞ; u0 ðxÞÞ ¼ const: geodetic problem with an obstacle: find the shortest path
dp
from A to B that goes around an obstacle, a bounded set X in
(It is sufficient to derive this expression with respect to x space. Lagrange’s method of variation tells us that the geo-
and to verify that the derivative is null). Inserting the desic ‘‘takes off’’ from the obstacle in such a way that it
expression for L and oL op and supposing for the sake of remains tangent to it. This is a particular case of a variational
simplicity that yA = 0, we find problem that was well-studied in the twentieth century,
sffiffiffiffiffiffiffiffiffiffiffiffiffiffi called the ‘‘obstacle problem’’ (Fig. 5).
1 þ y02 ðy02 Þ Examples in economics of the problem under constraints
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ const:
2gy abound: maximising the utility function under constraints
2gyð1 þ ðy0Þ2 Þ
of resources, etc.
123
Lett Mat Int (2014) 2:39–46 45
123
46 Lett Mat Int (2014) 2:39–46
2004, a EURYI award in 2007, the Henri Poincaré Prize of the IAMP
Sylvia Serfaty is a professor of
in 2012, and the Grand Prix Mergier-Bourdeix of the Académie des
mathematics at the Université
sciences de Paris in 2013. Her work has addressed partial differential
Pierre et Marie Curie–Paris 6,
equations and variational models coming mostly from physics, and
and a Global Distinguished Pro-
particularly vortices and phase transitions in the Ginzburg–Landau
fessor at the Courant Institute of
model of superconductivity, models for micromagnetics, and Cou-
Mathematical Sciences, New
lomb systems (photo copyright Olivier Boulanger).
York University. She studied at
the École Normale Supérieure in
Paris and earned her Ph.D. in
mathematics in 1999 at the Uni-
versity of Paris Sud–Orsay. She
was then a CNRS researcher, and
on the faculty at the Courant
Institute of New York University
from 2001 to 2008. She was an
invited speaker at the Interna-
tional Congress of Mathematicians in 2006. She received an NSF
Career award in 2003, a European Mathematical Society Prize in
123