Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Heaviside and the Operator Calculus

Tianxiang Xiong
December 5, 2011
1 Biography
From time to time there live among us men of genius whose creative work is
inadequately recognized by their contemporaries but who, later, are found to have
rendered a service of the highest value to the advancement of Sciencesuch a man
was Oliver Heaviside.
Foreword, J. Inst. Elect. Eng., April 1950
Oliver Heaviside was born on May 18, 1850 as the youngest of four sons of Thomas Heaviside,
a wood-engraver, and Rachel West, a governess. Although his immediate family had no
connection to science and engineering, his uncle-in-law on his mothers side was Sir Charles
Wheatstone, famous telegraphist and inventor.
At a young age, Heaviside exhibited both the intellectual capability and individualist streak
that would come to characterize him in later life. He attended Camden House School in
London and did well in his examinations, taking a top prize in the natural sciences but
receiving a bad mark in geometry, supposedly for disagreeing with the teaching method,
with its focus on rigor. This does not mean that Heaviside did not think that geometry was
an important subject; on the contrary, he recognized its importance, and was upset mainly
at the pedagogical methods he experienced as a schoolboy [9, 230].
Euclid is the worst. It is shocking that young people should be addling their
brains over mere logical subtleties...when they might be learning geometry, a
most important fundamental subject. I hold the view that it is essentially an
experimental science, like any other, and should be taught observationally, de-
scriptively, and experimentally...
He left the school at 16 years of age with no evidenced desire of seeking further education, but
his uncle took a keen interest in him and exhorted him to continue studying, in particular,
the natural sciences.
It was no doubt due to his uncles inuence that he found work with a Danish telegraph
company soon after leaving school. He was involved with the laying of the rst Anglo-Danish
telegraph line, and observed for the rst time a number of strange phenomena that no one
could explain and to which he would dedicated much of his later life studying. In 1870, after
the Great Northern Company acquired the Danish company at which he was employed, he
was promoted and returned to England. Again, this was probably due in no small part to
nepotism, for his brother Arthur was then an engineer with the Post Oce in Newcastle.
His early work at the Great Northern Company was highly practical. In particular, he was
interested in the problem of self-inductance in line transmission, which at the time took into
account only resistance and capacitance. In 1873, he obtained a rst edition of Maxwells
1
Electricity and Magnetism and became enthralled with Maxwells ideas. From then on, he
dedicated himself to studying and furthering Maxwells works.
In 1874 Heaviside decided that his day job was interfering too much with his theoretical
work and, at the age of 24, quit the Great Northern Company and returned to living with
his parents. From 1874 on he worked in seclusion, publishing voluminously and establishing
a name for himself as an expert on electromagnetism [3]. Aside from numerous articles,
Heaviside is known for two major works, Electrical Papers and Electromagnetic Theory.
Then as now, his mathematical work has been overshadowed by his work in engineering and
physics, perhaps because his mathematical developments always played a supporting role.
Nonetheless, Heaviside was no mean mathematician. His study of telegraphs, telephones, and
electrical circuits required sophisticated mathematical theory, which led to his formulation
of vector analysis and operator calculus. Using vectors, he recast Maxwells 20 quaternion
equations in 20 variables to the modern form of 4 vector equations.
We will not talk much about Heavisides work on vector analysis, which was revolutionary
but quickly accepted. Instead, our focus will be on his operator calculus, which was highly
controversial and drew him into bitter conicts with both mathematicians and engineers.
We will examine at length the nature of operational calculus and the criticism it received,
as well as the ugly personal and political sides of scientic research.
2 Historical Background of Operator Calculus
Operator calculus is a technique by which problems in analysis are transformed into algebraic
problems. Sometimes called operational mathematics [10], operator calculus has a long
history dating back to the earliest developments of calculus. G. Leibnizs dierential notation
was meant to represent an algebraic quantity and was treated as such by many ensuing
mathematicians, including P. Laplace (1749 - 1827), J. Lagrange (1736 - 1813) and A. Cauchy
(1789 - 1857).
F. Servois (1768 - 1847) believed that algebraic treatment was applicable to dierential
operators because they obeyed the commutative and distributive laws [8]. This idea was
extended by D. Gregory and G. Boole (1815 - 1864).
Much of this mathematics was obscure in Heavisides time. Therefore, the creation of oper-
ator calculus is often erroneously accredited to him. His contribution was in demonstrating
that the methods were practical and useful, and thereby led to their wide adoption.
Interestingly, the techniques which would supplant Heavisides methods, the Laplace and
Fourier transforms, were developed much earlier, albeit not in modern form.
2
3 Heavisides Operator Calculus
Heaviside used the operator p to represent dierentiation, p =
d
dt
. Then 1/p must represent
integration to satisfy p(1/p)f(t) = f(t). In particular, 1/p is taken to be denite integration,
dened as (1/p)f(t) =
_
t
0
f()d. These operators are manipulated like algebraic symbols.
For example, by repeated integration, we nd that (1/p
n
)U(t) = t
n
/n! (see Appendix A).
Let us look at a simple example (adapted from [1]) in which Heavisides operators can be
applied. Consider the circuit shown in Figure 3.1, which connects a resistor and inductor in
series with a DC voltage source with voltage E. At time t < 0, the switch is open. At t = 0,
the switch is closed. Therefore, a voltage of EU(t) is applied to the circuit. Using familiar
Figure 3.1: Transient Circuit
laws of circuit components
1
we describe the circuit using the dierential equation and its
initial condition
L
di
dt
+ Ri = EU(t)
i(0) = 0
In Heavisides operator notation, this is
pLi + Ri = EU(t)
i =
E
pL + R
U(t)
i =
E
L
_
1
p +
R
L
_
U(t)
1
See Appendix B
3
Let
R
L
= k. Then, using the properties of geometric series,
1
p + k
=
1
p
_
1
1 +
k
p
_
(3.1)
=
1
p

m=0
(1)
m
_
k
p
_
m
(3.2)
=

m=0
(1)
m
_
k
m
p
m+1
_
(3.3)
Recall that (
1
p
n
)U(t) = t
n
/n!. Then
i(t) =
E
L

m=0
(1)
m
_
k
m
p
m+1
_
U(t) (3.4)
=
E
L

m=0
(1)
m
(k
m
)
t
m+1
(m + 1)!
(3.5)
=
E
kL

m=0
(1)
m+1
(k
m+1
)
t
m+1
(m + 1)!
(3.6)
=
E
R

m=1
(1)
m
(k
m
)
t
m
(m)!
(3.7)
The MacLaurin series for e
t
is
e
t
=

m=0
(1)
m
t
m
m!
(3.8)
Then
i(t) =
E
R
(1 e
kt
) (3.9)
which can be checked by other means and shown to be the correct answer.
4 Criticism of Heavisides Operator Calculus
Both mathematicians and engineers found plenty to dislike about Heavisides operator calcu-
lus. Mathematicians found Heavisides methods to be too unrigorous, while engineers found
it too mathematically involved.
Regarding the mathematicans objections, we will touch upon two in particular.
4
The rst is that Heavisides algebrizing of dierential equations often necessitated series
expansions, which he implemented without taking into consideration the necessary condi-
tions. As an example, in the simple circuit problem shown in Section 3, the expansion of
1
1+k/p
as a geometric series was unjustiable. The geometric series identity is only valid for
|k/p| < 1, and it is not clear what this even means for an operator.
The second is that when continuous systems are considered (such as a telephone circuit of
innite length), fractional operators such as p
1
2
arise. Like operational calculus, fractional
dierentiation preceded Heaviside by centuries, but his development of the subject failed to
impress contemporary mathematicians.
As far as engineers were concerned, his work was little more than witchcraft. Sir William
Preece, Engineer-in-Chief of the British Post Oce, lamented Heavisides use of complex
mathematics in expounding his ideas such as self-induction in transmission lines [9, p.239].
...self induction in various forms has proved a bete noire which required all our
knowledge and all our skill, not only to master, but to comprehend.
It did not help that Heavisides publications were infamously dicult to decipher. His work
was lled with monstrously long and involved mathematics [7, p.xxv] and was incredibly
dense. His Electrical Papers contained 2, 500 pages of close type and more than 1.5 million
words and formulas [4, p.991]. Led by strong physical intuition, Heaviside rarely wrote down
proofs or detailed justications of his methods, which must have mad his work maddeningly
dicult to follow. Indeed, Nahin claims that readers tore their hair out trying to read him
[7, p.xxv], and Preece certainly agreed [9, p.238].
Mr. Heaviside...asserts (as I think somewhat impertinently) that it was very
much his own fault if Maxwell is still fully to be appreciated...Well, if Maxwells
expositions are anything approaching Heavisides in obscurity, no wonder.
5 Heavisides Conict with the Scientic Community
For his part, Heaviside did little to appease his critics. Heavisides attitude towards mathe-
matical rigor could be summarized by the following quote [Oliver Heaviside, Physical Math-
ematician]:
To have to stop to formulate rigorous demonstrations would put a stop to most
physico-mathematical enquiries. Am I to refuse to eat because I do not fully
understand the mechanism of digestion?
5
Perhaps it was unfair to expect Heaviside to have a good grasp of mathematical rigor. He
was, after all, a self-taught scientist with no formal education after the age of 16, whose
exposure to mathematics in school ended with basic algebra and trigonometry [9, p.232].
And had he lived a century or two earlier, when mathematical rigor was not held in such
high esteem, he might have escaped many headaches in this area.
Unfortunately for Heaviside, mathematicians of his era were highly concerned with rigor and
were in the process of putting the mathematics of their predecessors on rigorous foundation
[7, p. xviii]. Mathematicians at Cambridge, who held a particular dislike for Heaviside,
were at the time also focused on separating pure mathematics from applied subjects such as
physics and engineering.
Heaviside found such attitudes unbearable. Regarding the debate over pure vs. applied
mathematics he was of the opinion that [7, p. xviii]:
Physics is above mathematics, and the slave must be trained to work to suit the
masters convenience .
Heavisides conict with the mathematical community had large negative repercussions on
his career. The Royal Society refused to publish his work on operators in the Proceedings of
the Royal Society. As one anonymous mathematician wrote [3, p.56],
There was a sort of tradition that a Fellow of the Royal Society could print
almost anything he liked in the Proceedings without being troubled by referees:
but when Heaviside had published two papers on his symbolic methods, we felt
that the line had to be drawn somewhere, so we put a stop to it.
This embittered Heaviside greatly, to the extent that he harbored ill will towards British
academia even ten years later. In response to a Nature article on the relative liberty of
the French Academy of Sciences in contrast with the strict controls of the Royal Society,
Heaviside denounced England as a stick in the mud country...in love with tradition and
antiquated forms [7, p.225].
As far as publication was concerned, he fared little better with the engineering community.
Heavisides discoveries on inductance were blocked from publication by William Preece,
Engineer-In-Chief of the British Post Oce, whose theories Heavisides work would have
supplanted [3, p.59]. Heaviside did not hide his contempt for Preece in an article, Mr. W.
H. Preece on the self-induction of Wires, meant for publication in 1887 but rejected until its
inclusion in his Electrical Papers in 1892.
It will be remembered that Mr. Preece...was formerly an advocate of thin wires
of high resistance for telephony; but that, perhaps taught by costly failures in
his own department...he recently signied his conversion.
6
Such a personal attack was not to be seen in his disputes with the mathematicians, although
that was perhaps due to his clustering of all antagonistic mathematicians under the umbrella
term Cambridge mathematicians.
6 The Eects of Heavisides Attitudes on His Career
Had Heaviside maintained better communication skills and closer ties to the scientic com-
munity, he may have found greater appreciation for his work. While not a bitter recluse as he
is apt to be portrayed, Heaviside had a strong individualist streak that won him few friends.
In 1875 he was elected to the Council of the Society of Telegraph Engineers, and did not
attend a single meeting. Unsurprisingly, he was not re-elected the following year [3, p.56].
When he was elected a fellow of the Royal Society in 1891 on Lord Kelvins insistence, he
would not travel to London for the occasion. He would never visit anyone, including friends,
outside of his immediate family [3, p.56].
However controversial he was, Heaviside was widely read and his work broadly known. He
had his share of admirers, including Lord Kelvin, who as president of the Royal Society
had pushed for Heavisides inclusion in the organization before the Royal Proceedings spat.
When it was clear that Heaviside was falling into poverty and ill health after the death of
his parents and the loss of his old housekeeper, prominent scientists such as Lord Kelvin,
Lord Rayleigh, and Sir Oliver Lodge contacted powerful friends in government to give him
a government pension.
He maintained a healthy correspondence with some prominent scientists such as H. Hertz, but
was never very close to any of them. A particularly unfortunate case was that of T. Bromwich,
a Cambridge mathematician who was interested in the applied mathematics and sought to
provide a rigorous foundation for Heavisides operator calculus. He succeeded around 1916;
the result utilizes function theory and contour integrals in the complex plane [6, p.179].
Though Bromwich and Heaviside corresponded for several years, their relationship soured
when Heaviside became dismissive of Bromwichs mathematical complexities [6, p.180].
This was too bad, because it was around this time that Heavisides methods became broadly
accepted by the scientic community. In the early 20
th
Century his dispute with Preece
was settled when experiments showed that his theories on inductance were correct, and
further experiments gave credence to some other controverial theories of his. In 1905 the
University of Gottingen awarded him an honorary Ph. D., and he was rapidly inducted into
the Institution of Electrical Engineers and the American Institute of Electrical Engineers in
1908 and 1918 respectively. And when the Institution of Electrical Engineers was founded
in 1921, he became the rst recipient of the Faraday Medal [4, p.993].
7
7 The Legacy of Heavisides Operator Calculus
Heavisides methods were popularized among engineers during the rst half of the 20
th
Century by several textbooks on electrical engineering, such as J. Carsons Electric Circuit
Theory and the Operational Calculus, rst published in 1926 [6, p.268]. Being so soon after
Bromwichs work, none gave satisfactorily rigorous justication for the use of p and 1/p.
Within the mathematical community, the operator method was quickly displaced by the
Laplace transform, which was deemed mathematically rigorous. As mentioned previously,
the Laplace transform had been developed centuries earlier in a slightly dierent form, called
the Laplace transformation, and quickly forgotten. The search for alternatives to Heavisides
methods was no doubt the impetus for rediscovering and rening this technique. This was
made possible due primarily to the eorts of G. Doestch (1892 - 1977), who created the
modern Laplace transform. Recall that the Laplace transform is given by
F(s) =
_

0
f(t)e
st
dt (7.1)
Interestingly, the formula for the inverse Laplace transform can be derived using a Bromwich
integral, which was rst used to rigorize the operator method. Let us look at a particular
case here
2
.
Suppose that F(s) has nite singularities. Then F(s) is analytic on and to the right of line
Re{z} = a for some a R (see Figure 7.2).
By Cauchys integral formula,
F(s) =
1
2i
_
C
F(z)
z s
dz (7.2)
=
1
2i
_
Ca
F(z)
z s
dz +
1
2i
_
a+ib
aib
F(z)
z s
dz (7.3)
Consider only the rst component, the integral over the semicircle C
a
. We know that F(z)
is analytic on C
a
and therefore continuous there. Thus there exists some M > 0 such that
2
This example is adapted from a document written up by Jessica M. Conway, Mathematics Dept., Uni-
versity of British Columbia, 1984 Mathematics road, Vancouver, BC, Canada
8
Figure 7.2: Contour integral of Laplace Transform
|F(z)| M for all z C
a
. Then

1
2i
_
Ca
F(z)
z s
dz

1
2
_
Ca
M
|z s|
dz (7.4)

M
2
_
Ca
1
|z a| |s a|
dz (7.5)
=
M
2
_
Ca
1
b |s a|
dz (7.6)
=
M
2
_
b
b |s a|
_
(7.7)
=
Mb
2(b |s a|)
(7.8)
=
M
2(1
|sa|
b
)
(7.9)
We recognize that for all z C
a
, lim
b
Re(z) = . By Equation 0.9 (see Appendix C),
we conclude that lim
b
M = 0. Then
lim
b
M
2(1
|sa|
b
)
= 0 (7.10)
lim
b
1
2i
_
Ca
F(z)
z s
dz = 0 (7.11)
9
This gives us
F(s) = lim
b
1
2i
_
a+ib
aib
F(z)
z s
dz (7.12)
It can be shown that the unilateral Laplace transform is injective. Then
f(t) = L
1
{F(s)} (7.13)
= L
1
_
lim
b
1
2i
_
a+ib
aib
F(z)
z s
dz
_
(7.14)
= lim
b
1
2i
_
a+ib
aib
F(z)L
1
_
1
z s
_
dz (7.15)
=
1
2i
_
a+i
ai
F(s)e
st
ds (7.16)
Doestch wrote engineering texts as well as mathematical ones and contributed to the rapid
switch from Heavisides operator method to the use of Laplace transforms in engineering. By
1950, the exchange was almost complete [2, p.265]. This was a stunningly quick transition,
for the Laplace transform was at the front line of research as late as the 1930s.
Since operator theory had already been made rigorous by Bromwich and others who followed
him, why was the Laplace transform embraced so quickly? For the most part, it was due
to its ease of use compared to Heavisides operators. Tables of transforms and inverse
transforms could be compiled, and no further testing is needed on any individual problem
(as Heaviside often had to do, to make sure that his fuzzy mathematical methods did not
lead to unexpected results).
Though Heavisides operators are rarely if ever used today, and have not been a topic of
research for decades (with some exceptions, see [5]), they remain historically important as
mathematical tools that aided physicists and engineers before better methods arose. For all
the criticism they garnered, they were responsible for stimulating research into alternatives
such as the Laplace transform, as well as more modern form of operational calculus that
remain topics of active research.
10
Appendix A
Functions
Unit Step Function
U(t) =
_
0 if t 0
1 if t > 0
(0.1)
Dirac Delta Function
(t) =
_
+ if t = 0
0 otherwise
(0.2)
_

(t)dt = 1 (0.3)
The Dirac delta function is the derivative of the unit step function. It is not a true function
and can be best understood as a distribution or a measure.
11
Appendix B
Circuit Properties
0.1 Kirchos Voltage Law
The voltage around any closed loop in a circuit must be zero.
0.2 Voltage Laws for Circuit Components
Let V be voltage, I be current, R be resistance, C be capacitance, and L be inductance.
V = IR (0.1)
V = C
dI
dt
(0.2)
V =
1
C
_
t

I(t)dt (0.3)
12
Appendix C
Laplace Transform
The unilateral Laplace transform and its inverse are given by
F(s) =
_

0
f(t)e
st
dt (0.1)
f(t) =
1
2i
_
c+i
ci
F(s)e
st
ds (0.2)
For a function to be Laplace-transformable, it must meet the Dirichlet conditions
1. f(t) is piecewise continuous
2. f(t)e
st
is absolutely integrable, i.e.
_

0
|f(t)e
st
|dt <
The second condition is equivalent to the following:
3. |f(t)| Be

0
t
for some B R
+
,
0
R
With the above, we derive an important result. Let s = + i. Then
|F(s)| =

_

0
f(t)e
st
dt

(0.3)

_

0
|f(t)| |e
st
|dt (0.4)

_

0
Be

0
t
e
t
dt (0.5)
=
_

0
Be
t(
0
)
dt (0.6)
=
B

_
e
t(
0
)

0
(0.7)
=
B

0
, >
0
(0.8)
13
Which gives us
lim
Re(s)
F(s) = 0 (0.9)
Let us nd the Laplace transform of a simple function, f(t) = e
at
. This will be useful later.
L{f(t)} =
_

0
f(t)e
st
dt (0.10)
=
_

0
e
at
e
st
dt (0.11)
=
_

0
e
t(as)
dt (0.12)
=
1
a s
_
e
t(as)

0
(0.13)
=
1
s a
, s > a (0.14)
There also exists a bilateral Laplace transform, where the region of integration is (, )
rather than (0, ). For real-world signals, the unilateral Laplace transform is preferred.
14
Bibliography
[1] J. B. Calvert. Heaviside, laplace and the inversion integral, 2002. URL http://mysite.
du.edu/
~
jcalvert/math/laplace.htm.
[2] Michael A. B. Deakin. The ascendancy of the laplace transform and how it came
about. Archive for History of Exact Sciences, 44:265286, 1992. ISSN 0003-9519. URL
http://dx.doi.org/10.1007/BF00377050. 10.1007/BF00377050.
[3] D. A. Edge. Oliver heaviside (1850-1927) - physical mathematician. Teaching Math-
ematics and its Applications, 2(2):5561, 1983. doi: 10.1093/teamat/2.2.55. URL
http://teamat.oxfordjournals.org/content/2/2/55.short.
[4] Willis Jackson. Life and work of oliver heaviside. Nature, 165.
[5] P.A. Kullstam. Heavisides operational calculus: Olivers revenge. Education, IEEE
Transactions on, 34(2):155 166, may 1991. ISSN 0018-9359. doi: 10.1109/13.81595.
[6] Jesper Lutzen. Heavisides operational calculus and the attempts to rigorise it. Archive
for History of Exact Sciences, 21:161200, 1979. ISSN 0003-9519. URL http://dx.
doi.org/10.1007/BF00330405. 10.1007/BF00330405.
[7] Paul Nahin. Oliver Heaviside: The Life, Work, and Times of an Electrical Genius of
the Victorian Age. The Johns Hopkins University Press.
[8] J.J. OConnor and E. F. Robertson. Francois joseph servois, May 2000. URL http:
//www-history.mcs.st-andrews.ac.uk/Biographies/Servois.html.
[9] T. K. Sarkar. History of Wireless. Wiley-IEEE Press.
[10] Eric Weisstein. Operational mathematics, November 2011. URL http://mathworld.
wolfram.com/OperationalMathematics.html.
15

You might also like