This article was downloaded by: [New York University]

On: 06 February 2015, At: 00:31

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

The London, Edinburgh, and Dublin

Philosophical Magazine and Journal
of Science: Series 7
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/tphm18

CXVI. Some general theorems for

non-linear systems possessing
resistance
a
William Millar
a
Atomic Energy Research Establishment , Harwell
Published online: 15 Jul 2009.

To cite this article: William Millar (1951) CXVI. Some general theorems for non-linear systems
possessing resistance, The London, Edinburgh, and Dublin Philosophical Magazine and Journal
of Science: Series 7, 42:333, 1150-1160, DOI: 10.1080/14786445108561361

To link to this article: http://dx.doi.org/10.1080/14786445108561361

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
 [ 1150 ]

CXVI. S o m e General Theorems f o r N o n - L i n e a r S y s t e m s Possessing

Resistance.

B y WILLIAM MILLAR,
Atomic Energy Research Establishment, Harwell*.
[Revised MS. received June 8, 1951.]

SUMMARY.
This paper introduces some ideas and general theorems pertaining to
non-linear or mechanical systems. Thus in the case of a resistive network,
the dissipation is divided into two p a r t s - - t h e " c o n t e n t " and
" co-content " - - w h i c h are duals of each other. The dissipation itself
Downloaded by [New York University] at 00:31 06 February 2015

has stationary properties in linear b u t not (in general) in non-linear

networks, but it can be shown t h a t the " content " and " co-content "
have stationary and additive properties in the non-linear case. The idea
of " content " is extended to reactive systems, and it is shown that the
total content of any system in motion is an invariant.
This paper is intended as a companion to the paper by E. C. Cherry.

§ I. INTRODUCTION.
THE mathematical analysis of non-linear systems is difficult, and often
tends to obscure essential behaviour. An attempt is made in this and a
companion paper to present some new" concepts and general theorems
which might assist in forming a more physical approach to the analysis
and synthesis of systems possessing either incidental or intentional
non-linearities. The arguments are presented in terms of electrical
networks, but the concepts are equally applicable to mechanical and
electro-mechanical systems. The approach here is based upon the
methods of generalized dynamics, which in essence is a way of obtaining
the differential equations pertaining to a dynamical system. Much
work has already been done on the solution of non-linear differential
equations ; here, however, attention is concentrated, not on mathematical
methods of solution, but rather on examining the fundamental aspects
of the subject and trying to extract useful physical concepts from them.
One of the first theorems relating to linear networks was due to
Maxwell (1873), the " Minimum H e a t Theorem ". (See § 2.) I t can be
shown t h a t the theorem is not applicable, as it stands, to non-linear
networks ; the heat generated is not in general stationary, but a related
quantity called the " content ", which is proportional to the heat in
the linear case, is shown in §4 to have a stationary property. A
summation theorem is also given, on the " content " of two-terminal
networks. I n § 6 the idea o f " c o n t e n t " is e±tended to reactive networks,
where it is shown to be an invariant of the motion of such systems.
* Communicated by the Author.
 Some General Theorems for Non-Linear Systems Possessing Resistance 1151

§2. MAXWELL'S STATIONARY HEAT THEOREM.

In any purely resistive medium the current and voltage are distributed
according to the two continuity laws:
div i = 0 . . . . . . . . (1)
and
~ Eds=0 . . . . . . . . . (2)

T h e analogous conditions in a l u m p e d circuit are given b y " Kirchhoff's

current l a w "
2:i=0 at a vertex . . . . . . . (3)
and " Kirchhoff's voltage l a w "
Downloaded by [New York University] at 00:31 06 February 2015

Zv----0 r o u n d a n y mesh . . . . . . . (4)

Maxwell's t h e o r e m (Maxwell 1873) for l u m p e d circuits, p u t in more m o d e r n
form, states t h a t if any linear resistive circuit is driven by voltage or current
generators or both, then of all possible distributions of current consistent with
Kirchhoff's current law, that which gives a minimum to the quantity
( W - - 2 P v ) is the only one consistent with Kirchhoff's voltage law, and is
therefore the actual distribution. H e r e W is the t o t a l power dissipated in
the resistors, and P v is the t o t a l power t a k e n from the voltage generators.
H e n c e for a purely current-driven n e t w o r k the t o t a l heat is stationary.
Notice t h a t Kirchhoff's current law is observed t h r o u g h o u t , b u t t h e
voltage law must be violated in taking the variations.
A n e t w o r k with m meshes m a y be considered to h a v e m degrees of
f r e e d o m for currents. A n y m n o n - i n t e r d e p e n d e n t c u r r e n t variables can
therefore be chosen as generalized coordinates to define the state of t h e
n e t w o r k ; these currents could for instance be all the mesh currents,
or some o f the b r a n c h currents. I f t h e y are Jr, where r = 1, 2, . . . m~
t h e n Maxwell's t h e o r e m leads to the m equations

ai-: ( w - 2 P v ) = 0 . . . . . . . . (5)

which are sufficient to determine all the currents (and hence all the
voltages) in the network.
T h e dual of Maxwell's t h e o r e m is often.useful, b u t the two theorems
are sometimes confused. H e r e Kirchhoff's voltage law is observed
t h r o u g h o u t , a n d the current law v i o l a t e d ; the s t a t i o n a r y q u a n t i t y is
now ( W - - 2 P I ) , where P1 is the power t a k e n f r o m the current sources.
T h e n as in equatio n (5), if we have n generalized voltage coordinates v~
in a n e t w o r k o f ( n + l ) vertices, we h a v e the n equations
a
(w-2P )=0 . . . . . . . . (6)

which again tell us e v e r y t h i n g a b o u t the network.

 1152 William Mfllar on some General Theorems

Maxwell's t h e o r e m applies only to linear networks, and it can in fact

be shown, b y t h e m e t h o d s o f § 4, t h a t ( W - - 2Pv) is not in general s t a t i o n a r y
if any of the elements are non-linear. T h e search for some other q u a n t i t y
which does have the s t a t i o n a r y p r o p e r t y forms the starting-point here.

§ 3. DEFINITIONS RELATING TO NoN-LINEAI~ NETWORKS.

3.1. Elements.--A v e r y general t y p e of 2-terminal element m a y be
specified b y a relation of the form

( . d i d2i dv d2v )
f ~ , ~ , d t 2 , . . . . v, dr' dt 2 . . . . t = 0 , . . . . (7)
Downloaded by [New York University] at 00:31 06 February 2015

where i and v are the current and voltage at the terminals. I f the
time-derivatives are absent the element is said to be non-reactive, a n d is
then specified b y a relation of the form f(i, v, t)-----O. P r o b l e m s in which
all the elements are non-reactive can be solved at any instant in t e r m s of
time-invariant non-reactive elements specified b y a relation of the f o r m
f(i, v ) : 0 . Discussion of non-reactive elements is therefore confined to
this latter case, in which elements are specified at all times b y the single
characteristic, f(i, v ) : O . Such an element is said to be "passive " if
the characteristic cuts the i-v axes n o w h e r e e x c e p t a t the o r i g i n ;
otherwise it is said to be "active ". I t is sometimes desirable to
distinguish between ":simple " and " non-simple " elements, according
as to whether or n o t the c u r r e n t a n d voltage are e v e r y w h e r e single-valued
functions of each other.
3.2. Generators.--A "generator" is a particular form of active
non-reactive element in which either the current or the voltage is a b s e n t
from the defining function f(i, v). T h u s generators are either of t h e
" constant voltage " or " c o n s t a n t current " type. I n w h a t follows,
expressions such as " all the elements of the n e t w o r k " will include the
generators uniess s t a t e d otherwise.
3.3. Networks.---A "passive non-reactive network " is defined to be one
which contains only passive non-reactive elements. I t is assumed t h a t
if two vertices of such a n e t w o r k are chosen as terminals, the whole
network between the terminals m a y be regarded as a n element with its
own passive non-reactive characteristic.
3.3.1. Although Thdvenin's n e t w o r k t h e o r e m is well known, certain
points which h a v e a bearing on this discussion are w o r t h clarifying.
This useful t h e o r e m really consists of two p a r t s ; as usually p r o v e d t h e y
are (for non-reactive n e t w o r k s ) :
(i.) A 2-terminal linear active resistive n e t w o r k can always be replaced
b y a new 2-terminal n e t w o r k consisting only of a constant-voltage
generator El, in series with a linear resistor R 1, so far as a n y linear
element outside t h e terminals is concerned.
 for Non-Linear Systems Possessing Resistance 1153

(ii.) Further, E 1 is given by the open-circuit voltage at the terminals

of the original network, and 1~1 is given by the resistance at the terminals
of the original network when the voltage generators are short-circuited
and the current generators removed.
I f the load is non-linear, the theorem may be shown still to apply,
although the usual proof for the linear case is no longer valid. I f the
network is non-linear the theorem of course does not apply ; all that can
now be said in place of part (i.) of the theorem is that, for external purposes,
the 2-terminal network m a y be replaced by a single active non-linear
element (-the "equivalent" element).
3.4. " Content " and " co-content ".--Suppose that a current i 1 is
passing through an element specified by the characteristic f(i, v)-=O,
such as that shown in fig. 1 (a). The voltage across it will then be vl,
Downloaded by [New York University] at 00:31 06 February 2015

where f(i~, vi)-=O. We now define a quantity G, which we shall call the
" content " of the element when it is in the particular state (iD v~), by
the integral

The content has the dimensions of power. The dual quantity, obtained
by interchanging current and voltage everywhere, we shall call the
" co-content ", denoted by J. Thus

The integrals G and J are shown as areas on fig. 1 (a). I f the element is
passive (fig. 1 b), G and J are respectively the areas below and above the
characteristic curve. I f the total dissipation in the passive element is W,
it can be seen from fig. 1 (b) t h a t
W----ilVl=G+J . . . . . . . . . (8)
A linear passive element is a special kind of passive element in which
f(i, v)=--gi--v, where R is a property (the " resistance ") of the element.
In this ease, G = J ~ W / 2 for all values of current and voltage. Fig. 1 (e)
shows the characteristic of an element which is a constant-vol~ge generator.
I n this ease, G-----P~, where P~ is the power supplied by the generator;
no meaning can be attached to J. Similarly, fig. I (d) shows a constant-
current characteristic. Here J~----PI, while no meaning can be attached
to G. I f the elements are non-simple, G and J must be interpreted with
care.

§ 4. THEOREMS OF " CONTENT " AND Co-CONTENT ".

"

4.1. In this section four network theorems are given. Theorems III.
and IV. are the duals of I. and II., and will be stated without proof since
they follow from the Principle of Duality. These theorems apply at
every instant of time.
 ] 154 William Millar on, some General Theorems

T h e o r e m i iI.i . } "
The co-content
content \J of a two-terminal network, regarded as
co-contents \
an element between the terminals, is the sum of the contents J of the
constituent elements.
Theorem IV. " I f in an active non-reactive network the sum j
o} of the
contents }
co-contents of all the constituent elements is expressed in terms of the
.. .current~ ..
defining number of general~zect voltage ~ coordinates in the network, subject
current\ G}
only to the restrictions of K irchhoff' s voltage J law, then is stationary for
Downloaded by [New York University] at 00:31 06 February 2015

currents \
the actual distribution of voltages J"
Fig. 1.
(i,v ) = o .y(/,v)=O
i,,V,)

0 g
"--.'7-"-- gj

(a) Active element. (b) Passive element.

y(i.v)=O
7J~ (it,t),)
............... (i,,v,)

vt G J

0 ~ Z= 0 __.,.. i,
(c) Constant-voltage generator. (d) Constant-current generator.
Typical characteristics of various kinds of element, showing content, G,
and co-content, J, as areas.
4.2. T o prove these theorems, Kirchhoff's two laws are used. F o r
T h e o r e m I., consider a n e t w o r k such as t h a t shown in fig. 2, in which each
branch is a non-reactive element. These branches m e e t in ( n + l) vertices,
 for Non-Linear Systems Possessinq Resistance 1155

n u m b e r e d 0, 1, 2, . . . . j, #, . . . . n. T a k e the vertices 0 a n d h a s
terminals, a n d let a c u r r e n t i o be flowing in a t v e r t e x 0 a n d out at v e r t e x n.
L e t the resulting voltage a t v e r t e x j be % (This use o f " v e r t e x voltages "
a u t o m a t i c a l l y i n c o r p o r a t e s K i r c h h o f f ' s voltage law in the proof.) W i t h o u t
loss of generality, we m a y p u t v,~--~0. Also let the resulting current in
b r a n c h j/c, m e a s u r e d f r o m v e r t e x j to v e r t e x #, be i~k. N o w suppose
t h a t i 0 is increased slightly b y an a m o u n t Ai o. All the b r a n c h currents
wit] t h e n change b y a n a m o u n t /lijk , a n d the v e r t e x voltages b y a n

[ 0
a m o u n t /Ivy.. I t follows f r o m K i r c h h o f f ' s current law t h a t

~i~= ~io, j=o, . . . . . (9)

k=O
Downloaded by [New York University] at 00:31 06 February 2015

-- Zli o, j = n .
Fig. 2.
O I

|j |.J . j. [~,,

I t
,-I
J| ,s--
• L/ t I
:
!
.... L -q ,¢ ....
i i

rq

fl
t I

r-, [ ....
1
i
I
O v

- L i
A typical network, showing vertices j, k and branch jk.
T h e change AGjk in t h e content of each e l e m e n t is given, f r o m T a y l o r ' s
theorem, by
z l G ~ = ( v j - - v ~ ) ~ i i ~ 4-½ Aij~( ~v ~ - /lvk)+ . . . .
= (v~--vk) Aij~, to first order.
 1156 William Millar on some General Theorems

Now since i j k = - - i k ~ , we have AGjk----zJGkj, and hence, if G is the sum

of the contents of each element, we have, to the first order,

2 A G : Z X AGj~
j k

= Z Z (vj--vk)zJij~
j

- ~ Z Z vjzJij~ ~ 2 Z Vk,~ikj
jk kj

= 2 Z X v~,di~k (from symmetry o f j and k)

j k
Downloaded by [New York University] at 00:31 06 February 2015

~-2vozlio, from equations (9), since v , ~ 0 ,

io eo A G = v 0 ~ i 0, to first order.
dG
Hence di--~o--_Vo
. . . . . . . . . . . . . . . . (10)

But if G' is the content of the network, regarded as an element between

the terminals, we have
dG'
dio --_v o. . . : . . . . . . (11)
Hence, from (10) and (11),

G - - G ' = (G)io=0-- (G')i0=0. . . . . . . (12)

Now it follows from the definition of content t h a t (G')i.=o----0. I f all the
elements are passive, then (ijk)io=o-~O, and so (G)io=0:0. Therefore
from (12) we have G--~G', which proves Theorem I. Notice t h a t we have
proved it only for passive networks ; it can be shown t h a t the theorem
does not hold (in general) if the network is active.
4.3. For Theorem II., again consider such a network as t h a t shown in
fig. 2, but without special terminals. Any of the elements m a y be active.
It will be assumed that only one element joins any two vertices, since
any network m a y be reduced to this form. I f there are any constant-
current generators they are not considered as branch elements, since
variations cannot be made in the currents through them. As before,
we write iCk for the actual values of the branch currents, and v~ for the
actual values of the vertex voltages. Now consider an adjacent set of
current values (ijk-F/liCk); the variations zlij~ will be equivalent to
variations in generalized current coordinates, provided t h a t they are
consistent with Kirchhoff's current law, i. e. provided t h a t

zJij~-0 . . . . . . . (13)
k=0
 for Non-Linear Systems Possessing Resistance 1157

(The adjacent current values can only obtain if Kirchhoff's voltage law is
violated, and hence cannot themselves be also actual values.) Let the
variation in the content of each element be AG~k, and in the sum of all
the contents be AG.
We have
2z~G----Z 27 AG~k
i k
~ZZ (v~--vk)Aij~ , to the first order, and hence, by the
i k
same reasoning as used in the proof of Theorem I.,
Downloaded by [New York University] at 00:31 06 February 2015

whence A G = 0 , from (13).

This proves the theorem.
4.4. I t should be noted t h a t either Theorem II. or Theorem IV. is
sufficient to determine all the currents and voltages in a network. This
is done by using the equations corresponding to (5) and (6) in § 2. They
are
0G
0i--~~-0 . . . . . . . . (14)
and
~J
. . . . . . . . (15)

where r~- 1, 2 . . . . . m and j---- 1, 2 . . . . . n, for a network of m meshes and

( n : l ) vertices. The expressions in (14) and (15) are written as partial
derivatives with respect to independent generalized coordinates, such as, for
%

example, mesh currents or vertex voltages. (In computing the number m

J

current \ }
of generalized voltage J coordinates, constant voltage
current generators, if
any, are not counted as branches, but they appear as making contributions
currents "~
of the branch voltages J when the latter are expressed in terms of the
generalized coordinates.)
An example of the case n----1 is given in the next section.

§ 5. A SIMPLE PRACTICAL EXAMPLE OF TI~EOREM IV.

This section illustrates the rather trivial one-variable case of
Theorem IV. Consider the " scale-of-two " circuit, whose essentials are
shown in fig. 3 (a). The right-hand portion of this circuit can be treated
as a passive element, which m a y be shown to have the characteristic
given by the full line in the upper part of fig. 4. I f now the two halves
of the circuit are connected together, the voltage across the right-hand
 1158 William Mfllar on s o m e General T h e o r e m s

element will t a k e u p one of t h e " a c t u a l values " c o n s i s t e n t - w i t h

Kirehhoff's t w o laws. T h e r e are t h r e e such values, as m a y be seen b y
drawing t h e " load-line " for R, s h o w n d o t t e d in fig. 4. (The question
Fig. 3.
A A

i
..k.--
D

l
I

f
Downloaded by [New York University] at 00:31 06 February 2015

T I

(a) (6)
Essentials of " scale-of-two " circuit. Schematic of fig. 3 (a).
Fig. 4.

O/
\

"', SLOPE = - I/R

",oE
i
)
\ J

i ~ Ji÷J2
I w ~ J 2

Co-contents J1 and J2 of the elements in fig. 3 (b), plotted as a function of

the voltage variable v. The sum (Jl-~-J2) is also shown.
of s t a b i l i t y is n o t discussed here since it i n v o l v e s t i m e s ; in a physical
circuit, the s t a b i l i t y of each of these t h r e e positions d e p e n d s u p o n
reactive elements i n e v i t a b l y present.) N o w consider t h e n e t w o r k to be
 f o r N o n - L i n e a r S y s t e m s Possessing Resistance 1159

d r a w n schematically as in fig. 3 (b). This is a two-element passive network

fed b y a voltage generator ; it has only one degree of freedom for voltages,
a n d so, taking the voltage at B as zero, the voltage v at A m a y be chosen
as variable. If, without using Kirchhoff's current law, the co-contents
J l and J2 are plotted as functions of v, the curves shown in the lower
h a l f of fig.. 4 are obtained. Also shown on the same figure is J, which is
the sum of J~ and J2- I t will be seen t h a t J has s t a t i o n a r y values at,
a n d only at, the " actual values " of v.

§ 6. ~I~VAI~IA:NTS OF THE MOTION FOR GENERAL CIRCUITS INCLUDING

I~EACTA~ICE.
Consider now a network consisting of general elements of the t y p e
defined b y equation (7), b u t with t not appearing explicitly in the defining
Downloaded by [New York University] at 00:31 06 February 2015

function fi Suppose t h a t a complete solution has been obtained for all

t h e currents and voltages in the network. Then, for each element, an
i - v diagram m a y be drawn ; this diagram is not in general a characteristic
of the element, but represents a t r a j e c t o r y in a t y p e of so-called " phase
space ". The change in content of the element in going from the position
(il, vl) to the position (i2, v~) m a y be defined by the integral

G2--GI----- vdi . . . . . . . (16)

il
a n d similarly
J 2 - - J l = / "~' i d v . . . . . . . . (17)
J ~)1
W e are interested here in studying only the change in content as the
currents Mter with time, so we m a y simply talk of the " content ", G,
w i t h o u t concerning ourselves with the original state to which it is referred.
Now Theorem II. as stated applies only to non-reactive elements, b u t
there is nothing in the a r g u m e n t to exclude reactive elements, since the
theorem depends only upon Kirchoff's laws. I f t h e n we write G or J for
the total content or co-content of the network (i. e. the sum of M1 the
Gj~'s or J~.k's), including t h a t of a n y generators, we have from (14) and
(15)
0G
ai-: =0 . . . . . . . . . (as)

and
OJ
Ov--~=-O. . . . . . . . . . (19)
where r = 1, 2 . . . . . m, and j=- 1, 2 . . . . . n. The total time-derivatives of
G and J are given by
dG OG 3G di 1 OG die OG dim
(20)
dt - - at q-Oi 1" dt q - - Oi 2" -dt
- + . . . . -~ ai m" dt
and
dJ OJ OJ dv 1 OJ dv2 OJ dv,~
dt q- -av
d-t ----- ~ - t ~f- O v 1 " - e" - -dt q- . . . . q - O v . " dt " " (21)
 1160 Some General Theorems for Non-Linear Systems Possessing Resistance

Consider now any period of time during which all the voltage and current
generators, if any, have steady voltages and currents respectively. During
this time, t does not appear explicitly in the defining functions of any o f
~G ~J
the elements, and so ~ - = ~ / = 0 . Hence from equations (18), (19),
(20) and (21),
dG
d--T= 0
and
dJ
d--/:0.

We can state this result as follows : -

Downloaded by [New York University] at 00:31 06 February 2015

Theorem V. I n any network of time-invariant elements (possibly

including generators), the total content and the total co-content are invariants
of the motion. This theorem could, for instance, be applied to any circuit
for all time following an impulsive change in a current or voltage generator,
or in, say, a capacitor element. Note that it is very general, and applies
even to dissipative systems in which the total energy is not invariant.
The content and co-content, used in the above sense, are akin to integral
invariants ; information on these is given b y Whittaker (1904).

ACKNOWLEDGMENTS.
The author is indebted to Mr. E. C. Cherry, of the Imperial College of
Science, London, for much stimulating discussion and valuable help;
and to the Director of the Atomic Energy Research Establishment,
Harwell, for permission to publish this paper.

REFERENCES.
MAXWELL,JAMES CLERK, 1873, A Treatise on Electricity and Magnetism, Vol. I.
Third Edition, p. 407. (See also footnote by J. J. Thomson on p. 408.)
WHITTAKER,EDMUND, 1904, Analytical Methods in Dynamics, Chapter X.