This article was downloaded by: [Marshall University]

On: 16 August 2013, At: 20:27

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

International Journal of Mathematical

Education in Science and Technology
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/tmes20

A short biography of Joseph Fourier

and historical development of Fourier
series and Fourier transforms
a
Lokenath Debnath
a
Department of Mathematics, University of Texas – Pan American,
1201 W. University Drive, Edinburg, TX 78539, USA
Published online: 01 Dec 2011.

To cite this article: Lokenath Debnath (2012) A short biography of Joseph Fourier and historical
development of Fourier series and Fourier transforms, International Journal of Mathematical
Education in Science and Technology, 43:5, 589-612, DOI: 10.1080/0020739X.2011.633712

To link to this article: http://dx.doi.org/10.1080/0020739X.2011.633712

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
 International Journal of Mathematical Education in
Science and Technology, Vol. 43, No. 5, 15 July 2012, 589–612

A short biography of Joseph Fourier and historical development

of Fourier series and Fourier transforms
Lokenath Debnath*

Department of Mathematics, University of Texas – Pan American, 1201 W.

University Drive, Edinburg, TX 78539, USA
(Received 10 June 2011)
Downloaded by [Marshall University] at 20:27 16 August 2013

The profound study of nature is the most fertile source of mathematical discoveries. Not
only does this study, by offering a definite goal to research, have the advantage of
excluding vague questions and futile calculations, but it is also a sure means of moulding
analysis itself, and discerning those elements in it which it is still essential to know and
which science ought to conserve. These fundamental elements are those which recur in
all natural phenomena.
Joseph Fourier
pure mathematics enables us to discover the concepts and laws connecting them, which
gives us the key to the understanding of the phenomena of nature.
Albert Einstein

This article deals with a brief biographical sketch of Joseph Fourier, his
first celebrated work on analytical theory of heat, his first great discovery
of Fourier series and Fourier transforms. Included is a historical
development of Fourier series and Fourier transforms with their properties,
importance and applications. Special emphasis is made to his splendid
research contributions to mathematical physics, pure and applied mathe-
matics and his unprecedented public service accomplishments in the history
of France. This is followed by historical comments about the significant
and major impact of Fourier analysis on mathematical physics, probability
and mathematical statistics, mathematical economics and many areas of
pure and applied mathematics including geometry, harmonic analysis,
signal analysis, wave propagation and wavelet analysis. Special attention is
also given to the Fourier integral formula, Brownian motion and stochastic
processes and many examples of applications including isoparametric
inequality, everywhere continuous but nowhere differentiable functions,
Heisenberg uncertainty principle, Dirichlets’ theorem on primes in
arithmetic progression, the Poisson summation formula and solutions of
wave and diffusion equations. It is also shown that Fourier coefficients cn(t)
in the Fourier expansion of a scalar field (, t) satisfy equations of the
simple harmonic motion.

*Email: [email protected]

ISSN 0020–739X print/ISSN 1464–5211 online

ß 2012 Taylor & Francis
http://dx.doi.org/10.1080/0020739X.2011.633712
http://www.tandfonline.com
 590 L. Debnath

Keywords: Joseph Fourier; Fourier series; Fourier transforms; Fourier’s

diffusion equation; greenhouse effect; wave equation
AMS Subject Classifications: 01A; 35K05

1. Joseph Fourier (1768–1830) and education

Jean-Baptiste Joseph Fourier was born on 21 March 1768 in an ancient town of
Auxerre in France. This small town has many cultural, religious, political and
educational traditions. It was one of the foremost centres of teaching and learning in
France in the ninth and tenth centuries. It is located on the bank of the river Yonne
with many magnificent tall building including the famous ancient Clock Tower, the
even more famous Abbey St. Germain founded by St. Germain himself in the fifth
century A.D., and the Gothic Cathedral St. Etienne. It became more famous as the
Downloaded by [Marshall University] at 20:27 16 August 2013

birthplace of Joseph Fourier who is one of the most brilliant mathematical scientists
France has ever produced.
Joseph’s father was a master tailor of Auxerre and his mother was a typical
housewife. When he was nine- or ten-years old, both his parents died. He first
entered the Pallais elementary school to receive lessons in Latin and French. He was
a gifted student in both languages, but showed extraordinary talents for mathematics
and mechanics. At the age of 13, he had completed an extensive study of six volumes
of Etienne Bézout’s (1730–1783) Cours de mathe´matique. In 1783, he received the first
prize for his study of C. Bossut’s (1730–1814) Me´canique. At a very young age,
Joseph presented a research paper on algebraic equations at the Académie Royale
des Sciences. Although he had early interest in training to become a priest, he gave
up his idea of studying theology. He was then admitted to the progressive École
Royale Militaire to study science and mathematics. In 1790, he became a
mathematics teacher at the Benedictine College, École Royale Militaire of
Auxerre, where he was a student. At the same time, he took a major role in his
own district in promoting the French Revolution in 1789–1794, and got involved in
unhappy political problems. Unfortunately, Fourier was arrested twice by the order
of the Committee for Public Safety for his possible revolutionary and terrorist
activities in the Revolution in Auxerre during 1793–1794 and subsequently released.
During the storms of the Revolution, Fourier participated in local revolutionary
politics which led directly to his two imprisonments from which he suffered with deep
pain. However, he had then an intense desire to get out of all political troubles, but
he was determined to fight for his life and freedom. In fact, the revolution left its
mark on Fourier’s life, but it was directly responsible for opening up a successful
research career for him in mathematics and physics in which he distinguished himself
later in his life. Later in 1794, Joseph was nominated to study at the École Normale
in Paris, which opened in 1795 as a model school for training teachers in France. He
joined this teacher-training school and got a rare opportunity to meet some of the
foremost French mathematicians of that time including J.L. Lagrange (1736–1813),
P.S. Laplace (1749–1827) and Gaspard Monge (1746–1818). In September 1795,
Fourier was appointed as an assistant lecturer at the École Polytechnique to support
the teaching of Lagrange and Monge. In a short period of two years, he succeeded
Lagrange to occupy the Chair of Analysis and Mechanics at the Polytechnique in
1797. In spite of his cordial relationship with Lagrange, Laplace and Monge, and his
 International Journal of Mathematical Education in Science and Technology 591

exceptional teaching record, Fourier’s research record at the time was not that
outstanding by any standard.
In 1798, Fourier was selected to join Napoleon Bonaparte’s (1769–1821)
Egyptian expedition as scientific advisor to liberate unhappy Egyptian people
from deep social and political problems and to provide them with all the benefits of
European civilization. He remained in this position until the downfall of Napoleon in
1815. During his occupation of Egypt, Napoleon established the Institute d’Egypt in
Cairo with Monge as its President and Fourier as Permanent Secretary. Fourier
continued his position during the entire occupation of Egypt and was fully
responsible for collating all literary and scientific discoveries made during that time.
In addition, he was very active in administering the archaeological research of the
Cairo Institute. This position led Fourier to publish the Description de l’Egypte
(Description of Egypt) with an elegant preface, partly edited by Napoleon himself.
In 1799, Napoleon returned to Paris to hold the absolute power of France and
Downloaded by [Marshall University] at 20:27 16 August 2013

subsequently, he crowned himself Emperor of the French. In 1801, Fourier also came
back to Paris to resume his position as Professor of Analysis at the École
Polytechnique. After his return from Egypt in 1802, Fourier was appointed Prefect of
Isére in Grenoble with many major and varied administrative duties and respon-
sibilities. As Prefect from 1802 to 1815, Fourier was totally successful in planning,
management and the draining operation of the largest swamps of Bourgoin, as the
swamps had been fully responsible for annual epidemics of dangerous fever causing
deaths of a large number of the surrounding inhabitants at a young age. After
completion of the draining of the swamps in 1812, the annual epidemics of fever
stopped leading to a marked improvement in health problems of the inhabitants.
This was one of the greatest public and social service accomplishments of Fourier.
His other major accomplishment as Prefect was the planning of a very long
spectacular highway from Grenoble to Turin, of which only the French Section was
built. Despite his unfriendly relationship with Napoleon, Napoleon awarded Fourier
the title of Baron in recognition of his major accomplishments as Prefect.
During 1787–1810, a series of personal letters between Fourier and C.L. Bonard
(1765–1819), mathematics teacher at Auxerre provided the unique opportunity to
establish a real and permanent friendship between them. These letters were full of
their many personal and family experiences and reminiscences of the old days,
especially during the revolution in Auxerre. As a family friend and true admirer of
Fourier, Bonard was always very happy and proud of Fourier’s mathematical and
other achievements. Unfortunately, Bonard died in 1819 just three years before the
publication of Fourier’s masterpiece work on Theory of Heat.
In 1827, like Jean D’Alembert (1717–1783) and Laplace before him, Fourier was
elected to the literary Académie Francaise. In the same year, after the death of
Laplace, Fourier was also elected President of the Conseil de perfectionnement of the
École Polytechnique. Several foreign scientific organizations including the Royal
Society of London and the Royal Swedish Academy of Sciences elected Fourier as
honorary member for his outstanding contributions to mathematical sciences.
Throughout his life, Fourier’s general health was not very good. Towards the end
of his life, he began to display some unusual symptoms which were thought to have
possibly been contracted during his stay in Egypt from extreme hot climate. In early
May of 1830, Fourier suffered from a heart attack and his condition gradually
deteriorated until he died on May 16 at the age of 62. He was buried at the cemetery
of Père Lachaise in Paris. His gravestone was decorated with an Egyptian motif to
 592 L. Debnath

recognize his outstanding service as Permanent Secretary of the Cairo Institute and
his collation of the landmark Description de l’Egypte.

2. Fourier and heat conduction

During his stay in Grenoble, Fourier found the time and intellectual energy to
discover the theory of heat conduction based on his new mathematical ideas,
methods, experiments and observations. He completed his classic memoir on the
conduction of heat entitled ‘On the propagation of heat in solid bodies’, and
submitted it to the Academy of Sciences of Paris in 1807 for a research prize. His
memoir was judged by a great scientific committee consisting of Lagrange, Laplace,
Monge, A.M. Legendre (1752–1833) and S.L. Lacroix (1765–1843). The committee
was generally impressed by the importance and novelty of Fourier’s work, but it was
Downloaded by [Marshall University] at 20:27 16 August 2013

split on the overall quality and rigour of the mathematical treatment involving the
use of trigonometric (Fourier) series expansions of functions without adequate
justification. The controversy created by his work on trigonometric expansions of
functions was due to his demonstration of the paradoxical property of equality in a
finite interval between algebraic results of a totally different form. Another major
criticism of his memoir was about his derivation of the heat equation in a continuous
solid based on inadequate physical principles. Unfortunately, his memoir was
rejected by the committee for the prize. Although some of his contemporary
mathematicians vociferously objected to his revolutionary work, the Academy was
generally impressed by the subject matter of Fourier’s memoir and encouraged him
to resubmit it for a Grand Prize to be awarded in 1812. At the same time, S.D.
Poisson (1781–1840) and J.B. Biot (1774–1862) who were his rivals, expressed serious
concerns with major criticisms of Fourier’s work on his mathematical and physical
treatment of the theory of heat in solids. Surprisingly, Biot praised Poisson’s less
original work on the same subject. Poisson’s harsh criticism is quoted below in full:
As the partial differential equation to which it corresponds is linear and has constant
coefficients, one can also satisfy it by an integral composed of an infinity of exponentials
of sines and cosines containing an infinite number of arbitrary constants: this integral is
contained in the preceding one; but it would be difficult to decide a priori if it has the
same degree of generality and if it can replace it identically, something which necessarily
throws doubt and obscurity on all solutions deduced from this second form of the
integral. M. Fourier, who did not go beyond a solution of this kind, remarks himself
that it is similar to that which Daniel Bernouilli gave to the problem of vibrating strings;
but it is well known that Euler, d’Alembert, and Lagrange, who occupied themselves at
the same time with the same problem, and who differed among themselves on various
points, were at one nevertheless in regarding Bernouilli’s solutions as incomplete and
less general than that containing arbitrary functions. This is not true of the formulas of
M. Fourier: I am sure that all the results he obtains are correct; but against his analysis
can be advanced the same objections as those advanced against that of Bernouilli and
repeated in other similar cases.
In general it seems to me that whenever an unknown quantity depends on a partial
differential equation, and when its values should reduce in fact to a sum of particular
integrals, the only way of disposing of all doubts and retaining for the mathematical
certainty result is not to suppose in advance such a form for the unknown quantity, but
to deduce it, on the contrary, from the general integral by a succession of direct and
rigorous transformations. This is what I have attempted to do in this memoir . . .
. . . I leave it to mathematicians to judge if I have attained the end that I have set myself.
 International Journal of Mathematical Education in Science and Technology 593

Fourier had real talent, and extraordinary courage to defend his work in an
absolutely unprecedented manner. His devastating response to Poisson and Biot’s
unfounded criticisms of his solution was contained in his unpublished Historical
Pre´cis, and in his letters to Laplace around 1808–1809 with a note on certain analytic
expressions in connection with the equations of the theory of heat. Thereafter, all
criticisms seemed to have subsided once for all and barely tarnished his permanent
reputation.
Fourier continued his research and resubmitted his revised memoir in 1811 with
some additional new material on the cooling of infinite solids, radiant and terrestrial
heat, comparison of his theory with experimental observations and equations of the
movement of heat in fluids. In spite of some disagreements, Fourier was awarded the
Grand Prize for his work in 1812, but his memoir was not recommended for
publication in the Me´moires of the Academy. In 1817, he was elected member of the
Académie des Sciences by an overwhelming majority. He continued to revise his
Downloaded by [Marshall University] at 20:27 16 August 2013

memoir for publication. In 1822, J.B. Delambre (1749–1822), who was Permanent
Secretary of the mathematics and physics division of the Académie des Sciences died,
and Fourier was elected as the Permanent Secretary. Fourier remained in this
position until his death in 1830. Shortly after he became the Secretary, the Académie
published his masterpiece work entitled La The´orie Analytique de la Chaleur (The
Analytical Theory of Heat) based on Isaac Newton’s (1642–1727) law of cooling, that
is, the flow of heat between two adjacent molecules is proportional to extremely
small temperature difference. In his preface of this great book, Fourier described his
own assessment of the work as follows:
First causes are not known to us, but they are subjected to simple and constant laws
that can be studied by observation and whose study is the goal of Natural
Philosophy . . . Heat, like gravity penetrates every substance of the universe; its rays
occupy all regions of space. The object of our work is to set forth the mathematical laws
that this element obeys . . . But whatever the extent of the mechanical theories, they do
not apply at all to the effects of heat. They constitute a special order of phenomena
that cannot be explained by principles of movement and of equilibrium . . .
The differential equations of the propagation of heat express the most general
conditions, and reduce physical questions to problems in pure analysis that is the proper
object of the theory.
His analytical theory of heat was highly mathematical based on a general
equation of propagation of heat with appropriate initial and boundary conditions of
different kinds. Fourier’s fundamental heat equation for the temperature distribu-
tion, u ¼ u(x, t) is given by

ut ¼  r2 u, ð2:1Þ
k
where x ¼ (x, y, z),  ¼ c is the diffusivity constant, k is the heat conductivity, c is the
@2 @2 @2
specific heat,  is the density and r2 ¼ @x 2 þ @y2 þ @z2 is the three-dimensional

Laplacian. Thus, the heat equation is of the first order in time derivative and of the
second order in space derivatives. This was a striking contrast with almost all
dynamical equations including the wave equation in strings or air which are of
second order in both time and space derivatives.
One of the French scientific philosophers and close colleagues of Fourier,
Auguste Compte (1798–1857) attended a course of lecture of Fourier on the role of
the positivist philosophy of science. In his published book entitled Cours de
 594 L. Debnath

Philosophie positive dedicated to Joseph Fourier, Compte expressed his eloquent and
complimentary views on Fourier’s celebrated theory of heat as follows:
In fact, in this work, of which the philosophical characteristic is so eminently positive
the most important and precise laws of thermal phenomena are discovered without the
author having once enquired about the intimate nature of heat.
Subsequently, Henri Poincaré (1854–1912), an extremely versatile and creative
mathematician and mathematical physicist, made a delightful comment on Fourier’s
fundamental work as follows:
Fourier’s theory of heat is one of the first examples of the application of analysis to
physics. Starting from simple hypotheses, which are nothing but generalized facts,
Fourier deduced from them a series of consequences which together make up a complete
and coherent theory. The results which he obtained are certainly interesting in
themselves, but what is still more interesting is the method which he used to arrive at
them and which will always be a model for all those who wish to cultivate any branch of
Downloaded by [Marshall University] at 20:27 16 August 2013

mathematical physics.
Even with lack of rigour in Fourier’s classic work on the analytical theory of
heat, a eminent British physicist, Lord Kelvin (1824–1907) regarded the work as
Fourier’s permanent legacy and called it ‘a great mathematical poem.’

3. The Fourier integral theorem

This section includes some of Fourier’s major contributions to other areas of
mathematics and physics and their foremost impact on mathematical physics,
mathematical economics, probability, statistics, pure and applied mathematics.
During 1817–1818, he made notable contributions to the theory of Fourier
transforms and their applications to partial differential equations including the
heat and the wave equations. His 1822 monumental treatise on the Analytical Theory
of Heat provided the modern mathematical theory of conduction of heat, Fourier
series and Fourier integrals with many new examples of applications. In his treatise,
he formulated a remarkable theorem which is now universally known as the Fourier
Integral Theorem which states that an arbitrary function f(x) has the Fourier double
integral representation in the form
Z Z1
1 1 ikx
fðxÞ ¼ e dk eik fðÞd: ð3:1Þ
2 1 1

This representation shows that f(x) has been decomposed into an infinite number of
waves
R of different wavenumbers k (or wavelengths  ¼ 2 k ), and amplitude
1 1 ik
2 1 e f ð  Þd, whereas a given function can be represented by a Fourier series
in terms of an infinite but discrete set of wave components.
This theorem has been expressed in several slightly different forms to better adapt
it for many particular applications. It has been recognized, almost from the start,
however, that the form which best combines mathematical simplicity and complete
generality makes use of the exponential oscillating function exp(ikx). Indeed, the
Fourier integral formula (3.1) is regarded as one of the most fundamental results of
modern mathematical analysis, and it has widespread physical and engineering
applications. The generality, applicability and importance of the theorem is
eloquently expressed by Lord Kelvin and Peter Guthrie Tait (1831–1901) as
follows: ‘. . . Fourier’s Theorem, which is not only one of the most beautiful results of
 International Journal of Mathematical Education in Science and Technology 595

modern analysis, but may be said to furnish an indispensable instrument in

the treatment of nearly every recondite question in modern physics. To mention
only sonorous vibrations, the propagation of electric signals along a telegraph wire
and the conduction of heat by the earth’s crust, as subjects in their
generality intractable without it, is to give but a feeble idea of its importance’.
This integral formula (3.1) is usually used to define the classical Fourier transform
of a function and the inverse Fourier transform. No doubt, the scientific
achievements of Joseph Fourier have not only provided the fundamental basis for
the study of heat equation, Fourier series and Fourier integrals, but also for the
modern developments of the theory and applications of the partial differential
equations.
Although Fourier is most celebrated for his work on the conduction of heat,
Fourier series and Fourier transforms, the new mathematics he created has proved to
be very useful in a wide variety of ways. His superb mastery of analytical treatment
Downloaded by [Marshall University] at 20:27 16 August 2013

and physical intuition played a fundamental role in his success. In mathematical

physics, Fourier’s contributions have proved to be much more useful than those of
his renowned contemporaries including Lagrange, Laplace, Poisson, Biot and
Augustin Cauchy (1789–1857). Both Cauchy and Poisson successfully applied
Fourier’s universal mathematical analysis to the subject of linear water waves and to
other areas. It has also been a remarkable fact that Fourier analysis eventually
resolved the long-standing controversy over the mathematical treatment of the
famous vibrating string problem which was vigorously investigated by several
eminent mathematical scientists including Daniel Bernoulli (1700–1782), Leonhard
Euler (1707–1783), D’Alembert and Lagrange. Fourier’s work was also remarkably
influential for the comprehensive development of James Clerk Maxwell’s
(1831–1879) electromagnetic theory and kinetic theory of gases. In 1864, Maxwell
first formulated the basic equations for all electromagnetic phenomena, known as
the Maxwell equations. Maxwell expressed his full admiration for Fourier’s great
work on the theory of heat.
Among his other research contributions to physical sciences, Fourier first
suggested that the earth’s atmosphere helps trap the heat of sunlight. This special
behaviour of the atmosphere is known as the greenhouse effect because it resembles
the action of the green roof of a greenhouse. In his series of articles published during
1824 and 1827, he investigated various possible sources of the additional observed
heat, and suggested that interstellar radiation may be responsible for a large amount
of warmth. Apart from this work, Fourier published many papers on terrestrial heat
and radiant heat which led to his mathematical derivation of the sine law for
emission of radiation at the surface of heated bodies based on John Leslie’s
(1766–1832) experimentally determined the sine law. He also studied the application
of the principle of virtual velocities to fluid flows, oscillations of system of bodies
about the state of equilibrium, and the subject of wave motions in elastic lamina.
He also took up a few aspects of Laplace’s work on probability and statistics dealing
with the significance of the normal distribution and the central limit theorem.
In particular, his interest in the density function fðxÞ ¼ p1ffiffiffiffi 2
exp x2 =2 of the
standard normal (or Gaussian) distribution occurs in a wide variety of mathematical
contexts, from probability theory to Fourier analysis to quantum mechanics. This
density function has some unique properties that are not shared by any other special
functions in mathematics and mathematical physics.
 596 L. Debnath

4. Fourier series
Historically, it was Fourier who first not only represented an arbitrary function in
trigonometric series of cosine and sine functions, but also raised the fundamental
question of convergence of the series. Once he made the following bold statements:
‘Thus, there is no function . . . which cannot be expressed by a trigonometric
series . . . [or] definite integral’. ‘Regarding the researches of d’Alembert and Euler
could one not add that if they knew this expansion they made but a very imperfect
use of it. They were both persuaded that an arbitrary and discontinuous function
could never be resolved in series of this kind, and it does not seem that anyone had
developed a constant in cosines of multiple arcs, the first problem which I had to
solve in the theory of heat’.
The first major result about the convergence of a series of functions was Peter
Gustav Lejeune Dirichlet’s (1805–1859) theorem of convergence of Fourier series of
piecewise continuous functions. In 1829, he was one of the first mathematicians who
Downloaded by [Marshall University] at 20:27 16 August 2013

proved the first theorem giving sufficient (and very general) conditions for the
Fourier series of function f(x) to converge pointwise to f(x) after 18 years of
Fourier’s discovery. The proof given by Dirichlet was a kind of refinement of that
sketched by Fourier in the final section of his treatise on the theory of heat.
Dirichlet’s method of proof was to calculate the limit of the nth partial sum of a
Fourier series, sn(x) as n ! 1. He proved that for any given value of x, the sum of
the Fourier series is f(x) if f(x) is continuous at that point x, and is
1
2 ½ fðx  0Þ þ fðx þ 0Þ if f(x) is discontinuous at that point. It was Dirichlet who
closely followed Fourier’s work and then in 1829 established for the first time a
rigorous mathematical theory of Fourier series. In 1835, he also established the
Poisson summation formula for the Fourier analysis, and then modernized the
concept of function that served as the fundamental basis for the nineteenth century
investigations of mathematical analysis. In connection with the study of convergence
of Fourier series, Dirichlet commended Fourier’s great work: ‘The sine and cosine
series, by which one can represent an arbitrary function in a given interval, enjoy
among other remarkable properties that of being convergent. This property did not
escape the great geometer (Fourier) who began, through the introduction of
representation of functions just mentioned, a new career for the applications of
analysis; it was stated in the Memoir which contains his first research on heat. But no
one so far, to my knowledge, gave a general proof of it’. Inspired by Dirichlet’s work
on Fourier series, Bernard Riemann (1826–1866) first formulated the concept of
Riemann integral in his memoir on trigonometric series in 1854 that served the
fundamental basis for his profound work on the convergence of Fourier series.
According to his fundamental theorem, for a bounded integrable function f(x),
the convergence of its Fourier series at a point x in [, ] depends only on the
behaviour of f(x) in an arbitrarily small neighborhood of that point x. It was
Riemann who first observed this behaviour and referred it as the localization
principle for the convergence of Fourier series. Both Dirichlet and Riemann
recognized that more than continuity is required to ensure the convergence of
Fourier series, namely the requirement of bounded variation of the function f(x) in a
neighbourhood of x. In 1854, Riemann praised Fourier’s work and said: ‘Nearly fifty
years had passed without any progress on the question of analytic representation of
an arbitrary function, when an assertion of Fourier threw new light on the subject.
Thus a new era began for the development of this part of Mathematics and this was
 International Journal of Mathematical Education in Science and Technology 597

heralded in a stunning way by major developments in mathematical Physics’.

Further investigations, continuing to the present, led to extensions of this result and
revealed that the general question of convergence of a Fourier series is very difficult
and deep subject in mathematical sciences. However, the problem of finding
necessary and sufficient conditions on f(x) so that its Fourier series converges to f(x)
remained unsolved for a period of century. The nature of the convergence of Fourier
series also received further attention after the introduction of the notion of uniform
convergence by several leading mathematical scientists including George Gabriel
Stokes (1819–1903), Philipp L. Seidel (1821–1896) and Karl Weierstrass (1815–1897).
Subsequently, based on measure theory Henri Lebesgue’s (1875–1941) theory of
integral and convergence in L2-norm were put Fourier analysis on the modern
mathematical map. His discovery of these fundamental ideas and methods
undoubtedly stimulated further development of Fourier series and integrals.
Despite various attempts, it was an open question and unsolved problem for a
Downloaded by [Marshall University] at 20:27 16 August 2013

period of a century whether a Fourier series of a continuous function converges at a

point. In 1915, a Russian mathematician Nikolai Luzin (1883–1950) conjectured that
if f 2 L2[0, 2], then its Fourier series converges almost everywhere. In 1966, Lennart
Carleson (born in 1928) [1] provided an affirmative answer of Luzin’s conjecture with
the deepest theorem which states that the Fourier series of any square integrable
function f(x) in [0, 2] (i.e. f 2 L2[0, 2]) converges almost everywhere to f(x). Shortly,
after Carleson’s great work, Hunt [2] proved that almost everywhere convergence of
Fourier series of functions in Lp[0, 2] for 1 5 p 5 1. One year later, Sjolin [3] gave
further refinement of Hunt’s theorem to include spaces of functions that are smaller
than L1, but larger than Lp for every p 4 1. Subsequently, Hunt and Taibleson [4]
proved the sharpest result on almost everywhere convergence of Fourier series on the
ring of integers of a local field. However, another great mathematician of the
twentieth century, A.N. Kolmogorov’s (1903–1987) 1926 classical counter example
shows that the result fails when p ¼ 1, that is, the Fourier series of L1 functions
diverges everywhere. This resolved an important question of convergence of Fourier
series which became very useful in mathematical analysis and in many other areas of
mathematical sciences.

5. Fourier and pure mathematics

Fourier had early strong interests in pure mathematics. In fact, Bezout’s book on
The´orie ge´ne´rale des equations alge´braiques (1779) stimulated Fourier’s active
research interest in classical algebra. At the age of 21 in 1789, he presented a
paper on algebraic equations at the Académie Royale des Sciences. He gave a new
proof of the famous rule of René Descartes (1596–1650) which describes how to
determine the number of positive and negative roots of algebraic polynomial
equations. His lifelong interest in the theory of equations was evident from his
incomplete book manuscript on determinate equations which was edited by his close
friend and a famous applied mathematician, Claude-Louis Navier (1785–1836) and
then published it posthumously with the title Fourier’s Analyse des e´quations
de´termine´es in 1831. Fourier’s other ideas on algebra eventually led to his student
Charles Sturm’s (1803–1855) basic theorem. The solution of Fourier’s general
equation of heat conduction in two or three dimensions required precise boundary
conditions. Fourier’s first mathematical treatment and then by his students Charles
 598 L. Debnath

Sturm and Joseph Liouville (1809–1882) opened up the whole new field of
eigenvalues and eigenfunctions of ordinary and partial differential equations of
such enormous importance for modern applied mathematics and mathematical
physics. Apparently, Fourier himself realized the extraordinary nature and true
significance of his discovery of the equation of heat conduction and its methods of
solution by Fourier series and Fourier transforms. This work displayed his unique
physical understanding and mathematical ingenuity. Other modern developments in
pure mathematics which can be traced back to Fourier’s analytical theory of heat
included the theory of orthogonal functions, Fourier’s expansion of an arbitrary
function in terms of eigenfunctions, the ideas of different types of convergence,
definite integral as the limit of a sum and theory of infinite determinants.
Thus, Fourier’s most important, and, indeed, revolutionary contribution to pure
mathematics was the representation of functions by trigonometric (Fourier) series,
its numerous mathematical properties including the theory of convergence,
Downloaded by [Marshall University] at 20:27 16 August 2013

term-by-term integration and differentiation, and then, the determination of the

sum of numerical infinite series which is often impossible by other methods of
summation of infinite series. There are examples which showed that the constructed
Fourier series whose values on a given interval may not, in general, be the same as
those of the function.
Albert Einstein (1879–1955) well-described the mutual relationship of pure
mathematics, applied mathematics and physical laws as: ‘pure mathematics enables
us to discover the concepts and laws connecting them, which gives us the key to the
understanding of the phenomena of nature’.
Fourier went even further than this by insisting on the way in which situations
observed in nature stimulate advanced study and research in pure mathematics:
‘The profound study of nature is the most fertile source of mathematical discoveries.
Not only does this study, by offering a definite goal to research, have the advantage
of excluding vague questions and futile calculations, but it is also a sure means of
moulding analysis itself, and discerning those elements in it which it is still essential
to know and which science ought to conserve. These fundamental elements are those
which recur in all natural phenomena’. He was exactly right in making such a
profound statement that some of the most abstract pure mathematics involved in the
theory of Fourier series and Fourier transforms had successfully been used to solve
many applied physical problems including the vibration of elastic strings, and to
develop the analytical theory of flow of heat in solid bodies of different geometric
configurations. During the subsequent development of Fourier series, Joseph
Willard Gibbs (1839–1903) gave a remarkable surprise by his discovery in 1899
what is now known as the Gibbs phenomenon which states that near a jump
discontinuity of a function, its Fourier series overshoots (or undershoots) it by
approximately 9% of the jump.
Admittedly, the idea of convolution of two or more functions arose naturally first
in the study of Fourier series and Fourier transforms. It also played a fundamental
role in Fourier analysis, number theory, probability theory, mathematical statistics,
harmonic analysis and served more generally in mathematical analysis of functions
in other settings. Georg Cantor (1845–1918) revolutionized mathematics in the
second half of the nineteenth century by creating the foundation of the set theory
which was motivated in part by questions of convergence and uniqueness of
trigonometric and Fourier series expansions. In the 1920s, Friedrich Riesz
(1880–1956) established the norm convergence for Fourier series, and almost
 International Journal of Mathematical Education in Science and Technology 599

simultaneously, Ernst Fisher (1875–1959) unequivocally proved the mean-square

convergence theorem for Fourier series. All these developments led to the theory of
singular integral operators at the centre of harmonic analysis. Fourier analysis, in
general and Fourier transforms, in particular, have been very useful for the solution
of linear partial differential equations with initial and boundary conditions of
different kinds.

6. Fourier’s heat equation and Brownian motion

It was shown by Einstein that the Fourier heat equation is closely associated with the
diffusion process of Brownian motion. In 1828, the British botanist Robert Brown
(1773–1858) observed the continuous random movement of small particles of pollen
suspended in water under the microscope. This never-ending very irregular and
Downloaded by [Marshall University] at 20:27 16 August 2013

wiggly motion caused by collisions of molecules is known as Brownian motion that

was also found in colloidal particles in chemistry and in many other physical and
biological phenomena. Brown pointed out that this irregular motion appeared to
obey no known mathematical or physical laws. During the nineteenth century,
several attempts had been made to understand the origin of the Brownian motion in
microscopic living animals and colloidal particles in chemistry. It has been proposed
that the Brownian motion of particles was caused by the electrostatic forces among
them, and the constant changes in direction in motion took place due to the
influences on particles from molecules of the surrounding medium. For sufficiently
small light particles, numerous collisions could have microscopic impact on their
random displacements. From a thermodynamic point of view, the thermal excitation
of molecules is increased due to increase in temperature of the system, and hence, the
physical explanation of Brownian motion is consistent with experimental observa-
tions. Fourier suggested that heat propagates almost like a collection of infinitesimal
Brownian particles, and hence, his diffusion equation is closely related to Brownian
motion.
In one of his three famous 1905 papers, Einstein developed a major physical
theory of the random motion of the individual particles in Brownian motion.
He described the collective motion of Brownian particles of density (x, t) at position
x and time t which satisfies the Fourier diffusion equation, t ¼ xx with the
solution
 
1 x2
ðx, tÞ ¼ pffiffiffiffiffiffiffiffiffiffi exp  , ð6:1Þ
4t 4t
where  is called the diffusion constant which measures how rapidly the distribution
spreads out with time. In order to verify Einstein’s theoretical molecular explanation
of Brownian motion, Jean Perrin (1870–1942) performed experiments and observed
that functions describing the paths of Brownian motion are nowhere differentiable,
even though they are continuous.
Using statistical methods together with heat equation, Einstein showed that if
Brownian motion of particles starts at the origin, then after a fixed time t, its position
is randomly distributed according to the three-dimensional Gaussian distribution
with mean zero and variance t. In other words, the mathematical description of the
Brownian motion of particles in terms of a random process Bt with transitional
 600 L. Debnath

probability densities is given by the diffusion (or Gaussian) kernel Gt(x, m) in n

dimensions in the form
 
1 ðx  nÞ2
Gt ðx, nÞ ¼ exp  : ð6:2Þ
ð4tÞn=2 4t
Since diffusion process results from the Brownian motion of particles, the exact
n
solution u(x, t) of the Fourier diffusion equation pffiffiffi inx R with the initial data
u(x, 0) ¼ u0(x) is equal to the expected value of u0 2 Bt , that is,
h pffiffiffi i
uðx, tÞ ¼ E u0 2 Bxt , ð6:3Þ

where Bxt is the n-dimensional Brownian motion starting at x. It is further noted that
a trajectory of Brownian motion is almost surely non-differentiable. This explains the
Downloaded by [Marshall University] at 20:27 16 August 2013

irregular randomness of the Brownian motion. Thus, result (6.3) describes the
relation between the solution of heat equation and probability theory and
stochastic processes. Subsequently, this relationship led to the celebrated
Feynman–Kac formula
  Zt 
pffiffiffi pffiffiffi
uðx, tÞ ¼ E u0 2 Bxt exp  V 2 Bxs ds , ð6:4Þ
0

which relates the Brownian motion of particles to the solution of the n-dimensional
heat equation

ut ðx, tÞ ¼ rn2 uðx, tÞ  VðxÞuðx, tÞ, ð6:5Þ

with a given initial data, u0(x), where rn2 is the n-dimensional Laplacian.
Subsequently, the idea of stochastic process or stochastic convergence of probability
distributions came into existence when interpreted as modes of convergence of their
Fourier transforms. Norbert Wiener (1894–1964) first developed the mathematical
theory of probability measure and integration in function space to construct rigorous
mathematical models for Brownian motion which was till then understood only
heuristically by Einstein and others. He also constructed a probability measure in the
space of all continuous functions on C[0, 1), universally known as the Wiener
measure. In fact, he discovered the systematic mathematical description of Brownian
motion known as the Wiener process (a continuous-time stochastic process) in 1923
based on the theory of Fourier series.
Subsequently, several great mathematical scientists including Kolmogorov, Paul
Levy (1886–1971) and Kiyoshi Itô (1915–2008) made some major contributions to
stochastic analysis for a further understanding of Brownian motion and diffusive
processes. Based on Fourier transforms, Levy formulated a new and simpler
approach to the general theory of stable probability distributions. From a
thermodynamic point of view, Brownian motion is irreversible in the sense that,
for a large number of diffusing particles, there exists a smoothing of the
probability distributions describing the state of particles as time increases. On
the other hand, Itô discovered a powerful method of stochastic calculus and
stochastic differential equations for an explicit description of the Markov diffusion
process.
 International Journal of Mathematical Education in Science and Technology 601

7. Some applications of Fourier analysis to diverse areas

We next give some general flavour of the broader impact of Fourier analysis on some
selected areas of mathematics, mathematical physics and mathematical economics.
Using Parseval’s formula for Fourier series, Adolf Hurwitz (1859–1919) gave a very
elegant solution of the famous isoperimetric problem in geometry. He proved that
the area A of the region enclosed by a simple closed curve  in R2 of length ‘ satisfies
the so-called isoperimetric inequality

‘2
A , ð7:1Þ
4
where equality holds if  is a circle. Hurwitz made some comments praising the use
of Fourier series in geometry: ‘Fourier series and analogous expansions intervene
very naturally in the general theory of curves and surfaces. In effect, this theory,
conceived from the point of view of analysis, deals obviously with the study of
Downloaded by [Marshall University] at 20:27 16 August 2013

arbitrary functions. I was thus led to use Fourier series in several questions of
geometry, and I have obtained in this direction a number of results which will be
presented in this work. One notes that my considerations form only a beginning of a
principal series of researches, which would without doubt give many new results’.
Several great mathematicians including Riemann, Weierstrass and G.H. Hardy
(1877–1947) investigated the famous problem of everywhere continuous but nowhere
differentiable function in real analysis. In 1861, Riemann suggested that the function
X1
1  
RðxÞ ¼ 2
sin n2 x , x 2 R, ð7:2Þ
n¼1
n

is everywhere continuous, but nowhere differentiable in one of his lectures without

proof. In order to give a formal proof of the Riemann function, Weierstrass
discovered another everywhere continuous but nowhere differentiable function
defined by
X
1
WðxÞ ¼ bn cosðan xÞ, ð7:3Þ
n¼1
 
where 0 5 b 5 1 and a is an integer greater than one. He proved that if ab 4 1 þ 32 ,
the Weierstrass function, W(x) is nowhere differentiable. In 1916, Hardy proved that
R(x) is not differentiable at all irrational multiples of  and at some rational
multiples of . In 1969, Gerver proved that
 the Riemann function is differentiable at
all rational multiples of  of the form p
q , where p and q are odd integers and is not
differentiable at other points. In the context of Fourier series, there is another
function
X
1 X
1
fðxÞ ¼ 2n expði 2n xÞ ¼ 2n ½cosð2n xÞ þ i sinð2n xÞ, ð7:4Þ
n¼0 n¼0

where 0 5  5 1. This is another continuous but nowhere differentiable function

with many vanishing Fourier coefficients. A Fourier series with many missing
Fourier coefficients, like (7.3) and (7.4) is called lacunary Fourier series. It can be
shown that the real as well as the imaginary part of f(x) are both nowhere
differentiable.
 602 L. Debnath

With ideas from Fourier series, Hermann Weyl (1885–1955) proved his famous
criterion which states that a sequence of real numbers fxn g1 1 in [0, 1] is
equidistributed for all integers k 6¼ 0 if and only if

1 XN
expð2ikxn Þ ! 0 as N ! 1: ð7:5Þ
N n¼1

This means that, to understand the equidistribution properties of a real sequence

{xn}, it is sufficient to estimate the size of the corresponding Fourier sum
P N
n¼1 expð2ikxn Þ. For example, it can be shown using the Weyl criterion that the
sequence of the fractional parts hn2 i ¼ n2  [n2 ], where [n2 ] is the greatest
integer  (n2 ) and is the integral part of n2 and is irrational, is equidistributed
1
in [0,1]. However, it can be proved  that  sequence fxn g1 , where xn is the
pffiffiffithe
n 1
fractional part of g where g ¼ 2 1 þ 5 is the golden number [5], is not
Downloaded by [Marshall University] at 20:27 16 August 2013

equidistributed in [0, 1].

As a remarkable application of the theory of finite Fourier series, we state
Dirichlet’s theorem on primes in arithmetic progression. If a and h are positive
integers with no common factor, then the arithmetic progression
a, a þ h, a þ 2h, ..., a þ nh,  ð7:6Þ
contains infinitely many prime numbers. In this theorem, it is the theory of Fourier
series on finite Abelian group Z*(a) that plays a fundamental role in the solution of
the problem, where Z*(a) is defined as the set of all integers modulo a that have
multiplicative inverses with multiplication modulo a as the group operation.
In 1924, Werner Heisenberg (1901–1976) first formulated his celebrated
uncertainty principle between the position and momentum of a moving particle in
quantum mechanics which states that it is not possible to determine the position and
momentum of a particle exactly and simultaneously. This principle has an important
physical interpretation as an uncertainty of both position and momentum of a
particle described by a wave function 2 L2(R). Mathematically, this principle can
be formulated in terms of a relation between a function and its Fourier transform. If
f 2 L2(R), then f and F fðxÞ ¼ FðkÞ ¼ b fðkÞ cannot both be essentially localized at
x ¼ 0, k ¼ 0. In other words, it is not possible that the widths of the graphs of j f(x)j2
and jF(k)j2 can both be made arbitrarily small. This fact underlines the Heisenberg
uncertainty principle in quantum mechanics and the bandwidth theorem in signal
analysis. All these can be explained more explicitly as follows:
The function fX(x) ¼ j (x)j2 can be interpreted as the probability density for the
random position variable X with its probabilistic interpretation
Z1 Z1
2  2
¼  
ðxÞ dx ¼ fX ðxÞdx ¼ 1: ð7:7Þ
1 1
 2  2
Similarly, fP ðkÞ ¼ F ðxÞ  ¼ bðkÞ is the corresponding probability density
function for its momentum P. The Heisenberg uncertainty principle states in the
form of an exact inequality that these two probability densities cannot be essentially
localized.
We then introduce the expectation of the random variable X2 by
Z1 Z1
   2 2
E X2 ¼ x2 fX ðxÞdx ¼ x2  ðxÞ dx ¼ x : ð7:8Þ
1 1
 International Journal of Mathematical Education in Science and Technology 603

This can be regarded as a measure of the horizontal spread of the wave function
(x). Similarly, the spread of its Fourier transform, bðkÞ over the k-axis can be
introduced by the integral
Z1 Z1  2 2
 
k2 fP ðkÞdk ¼ k2  bðkÞ dk ¼ k b : ð7:9Þ
1 1

In terms of these two quantities, the uncertainty principle can be formulated in the
form of a precise inequality, known as the Heisenberg inequality,
1 2
x  kb  , ð7:10Þ
2
where the equality holds when (x) is the Gaussian function, that is,
(x) ¼ C exp(ax2), a 4 0, and C is a constant. For a proof, see [6].
It is important to point out new and modern impact of Fourier analysis on
Downloaded by [Marshall University] at 20:27 16 August 2013

mathematical economics and finance. Based on the principle of Brownian motion for
random fluctuations of share prices, Black and Scholes [7] first formulated a dynamic
replication model of price and show that the option price function, C(s, t) satisfies a
partial differential equation with an extra drift term in the form
1 2 2
Ct þ rs Cs þ s Css ¼ rC, 0 5 t 5 T, ð7:11Þ
2
with the final boundary condition, C(s, T ) ¼ f(s), where r and 2 are constants. This
equation is now known as the Black–Scholes (or BS) equation. With a suitable
change of variables, the BS equation can be reduced to the one-dimensional Fourier
diffusion equation. Thus, the Fourier transform of C(s, t) with respect to s can be
used to find solution of the BS model equation. Although the BS dynamic replication
approach satisfies the principle of market completeness, one problem of this approach
is that the price of a derivative does not depend on the drift of the share price. So,
there is a need for an alternative approach to derivatives pricing theory, known as
the risk-neutral expectation pricing theory which has also subsequently been
developed. Thus, there are two major methods for pricing: the BS replication
approach and the risk-neutral expectation approach. Surprisingly, the Feynman–
Kac theory describing the relation between the solution of the Fourier diffusion
equation and probability theory helped combine the above two approaches together
so that certain second-order linear partial differential equation can be solved using
expectations of diffusive processes. However, there is considerable evidence that log
of the share price does not follow the principle of Brownian motion with drift. In
particular, when stock market crashes that corresponds to a big jump in the share
price leading to a failure of the BS replication strategics. So, the BS model fails to
capture some important features of share-price evolution because all paths in
Brownian motion are continuous but non-differentiable. This leads to many
generalization of the BS model, but most generalizations do not retain the principle
of market completeness which means that they give rise to multiple option prices
rather than just one price. The analysis of real market data on small-scale share
movements reveals that they do not resemble a diffusion process. In fact, they behave
more or less like a series of jumps than a Brownian motion. So, rescaling time based
on the number of trades that have occurred rather than on calendar time leads to
returns that do become approximately normal distribution. Thus, the BS model can
 604 L. Debnath

be generalized based on trading time. An example of such a model is known as

the variance gamma model. In general, the theory of Levy processes can be
employed to develop more general theories of price movements for shares and
other assets.
Since Fourier’s 1822 celebrated work on the theory of heat, the sine and cosine
functions have played the fundamental basis functions in the subsequent develop-
ment of Fourier series and Fourier transforms. However, in many applications,
especially in the space-wavenumber (or time-frequency) analysis of a function (or
signal), the standard Fourier analysis is not adequate because the Fourier transform
of the function (or signal) does not contain any local information. In fact, it excludes
the idea of wavenumbers changing with space (or frequencies changing with time), or
equivalently, the notion of finding the wavenumber spectrum of a function with
space (or frequency spectrum of a signal with time). So, the Fourier transform
analysis of functions cannot be used for the analysis of functions (or signals) in a
Downloaded by [Marshall University] at 20:27 16 August 2013

joint space wavenumber (or joint time-frequency) domain. So, in 1980s, a whole new
idea of wavelets has been introduced as a family of functions constructed from
translations and dilations of a single function called the ‘mother wavelet’ (x).
They are defined by
 
1 xb
a,b ðxÞ ¼ pffiffiffiffiffiffi , a, b 2 R, a 6¼ 0, ð7:12Þ
jaj a
where a is called a scaling parameter which measures the degree of compression or
scale, and b is a translation parameter which determines the space location of a
wavelet. If 2 L2(R) and a,b(x) is given by (7.12), the continuous wavelet transform
is defined by
Z1
 
W ð f Þ ða, bÞ ¼ a,b ðxÞ fðxÞdx, ð7:13Þ
1

where a,b(x) plays the same role as the kernel exp(ikx) in the Fourier transform.
Like the Fourier transform, the continuous wavelet transform is linear, and it is not a
single transform, but any transform obtained in this way. The inverse wavelet
transform can be defined so that f(x) can be reconstructed by a double integral
formula [8]. It is now well-known that, analogous to the Fourier analysis, the
wavelet analysis has provided a famous new method for decomposing a function or a
signal in 1980s. It is more useful and better suited to a wide variety of problems in
applied and computational mathematics and also to many questions in real and
harmonic analysis. The most significant contributions of both discrete and
continuous wavelets can be found in signal processing and digital image
compression.

8. Historical development of Fourier series

Historically, in 1747, D’Alembert first formulated the wave equation with the first
solution of a partial differential equation. In fact, Fourier series originated from the
early study of the initial-boundary problem of the one-dimensional wave equation
for vibration of an elastic string fixed at its ends in the eighteenth century and of the
Fourier heat equation of a solid body in 1800s. Euler, D’Alembert and Daniel
 International Journal of Mathematical Education in Science and Technology 605

Bernoulli made earliest attempts to solve the initial-boundary value problem of the
classical one-dimensional wave equation [10]

utt ¼ c2 uxx , 0 5 x 5 ‘, t 4 0, ð8:1Þ

where u ¼ u(x, t) is the displacement of the string of length ‘ at the position x and
time t, c2 ¼ (T/), T is the tension and  is the mass density of the string.
The initial and boundary conditions of this problem are given by

uðx, 0Þ ¼ fðxÞ, and ut ðx, 0Þ ¼ gðxÞ, 0 5 x 5 ‘, ð8:2abÞ

uð0, tÞ ¼ 0 ¼ uð‘, tÞ, t 4 0, ð8:3abÞ

where f(x) and g(x) are arbitrary functions of x.
Both Euler and D’Alembert made the first attempt to solve (8.1)–(8.3ab) in the
Downloaded by [Marshall University] at 20:27 16 August 2013

travelling wave solution in the functional form

uðx, tÞ ¼ ðx  ctÞ þ ðx þ ctÞ, ð8:4Þ
where (x  ct) represents a wave travelling to the right with constant velocity c and
(x þ ct) represents a wave moving to the left with constant speed c. The solution
(8.4) is the general solution of (8.1) representing the superposition of two travelling
wave solutions.
On the other hand, Daniel Bernoulli derived the solution of (8.1) with (8.3ab)
satisfied in an infinite trigonometric series form
1 h
X nct nct i nx
uðx, tÞ ¼ an cos þ bn sin sin , ð8:5Þ
n¼1
‘ ‘ ‘

where an and bn are to be determined from (8.2ab) so that

X
1
nx
fðxÞ ¼ uðx, 0Þ ¼ an sin , ð8:6Þ
n¼1
‘

c X
1
nx
gðxÞ ¼ ut ðx, 0Þ ¼ nbn sin : ð8:7Þ
‘ n¼1 ‘

This means that arbitrary functions f(x) and g(x) have been expanded in
trigonometric (or Fourier) series. Since (8.6) or (8.7) is an infinite series, the
question of convergence arose. But many mathematicians including Euler,
D’Alembert and Laplace were not entirely convinced about the existence of
Bernoulli’s Fourier series solution (8.5). This led to a serious controversy as the
solution could hold if an arbitrary function could be represented in Fourier series.
In 1759, a young mathematical scientist, Lagrange joined in this controversial debate
on the various forms of the solution of the vibrating string problem and published a
paper which was criticized by both Euler and D’Alembert. In order to resolve this
controversy, Lagrange gave an entirely new treatment of the problem. From 1768,
D’Alembert wrote a series of notes in which he attacked Euler’s views on the
solution of the problem. In 1779, Laplace entered into this debate and supported
D’Alembert views. They all discovered solutions in several different forms and the
 606 L. Debnath

merit of their solutions and their relations among these solutions extending over
more than 25 years. Their major concerns were whether an arbitrary function can
be expanded in Fourier series and convergence of such series. Unfortunately,
the outcome of this controversial issues remained inconclusive for a long period
of time.
In 1807, it was Joseph Fourier who resolved this long-standing controversy by
discovering the solution of the one-dimensional heat equation in terms of Fourier
series. Indeed, he formulated and solved the initial-boundary value of the
one-dimensional heat equation in a rod of length ‘
ut ¼ uxx , 0 5 x 5 ‘, t40 ð8:8Þ
with the initial and boundary conditions
uðx, 0Þ ¼ fðxÞ, 0 5 x 5 ‘, ð8:9Þ
Downloaded by [Marshall University] at 20:27 16 August 2013

uð0, tÞ ¼ 0 ¼ uð‘, tÞ, t 4 0: ð8:10Þ

Using the method of separation of variables, Fourier obtained the most general and
rapidly convergent solution for the temperature distribution u(x, t) in the form
X1  
n 2 nx
uðx, tÞ ¼ bn exp  t sin , ð8:11Þ
n¼1
‘ ‘

where the initial condition (8.9) implies that

X
1
nx
fðxÞ ¼ bn sin : ð8:12Þ
n¼1
‘

The upshot of Fourier’s analysis led to expanding an arbitrary function in an infinite

series of sine functions and then finding the infinite number of coefficients
Z ‘
2 nx
bn ¼ fðxÞ sin dx: ð8:13Þ
‘ 0 ‘

In his study of heat conduction problem, Fourier boldly undertook to demonstrate

the validity of the Fourier series expansion and to obtain a formula for its
coefficients. Although his argument was not rigorous, it must be considered a major
conceptual breakthrough. Thus, Fourier’s conviction and work eventually led
Dirichlet, Riemann and others to a complete proof that an arbitrary function could
be represented by a convergent Fourier series.
Fourier also considered trigonometric series expansion of a periodic, integrable
function f(x) of period 2‘ defined in (‘, ‘) of the form [11]
1 h
1 X nx nx i
fðxÞ ¼ a0 þ an cos þ bn sin , ð8:14Þ
2 n¼1
‘ ‘

where
Z ‘
1 nx
an ¼ fðxÞ cos dx, n ¼ 0, 1, 2, . . . , ð8:15Þ
‘ ‘ ‘
 International Journal of Mathematical Education in Science and Technology 607

and
Z ‘
1 nx
bn ¼ fðxÞ sin dx, n ¼ 1, 2, 3, . . . : ð8:16Þ
‘ ‘ ‘

Fourier also presented series of the form

1 X1
nx
fðxÞ ¼ a0 þ an cos : ð8:17Þ
2 n¼1
‘

The series (8.12), (8.14) and (8.17) are called the Fourier sine series, Fourier series and
Fourier cosine series, respectively.
It is sometimes convenient to represent a function f(x) by a Fourier series in
complex form. This representation can be easily derived from the Fourier series
Downloaded by [Marshall University] at 20:27 16 August 2013

(8.14) of a periodic integrable function of period 2 defined in (, ) with an and bn

given by (8.15)–(8.16) in the form

a0 X1
f ðxÞ ¼ þ ðan cos nx þ bn sin nxÞ
2 n¼1
1  
a0 X 1  inx inx
 bn  inx inx

¼ þ an e þ e þ e e
2 n¼1 2 2i
1  
a0 X 1 inx 1 inx
¼ þ ðan  i bn Þe þ ðan þ i bn Þe ð8:18Þ
2 n¼1 2 2

X
1  
¼ c0 þ cn einx þ cn einx
n¼1
X
1 X
1
¼ cn einx ¼ b
fðnÞeinx , ð8:19Þ
n¼1 n¼1

where
Z 
a0 1
c0 ¼ ¼ f ðxÞdx, ð8:20Þ
2 2 

Z
1 1 
cn ¼ ðan  i bn Þ ¼ f ðxÞðcos nx  i sin nxÞdx
2 2 
Z
1
¼ f ðxÞeinx dx ¼ b
fðnÞ, ð8:21Þ
2 

Z
1 1 
cn ¼ ðan þ i bn Þ ¼ f ðxÞðcos nx þ i sin nxÞdx
2 2 
Z
1
¼ f ðxÞ einx dx ¼ cn : ð8:22Þ
2 
 608 L. Debnath

Thus, we obtain the Fourier series of f(x) in complex form (8.19), where cn is given
1
by (8.21). Multiplying (8.19) by 2 f ðxÞ and integrating the result from  to  gives
the celebrated Parseval formula for a complex Fourier series:
Z X1 Z
1  2 1 
f ðxÞ dx ¼ cn f ðxÞeinx dx
2  n¼1
2 
X1
¼ cn  cn
n¼1
X1 X
1
¼ cn cn ¼ jcn j2 : ð8:23Þ
n¼1 n¼1

If the Fourier series (8.18) converges to f(x), then

Z
a20 X1   1  2
Downloaded by [Marshall University] at 20:27 16 August 2013

2 2
þ a þ bn ¼ f ðxÞdx: ð8:24Þ
2 n¼1 n  

This is universally known as Parseval’s relation and is one of the fundamental results
in the theory of Fourier series.
We next derive the Fourier integral formula (3.1) from the complex Fourier series
of a function f(x) of period 2‘ defined in (‘, ‘) in the form [9].
X1  
inx
fðxÞ ¼ cn exp , ð8:25Þ
n¼1
‘

where the Fourier coefficients, cn are given by

Z  
1 ‘ in
cn ¼ fðÞ exp  d: ð8:26Þ
2‘ ‘ ‘
We write n 
‘ ¼ kn and kn ¼ ðknþ1  kn Þ ¼ ‘ in (8.25)–(8.26). In the limit as ‘ ! 1,
the interval (‘, ‘) grows indefinitely, the difference kn ! 0 and the values kn get
closer and closer and tends to a continuous real variable k and ‘ ¼ kn ! dk.
Substituting the values of cn from (8.26) into (8.25) gives
Z ‘ 
1 X 1

fðxÞ ¼ fðÞeikn d expðixkn Þ: ð8:27Þ
2 n¼1 ‘ ‘

In the limit as ‘ ! 1, kn, kn can be replaced by k and dk, respectively, and the
infinite sum in (8.27) can be replaced by an integral so that (8.27) reduces to the
famous Fourier integral formula (3.1).

9. Fourier transforms
The Fourier integral representation (3.1) of f(x) can be used to give the direct
definition of the Fourier transform of f(x), denoted by F { f(x)} ¼ F(k), k 2 R ¼
(1, 1), as follows:
Z1
1
F fðxÞ ¼ FðkÞ ¼ pffiffiffiffiffiffi eikx fðxÞdx: ð9:1Þ
2 1
 International Journal of Mathematical Education in Science and Technology 609

This simply means that the Fourier transform of f (x) on R is a continuous version of
the Fourier coefficients, cn defined by (8.26) as ‘ ! 1. Further, the definition (9.1)
shows a symmetry between a function and its Fourier transform as the Fourier
transform of a function on R is again a function on R. Thus, F(k) is a continuous
function of wavenumber k ¼ 2  ,  is the wavelength, and is the Fourier transform of
f(x), where x 2 R is the spatial coordinate.
The Fourier integral formula (3.1) allows us to also define the inverse Fourier
transform, F 1{F(k)} ¼ f(x), in the form
Z1
1
F 1 FðkÞ ¼ fðxÞ ¼ pffiffiffiffiffiffi eikx FðkÞdk: ð9:2Þ
2 1
Clearly, (9.2) shows that f(x) is decomposed into waves  of different
 wavenumbers, k
(or different wavelengths  ¼ 2 ) and amplitude p1ffiffiffiffi FðkÞ. Physically, the inverse
k 2
Fourier transform provides a way of reconstructing functions on the whole real line
Downloaded by [Marshall University] at 20:27 16 August 2013

R ¼ (1, 1) as integral superposition of an infinite number of sinusoidal

oscillations in much the same way that Fourier series is used to reconstruct
functions on a finite interval (‘, ‘) or (, ) from its Fourier coefficients cn ¼ b fðnÞ
given by (8.21). It is noted that F and F 1 are both linear integral transformations.
One final remark on the definitions of the Fourier transform and the inverse
Fourier transform is that they are obtained from the limit (‘ ! 1) of the Fourier
coefficient (8.26) and the Fourier series (8.25) over a finite interval (‘, ‘),
respectively.
Making reference to Debnath and Bhatta [6], we now state the Poisson summation
formula in Fourier analysis in the form
pffiffiffiffiffiffi 1  
X1
2 X n inx
fðx þ 2naÞ ¼ F exp , ð9:3Þ
n¼1
2a n¼1 a a

where the series on the left-hand side converges absolutely for almost all x in (  a, a).
In particular, when 2a ¼ 1 and x ¼ 0, (9.3) reduces to an elegant form
X
1 X
1 pffiffiffiffiffiffi X
1
fðnÞ ¼ 2 Fð2nÞ ¼ b
fð2nÞ, ð9:4Þ
n¼1 n¼1 n¼1

where bfðkÞ is the Fourier transform of f(x) without the factor p1ffiffiffiffi
2
. The formula (9.4)
represents the sum of the values of any function at all integers is equal to the sum of
the values of its Fourier transform at integral multiples of 2. It is a very useful of
transforming an infinite series from one form to another form which may be more
useful for computation of a series in closed form.

10. Time-dependent Fourier coefficients and simple harmonic motion

We conclude this article by showing that time-dependent Fourier coefficients cn(t) in
the Fourier expansion of a real-valued scalar field (, t) satisfy the equation of the
simple harmonic motion. We write the Fourier expansion of (, t) as
X
1
ð, tÞ ¼ cn ðtÞ expðinÞ, 0    2, t 4 0, ð10:1Þ
n¼1
 610 L. Debnath

where c0, cn and cn are the Fourier coefficients given by (8.21)–(8.22) with
cn ¼ cn ¼ cn , because (, t) is real-valued. It is then useful to use William Rowan
Hamilton’s (1805–1865) principle of least action with the action integral A which is
the time integral of the Lagrangian L ¼ T  V, the difference between the kinetic
energy T and the potential energy V of a dynamical system. More precisely, the
action integral is defined by
Z t2 Z t2
A¼ L dt ¼ ðT  V Þdt: ð10:2Þ
t1 t1

The Hamilton principle of least action asserts that

Z t2
A¼ L dt ¼ 0: ð10:3Þ
t1
Downloaded by [Marshall University] at 20:27 16 August 2013

This means that the actual motion of the system is such as to make the value of the
action A either a minimum or a maximum. For example, if x(t) is an arbitrary path
of a particle of mass m, the kinetic energy T ¼ 12 mx_ 2 and the potential energy
V ¼ 12 kx2 ðk 4 0Þ so that
Z t2 Z t2  
1 2
A¼ L dt ¼ mx_  V dt ¼ 0: ð10:4Þ
t1 t1 2
This leads to the Euler–Lagrange equation for L given by
 
@L d @L
 ¼ 0: ð10:5Þ
@x dt @x_
Or,
mx€ þ rV ¼ 0 ¼ mx€ þ kx ¼ 0: ð10:6Þ
This is the celebrated Newton law of motion representing the simple harmonic
oscillator with pforce
ffiffiffiffiffiffiffiffiffi F ¼ rV. Thus, the solution of (10.6) is oscillatory with
frequency ! ¼ k=m and periodic time 2 ! .
Finally, the simplest action integral A for the scalar field (, t) given by (10.1)
can be written as
Z t2 Z 2  
1  _2 
A¼  02 dt d, ð10:7Þ
t1 0 2
where _ ¼ @@t and 0 ¼ @@ . Substituting the values of _ and 0
from (10.1) in (10.7)
and then evaluating the -integral gives
Z t2 " X
1
#
2 2 2
A¼ c_ ðtÞ  n cn ðtÞ dt: ð10:8Þ
t1 n¼0

Thus, the application of the Hamilton principle of least action, A ¼ 0 leads to the
equation
X
1
c€n ðtÞ þ n2 cn ðtÞ ¼ 0: ð10:9Þ
n¼0
 International Journal of Mathematical Education in Science and Technology 611

This represents an infinite number of equations of simple harmonic oscillators for

cn(t) except for c0 which corresponds to a free particle with zero potential.

11. Concluding remarks

It is clear from the above discussion that Joseph Fourier was undoubtedly one of the
most brilliant and influential mathematical scientists of the eighteenth century and
beyond. He revolutionized mathematics and mathematical physics through his major
discovery of the theory of heat, Fourier series and Fourier transforms. There is no
doubt at all about Fourier’s profound and everlasting influence on mathematical
sciences and the scientific community of the world. His lifelong interest, thoughtful
and eloquent views on mathematical theory, experiment, observation and philosophy
of science can be described by his own memorable quotations. In reference to
Downloaded by [Marshall University] at 20:27 16 August 2013

mathematics, Fourier made extraordinary comments on its universality, simplicity,

clarity, elegance and beauty. Based on his belief of these principal attributes of
mathematics, Fourier also made the following statements: ‘There cannot’, ‘he
believes’, be a more universal or simple language, one more exempt from errors and
obscurities’. His notable views on the role of the mathematical theory and its
comparison with experimental observations can also be described by his delightful
words: ‘the theory itself will direct all these measures and will assign to them their
precision. There cannot be henceforth any considerable progress which will not be
founded on these experiments; for mathematical analysis is able to deduce from
general and simple phenomena expressions of laws of nature: but the special
application of these laws to effects which are very complicated requires a long theory
of exact observations’. ‘ . . . mathematical analysis is as extensive as nature itself; it
defines all perceptible relations, measures time, spaces, forces, temperatures; this
difficult science is formed slowly, but it preserves every principle which it has once
acquired; it grows and strengthens itself incessantly in the midst of many variations
and errors of the human mind’. ‘Mathematics compares the most phenomena and
discovers the secret analogies that unite them’.
In Fourier’s own words: ‘primordial facts whose causes are not considered by
mathematicians, but which they admit as resulting from ordinary observation and
confirmed by experiments’. With regard to philosophy of science based solely on
observable scientific facts and their relations to each other, he was an exponent of
positivism and expressed his views as follows: ‘The primordial causes are unknown
to us, but they are simple and constant laws which can be discovered by observation
and whose study is the object of natural philosophy’.
Fourier will be remembered forever not only for his splendid scientific
achievements, but also for his great contributions to the scientific community of
the world, in general and his unique political, social and administrative public service
contributions to the welfare of France and Egypt, in particular. Indeed, he was a
great statesman and a successful diplomat, and his diplomatic skills enabled him to
survive the many changes in the socio-political landscape of France in his time.
By any standards, he would belong to a select group of great scientists including
Galileo (1564–1642), Newton, Euler, Max Planck (1858–1912), Maxwell and
Einstein who produced revolutions in various branches of mathematical and
physical sciences.
 612 L. Debnath

Acknowledgement
The author expresses his sincere thanks to the referees for suggesting some changes for the
improvement of the exposition of this article.

References

[1] L. Carleson, On convergence and growth of partial sums of Fourier series, Acta. Math. 116
(1966), pp. 135–137.
[2] Hunt, R., On the convergence of Fourier series, Orthogonal Expansions and their
Continuous Analogues (Proc. Conf. Edwardsville, IL, 1967), Southern University Press,
Carbondale, IL, 1968, pp. 235–255.
[3] P. Sjölin, An inequality of Paley and convergence a.e. of Walsh-Fourier series, Ark.
Math. 7 (1969), pp. 551–570.
Downloaded by [Marshall University] at 20:27 16 August 2013

[4] R. Hunt and M. Taibleson, Almost everywhere convergence of Fourier series on the ring of
integers of a local field, SIAM J. Math. Anal. 2 (1971), pp. 607–625.
[5] L. Debnath, A short history of the Fibonacci and golden numbers with their applications,
Int. J. Math. Educ. Sci. Technos. 41 (2010), pp. 1–31.
[6] L. Debnath and D. Bhatta, Integral Transforms and Their Applications, 2nd ed., Chapman
& Hall/CRC, Boca Raton, FL, 2007.
[7] F. Black and M. Scholes, The valuation of options and corporate liabilities, J. Political
Econ. 81 (1973), pp. 637–654.
[8] L. Debnath, Brief introduction to history of wavelets, Int. J. Math. Educ. Sci. Technol. 29
(1998), pp. 677–688.
[9] L. Debnath, The legacy of Leonhard Euler – A tricentennial tribute, Int. J. Math. Educ.
Sci. Technol. 40:3 (2009), pp. 353–388.
[10] L. Debnath, The legacy of Leonhard Euler – A Tricentennial Tribute, Imperial College
Press, London, 2010.
[11] T. Myint-U and L. Debnath, Linear Partial Differential Equations for Scientists and
Engineers, 4th ed., Birkhauser Verlag, Boston, MA, 2007.