The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

Action on Stage: Historical Introduction

Jon Ogborn, Institute of Education, University of London, UK

Jozef Hanc1, Institute of Physics, P.J. Safarik University in Kosice, Slovakia,
Edwin F. Taylor2, Department of Physics, Massachusetts Institute of Technology, USA

Abstract
The action principle is a powerful tool for understanding, applying, and building
bridges among fields of physics, from quantum theory through relativity to current
research. We dramatize those who devised the action principle and its precursors –
Fermat, Huygens, Maupertuis, Euler, Hamilton, Einstein and Feynman – with the au-
thors performing the roles of these great physicists and mathematicians. We accept
no responsibility for the accuracy of the words of our characters! This is an effort to
introduce fundamental physical principles, not to reconstruct the actual historical de-
velopment of these principles.

Action on Stage

Animateur:
This symposium is about building bridges between things students would like to learn– rela-
tivity, quantum theory, particle physics – and things they have to learn – notably classical me-
chanics. We are interested in simplicity and unity in physics, as well as with exciting students
about physics.
The idea that links these topics is the concept of stationary action.
Now we have a problem. Either you know nothing at all about the physical quantity
called action, or you learned it in a difficult course of theoretical mechanics. Both will make
you hostile to our proposals. Either, “I never studied it, so it can’t be important”, or “Any-
thing I don’t understand must be too hard for students”.
So – we have to tell you about these ideas, starting from zero. How better than to ask
the people who invented them to explain what they were doing? First up is the Frenchman
Pierre de Fermat.

Pierre Fermat (1601-1665):

Although I never published a scientific paper, my reputation as one a
leading mathematician came from my correspondence with other sci-
entists and from them publishing my ideas and methods in their work.
I am known primarily for my work in number theory. I also developed
analytic geometry independent of Descartes and worked in many
other mathematical fields – completely as an amateur.
I had a terrible fight with Descartes. He thought that light is
transmitted instantaneously from point to point "like the cane of a
blind man," so I had to express my optical theory in terms of "resis-
tance" of different media through which light passes. You have no Pierre de Fermat
such difficulty, and Fermat's principle of least time is the oldest varia-
tional principle; one that you still use. The idea is simple: the path that light takes is just the
one that takes the least time. Among all possible paths, the minimum total time picks out the
unique path between fixed initial and final points. What could be easier?

Animateur:
Monsieur Fermat, there’s an obvious objection to your idea. How does the light know in ad-
vance which path will be the quickest?

1
email: <[email protected]>
2
email: <[email protected]>, website: <http://www.eftaylor.com>

Ogborn, Hanc, Taylor 1

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

Fermat:
When I was alive I could not answer your question. The objection was not overcome until
long after my death, when you came to see that every point on an advancing wave acts as a
source of little wavelets. Then between point source and point detector the wavelets add up
with coherent phase along the path of stationary time. I hope you will tell us how some
Dutchman figured it out.

Animateur:
We had hoped that, as a Dutchman, Christiaan Huygens could join us
here in Amsterdam, but unfortunately he is away at the Royal Court
in France. His big idea was that light is a wave, and that where the
wave goes next can be predicted by supposing that each point on the
wave front acts as a source of little wavelets. The many wavelets all
superpose, adding up in constructive interference to generate the new
wave front, but canceling in destructive interference everywhere else.
Centuries later, Richard Feynman was to adapt the same idea to build
a new formulation of quantum mechanics: the “many paths” ap-
proach. Christiaan Huygens
Now we jump a hundred years, and our next guest is another
Frenchman, Pierre Louis de Maupertuis.

Maupertuis (1698-1759):
I am Pierre-Louis Moreau de Maupertuis. My father, a wealthy pirate,
gave me every advantage. I led an expedition to Lapland to measure
the length of a degree along the Earth’s meridian, proving that the
Earth is hamburger-shaped. Its fame led to my becoming president of
the Prussian Academy and a favorite in the court of Louis the fif-
teenth.
I conceived the principle of least action, that in all events of
Nature there is a certain quantity, called action, which is always a
minimum; that collisions of bodies or refraction of light occur in such
Pierre de Maupertuis a way that the amount of the quantity mvs is as little as possible. My
original definition of action as the product of mass, speed and distance
traveled by a moving object was later restated by my friend Leonhard Euler [see eq. (1)].
My paper was titled, "The laws of motion and rest deduced from the attributes of
God" and stated: "Here then is this principle, so wise, so worthy of the Supreme Being:
Whenever any change takes place in Nature, the amount of action expended in this change is
always the smallest possible." I am horrified to hear that people think that action can some-
times be a saddle point; I reject this idea entirely because the perfection of God is incompati-
ble with anything other than utter simplicity and minimum expenditure of action.

Animateur:
Ignorant people say that nothing of intellectual distinction greater than the cuckoo-clock ever
came out of Switzerland. To give them the lie, we now hear from the great Swiss mathemati-
cian Leonhard Euler, who supported and developed Maupertuis’ idea.

Leonhard Euler(1707-1783):
It is no boast to say that I am the most prolific mathematician of all
time, producing about 900 papers and books in my lifetime. I spent my
years largely in the courts of the Tzars of Russia and in the court of
Frederick the Great.
Maupertuis is a great buddy of mine, but sloppy in formulating
his action principle. I realized that without the law of conservation of
energy the action quantity of Maupertuis loses all significance. So I

Leonard Euler

Ogborn, Hanc, Taylor 2

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

cleaned it up, formulating the principle of least action as an exact dynamical theorem and giv-
ing his action a correct mathematical form:
final
position

W= ∫ mv ds
initial
(assume energy conserved), (1)
position

(The integral is calculated along a path of a moving particle.)

My statement “since the plan of the universe is the most perfect possible and the work
of the wisest possible creator, nothing happens which has not some maximal or minimal
property!” was my acknowledgement of Maupertuis as originator of the action principle.
I also developed a simple, intuitive, geometrically understandable way of finding the
minimum or stationary action path (see figure 1)

Figure 1. Euler realized that if the action integral is minimal along the entire path, it must also
be minimal for every subsection of the path: triplets of nearby points on the path, e.g. mno in
my figure. Minimal action means that any change in the path, e.g. point n varied slightly to
point v, leads to zero first order change in action. If this condition is be satisfied for each triplet
and we go to the limit in which lengths of segments tend to zero, we get a differential equation
(the Euler-Lagrange equation), whose solution is the stationary action path.

Later, I helped the career of the young Joseph Louis Lagrange (1736-1813), who wrote to me
about his elegant mathematical way to express conditions of minimum action. His ideas led
me to drop my intuitive graphical approach and coin the phrase "calculus of variations". Great
for mathematics and theoretical physics, but a disaster for physics education! Lagrange's ab-
stract method has dominated your advanced mechanics classes. Too bad, because my graphi-
cal method is perfect for modern computers (fig. 2)

Ogborn, Hanc, Taylor 3

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

Figure 2. The universality of Euler’s graphical approach is demonstrated by this computer

display used in modeling Fermat’s principle. Click the computer mouse to select an arbitrary
moveable intermediate point on the path, then drag the point up and down, looking at the value
of the total time, to find the minimal (stationary) time of that point. Then do the same for other
points, cycling through them until the time for each results in the least (stationary) value of the
total action or time. This method of successive displacements or hunting for the least time path
is straightforward but tedious. However the task can be done quickly by computer.

Animateur:
We jump a hundred years again. Here is the Irishman William Rowan Hamilton to tell you
about his new version of action, more powerful than ever.

Hamilton (1805-1865):
My name is William Rowan Hamilton. I wanted to develop a com-
mon mathematical language for particles and waves, so starting from
Fermat's least time principle and using the Lagrangian, I found what
you call Hamilton's action S:
final final
event event

S= ∫ L dt = ∫ ( K − U ) dt
initial initial
(2)
event event

Maupertuis’ action, remember, determines trajectories in space be- William R. Hamilton

tween fixed initial and final locations and requires that energy be
conserved. In contrast, my action principle determines worldlines in spacetime between initial
and final events and is true even if the potential energy is a function of time as well as posi-
tion, in which case the energy of the particle may not be a constant. I understood that action
along an actual worldline is not necessarily a minimum but is always stationary compared
with action along adjacent alternative worldlines between the same fixed initial and final
events. Actually the word “worldline” is a stranger to me; it took Albert Einstein to make the
term important. For me it was just the path between fixed places and times – between what
Einstein called fixed events.

Ogborn, Hanc, Taylor 4

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

Einstein:
I, Albert Einstein, am Swiss by nationality – another blow to cuckoo-
clock theory! It was my idea that the physical world has to be structured
as space-time events. I emphasized the fact that such events are con-
nected by worldlines in space-time. I could show that the natural, un-
forced, path from one event to another was that for which ageing –
wristwatch time – is a maximum.
In relativity, the Hamilton action S for a free particle is just:
final
event
Albert Einstein S = −mc 2 ∫ dτ
initial
(3)
event

Minimal (or stationary) action along a real worldline makes the total proper time τ = ∫ dτ
maximal (or stationary). Therefore, because of the minus sign in front of the integral, the rela-
tivistic principle of least action is the same as the principle of maximal proper time, called by
Dr. Taylor the principle of maximal aging.
Moreover it is not difficult to show that for small velocities Eq. (3) gives the same re-
sults as classical nonrelativistic Hamilton action.

Animateur:
You will have noticed that these contributions all came from Europe. But in the last century,
American physics blossomed, and one of its finest products was Richard Feynman, who com-
pletes our story.

Richard Feynman:
When I was in high school my physics teacher Mr Bader told me that
Newton’s laws could be stated not only in the form F = ma, but also in
the form: “average kinetic energy minus average potential energy is as
little as possible for the path of an object going from one point to an-
other.” I next got involved with least action as a PhD student with
John Archibald Wheeler. This led to the development of the “many
paths” version of quantum mechanics, a third formulation mathemati-
cally equivalent to the Schrödinger and Heisenberg versions. It also
helped me develop my “Feynman diagrams” for doing calculations in Richard Feynman
quantum electrodynamics. For this work I shared the Nobel Prize with
Schwinger and Tomonaga.
So how does a mindless particle recognize the least action path or worldline? Does it
smell neighboring paths to find out whether or not they have increased action? According to
my formulation, yes! The electron explores all worldlines between source and detector. For
each possibility there exists a little rotating stopwatch whose hand, or arrow, makes a total
number of turns equal to Hamilton’s action S divided by Planck constant h (see Fig. 3). The
Lagrangian L, divided by h, is nothing other than the rate of arrow rotation.

Ogborn, Hanc, Taylor 5

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

Probability
amplitude

Figure 3. In the "many paths" version of quantum mechanics the electron explores all possible
worldlines from initial emission event to final detection event. The figure shows a single one of
these worldlines. Along this path a little stopwatch hand rotates at the rate L/h, leading to a
contribution to the final amplitude at the detection event.

All these quantum arrows (probability amplitudes) add up constructively (line up) if
they have similar phases. This is so for worldlines close to the stationary action path (the blue
pencil of paths in Fig.4). The arrows cancel out or curl up for other sets of worldliness, as you
can see in a piece of Dr. Hanc’s program. The bigger the mass of an object the narrower is the
pencil of nearby worldlines that significantly contributes to the resulting amplitude.

Probability
amplitude

Figure 4. Some of the infinite number of possible worldlines connecting fixed initial and final
events. The squared magnitude of the resultant arrow is proportional to the probability of de-
tecting the particle at the final event. As the particle mass increases, the pencil of worldlines
contributing significantly to the resultant arrow (shown in blue) becomes narrower and nar-
rower, approaching the single path of classical mechanics.

Animateur:
Christiaan Huygens has sent me a letter claiming priority for your “many paths” idea. He
claims that it is just his idea of wavelets, in modern clothing. What do you say to that?

Ogborn, Hanc, Taylor 6

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

Richard Feynman:
It’s true (I acknowledged Huygens in my PhD thesis). However the use of Planck’s constant,
and the fact that the idea also works for particles like electrons or atoms, goes beyond Huy-
gens. The idea for the number of quantum stopwatch rotations came from Paul Dirac.

Animateur:
OK, that’s it. We will be describing in the papers that follow, how the scalar quantity action
adds to the physicist’s toolkit for analyzing and predicting motion. It looks like this:

1. Use Maupertuis action W when we fix in advance the initial and final POSI-
TIONS, and energy is conserved.
2. Use Hamilton action S when we fix initial and final EVENTS and energy is
may or may not be conserved.
3. Use Newton or Lagrange when we do NOT know where the motion is going
from its initial conditions.
4. Use Newton when friction is significant, so vectors are inevitable.

We hope that we have started to break up some of your thought-glaciers about action.

References
Background papers with historical references are available at the web-site:
http://www.eftaylor.com/leastaction.html

Photos

Ogborn, Hanc, Taylor 7

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

What is the action model?

Introducing and modeling principles of least action.

Jozef Hanc1, Institute of Physics, P. J. Safarik University in Kosice, Slovakia,

Abstract
The Action Principle predicts motion using the scalars energy and time, entirely
avoiding vectors and differential equations of motion. Action is the tool of choice
when we want to specify both initial and final conditions. Maupertuis-Euler action
finds the trajectory when initial and final positions are prescribed in advance, but re-
quires that energy be a constant of the motion. Hamilton action finds the worldline
when initial and final events are prescribed in advance and easily describes motion
when potential energy is a function of time as well as position. A simple toolkit of
motion tells us when to use action, when to use Lagrange's equations, and when we
must return to the vector methods of Newton. The original Euler method of handling
action also provides a basis for computer modeling. Interactive software allows stu-
dents to employ basic concepts of the principle of least action and increase concep-
tual understanding.

Least action approach in teaching

The least action principle approach, so important for modern physics, is widely considered to
be a difficult topic and is usually only used in advanced mechanics textbooks and courses.
Why does it seem a peculiar way to introduce and teach classical mechanics? Why is action as
a physical quantity understood as being very abstract and unsuitable for introductory physics,
despite the fact that it is a scalar very similar to energy − one of the central concepts of intro-
ductory courses?
The reasons are that in the majority of standard advanced texts [like Landau & Lifschitz
1976, Goldstein et al 2002, Marion & Thornton 2003, Hand & Finch 1998], ⎯ (1) the
mathematics used is the calculus of variations, which is not part of the common mathematical
toolkit acquired at introductory college level; (2) action is usually introduced extremely
briefly and is only used for a quick variational derivation of Lagrange’s equations; (3) action
is not immediately illustrated by examples; texts typically include no or a very few examples;
(4) there is also no study of the properties of action after introducing it, since they are taught
at the end of courses and texts, (5) and finally you find no computer modeling, which means
that students cannot obtain direct experience and intuition.
So the important question is how to introduce least action principles? Our experience says
that it is possible in the frame of introductory courses provided that we concentrate on the
following crucial issues:
• Starting with one dimensional cases and using a powerful graphical language. More
general cases only bring in more complicated mathematical expressions, but essen-
tially nothing new in physics ideas
• Using concrete, but easily generalizable examples. We have chosen Newton’s falling
apple, but the arguments will work for any reasonable potential energy function.
• Building a clear connection to Newton’s laws in terms of comparison.
• Using ingeniously simple original arguments of the greatest physicists and mathema-
ticians of all times: (1) Newton’s argument from his celebrated Mathematical Princi-
ples of Natural Philosophy (1687) and (2) Euler’s argument from his pioneering work
on the variational calculus The method of finding curved lines enjoying properties of
maximum or minimum (1744). As result we will not need advanced mathematics (all
arguments require only high school algebra such as expressions (a ± b)2 and basic
properties of parabola). Moreover we also obtain an excellent foundation for com-
puter modeling, which is important in getting good intuition and experience.

1
email: <[email protected]>

Ogborn, Hanc, Taylor 8

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

So how does classical mechanics explain the motion of a falling apple? We will show
three different approaches − tools for answering this question: Newton’s laws, Hamilton’s and
Maupertuis’ principle of least action. As we will see below, together they form a simple tool-
kit of mechanics in which the question being asked about any system determines directly
which tool should be used to predict the motion of that system.

Motion of a falling apple from different points of views

Newton’s laws of motion

We start with the well-known Newton’s laws of motion, which are already taught at high
schools. Firstly Newton says: “Give me the initial state of the apple, which means the initial
position and velocity of the apple”.
However giving the initial velocity and position means experimentally measuring two
nearby positions at very close instants. In this case the initial (and indeed any) velocity is
graphically nothing else than the slope of the position vs. time graph, or in the language of
spacetime physics the slope of the apple’s worldline.
Then Newton offers us his laws of motion and answers the questions: What happens
next with the apple? That is, what is the position of the apple at the next instant, if there is
Earth’s gravity or in general some force F (see fig.1)?

0
1 2
? F

velocity = slope of the graph

(or slope of apple’s worldline)

∆t ∆t
t
very small time intervals
Figure 1. Newton’s laws answer the question what is the position of the apple at the
next instant, if we know the initial velocity and position or in other words, we know
two very nearby positions of the apple.

If there were no acting force, then according to the first law of motion, the principle
of inertia, the apple would continue in motion at the same velocity, so graphically it would
follow a straight-line worldline (see fig.2). Instead Earth’s gravity (or generally some force F)
causes a component of motion in the direction of the applied force, as described by the second
law of motion, the momentum principle m∆v = F∆t. Competition between these two tenden-
cies results in the parallelogram of which the diagonal represents the worldline of the actual
motion (fig.2). This process of constructing worldlines (which is conceptually the same for
trajectories) is simple, repetitive and universally applicable, so it provides a first-rate founda-
tion for computer modeling. Since today’s computers are very fast there is really no need for
fancy algorithms in introductory physics teaching. To get a better approximation to the actual
motion we simply take smaller time steps.

Ogborn, Hanc, Taylor 9

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

x inertial motion

0
1 F

motion due to
force

2
∆t ∆t
t
very small time intervals
Figure 2. According to Newton’s laws the motion of the apple is produced by two “ef-
fects”: the apple’s inertial motion at constant velocity, if no forces act upon it and the
motion due to an acting force F. The net motion is then given by the diagonal of the par-
allelogram of the separate motions that would have occurred.

This numerical method appears in some classics physics texts notably in Chapter 9 in
Vol.1 of Feynman’s lectures on physics (1964). But it is also very effectively applied in the
modern introductory physics curriculum, e.g. in Modern Mechanics of Chabay and Sherwood
(2002) or in Unit N of Moore’s Six Ideas That Shaped Physics (2003).

Hamilton’s principle of least action

Now we apply a first action model to our falling apple based on energy concepts and Hamil-
ton’s least action principle. Hamilton tells us: “Give me both initial and final positions and
times of the apple,” called in spacetime physics events. If we specify the initial and final
events in advance, then Hamilton’s principle can successfully answer the following question:
What is the middle event for the apple? Or which worldline is followed by the falling apple
between the initial and final events, if the apple has potential energy U(x) (see fig.3)?

0
? apple has
potential
1 energy U(x)

2
∆t ∆t
t
very small time intervals
Figure 3. Hamilton’s principle answers the question what is the middle position or more
generally the middle event for the apple, if we know the apple’s initial and final events.

Ogborn, Hanc, Taylor 10

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

Now what special property does the actual worldline obey? The principle of least ac-
tion discovered by Hamilton says that the apple follows the worldline for which the average
kinetic energy minus the average potential energy is as little as possible or put more briefly
worldline has the least action, because Hamilton’s action S is defined as
⎛ difference bewteen ⎞
⎜ ⎟ ⎛ time duration ⎞
S ≡ ⎜ average kinetic and potential ⎟ ⋅ ⎜ of motion ⎟ or S = ( K − U ) ⋅ (t final − tinitial ) (1)
⎜ energy along the worldine ⎟ ⎝ ⎠
⎝ ⎠
Using integral calculus the definition (1) has the form
t final t final

S= ∫ ( K − U )dt = ∫ Ldt ,
tinitial tinitial
(2)

where the difference K − U is called the Lagrangian L, the quantity that appears in La-
grange’s equations of motion.
In the case of our falling apple (and also in general case for small ∆t) it is easy to cal-
culate all terms in the expression (1) for action S along any worldline 012 (fig. 3). The time
duration tfinal − tinitial equals 2∆t. The average kinetic energy 〈K〉 is given by (KA + KB)/2, the
average of kinetic energies for the first and second segment of the worldline, that is, by
(1/2)(mvA2/2 + mvB2/2), where vA = (x1 − x0)/∆t and vB = (x2 − x1)/∆t.
We now have to pay attention to the potential energy U(x). The shortness of ∆t allows
us to approximate U(x) by a linear function Cx in the region near point 1. (An additive con-
stant is not important, because it is always zero after an appropriate choice of a reference
point.) For the apple constant C is positive and equals mg. Generally we will consider it here
as some positive constant. From the viewpoint of the force concept used previously in New-
tonian analysis it represents a force F = − ∆U/∆x = − C∆x/∆x = − C, a force in the downward
direction. Then 〈U〉 equals (UA + UB)/2 = (1/2)[C(x0 + x1)/2 +C(x1 + x2)/2]. Since the events 0
and 2 are fixed and only position of the middle event 1 is variable, the apple’s action S must
be only a function of x1, in which case it is a quadratic function.
To find a worldline with the least action therefore means that we must vary and find a
position x1 which makes the action a minimum. There are two natural ways to do this. One is
the trial-and-error method, perfectly suited for a computer which can quickly calculate and
compare the action (1) for millions of worldlines. The detailed description of computer mod-
eling based on the so-called Euler variational method is described in our symposium contribu-
tions Action on Stage (see fig. 1, 2) and Use, Abuse, and Unjustified Neglect of the Action
principle (see fig. 1).
The second way to find the least action worldline is the use of mathematical methods.
According to Hamilton’s principle the action has to become larger, if we change the position
x1 of the middle event 1 of the actual worldline by any small displacement δx. Using only
high school algebra one can obtain the following expression for the corresponding change in
action:

δ S = S ( x1 + δ x) − S ( x1 ) = (m∆v − F∆t )δ x + 2
m
(δ x) 2 (3)
∆t
Mathematically equation (3) represents a simple quadratic function with respect to δx whose
graph is a parabola. The graphical method proves that the least-action worldline is identical
with the worldline predicted by Newton’s laws (fig. 4). The method gives students an intui-
tive and visual understanding of the meaning of the least action principle, as does the com-
puter modeling described earlier.

Ogborn, Hanc, Taylor 11

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

δS δS

δS > 0
δS > 0
δS > 0

δS < 0 δx δx

Figure 4. Both parts of the figure display changes in action with respect to dis-
placement δx. In the left part the linear term in eq. (3) is not zero, i.e. m∆v−F∆t ≠ 0.
The action demonstrates both negative and positive changes, so the chosen worldline
does not yield a minimal action. In the right part the condition that the linear term be
zero, m∆v−F∆t = 0, gives a required minimum.

Maupertuis’ principle of least action

Finally we will analyze the apple’s motion from the viewpoint of the second least action prin-
ciple called Maupertuis’ principle of least action. Maupertuis requires: “Consider a conserva-
tive system. Give initial and final position and total energy of the system.”
The total energy and its conservation (we again assume knowledge of the potential
energy), lead to knowing the apple’s initial speed, which implies two possibilities of motion −
with the upward or downward direction. But in the case of the falling apple we are interested
only in the downward motion. Then according to Maupertuis we are able to answer the ques-
tion: “What is the apple’s final event, if we know its initial and final positions (see fig. 5)?”

0
1
?
2
∆t t
Figure 5. The Maupertuis action principle can answer the question “What is the ap-
ple’s final event, if we know its initial and final positions?”

Since we consider a conservative system the actual motion of the apple must satisfy en-
ergy conservation. From Newtonian mechanics we know that it is the same motion as pre-
dicted by Newton’s laws. Everything seems to be good. So the natural question arises: where
is the action principle? But we now see that we did not realize that energy conservation alone
actually allows other worldlines, strange and unrealistic with respect to Newton’s laws. One
example is shown in fig. 6. How to recognize a motion as actual or unrealistic?

Ogborn, Hanc, Taylor 12

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

The criterion is just the Maupertuis action. It can be shown in a very similar way as in
Hanc et al. 2005 that a useful graphical tool in the case is the velocity vs. position diagram
called in mechanics the phase diagram, which says that for unrealistic worldlines the area
under the phase curve is always bigger than for the actual one. This geometric idea provides a
foundation for the definition of the second version of action, Maupertuis action W:
final
⎛ area under ⎞ ⎛ object' s ⎞
position

W ≡ ⎜⎜ ⎟⎟ ⋅ ⎜⎜
⎝ phase curve ⎠ ⎝ mass ⎠
⎟⎟ or W= ∫ mv ds
initial
(4)
position

0 0’ 1’
1

2 2’
∆t ∆t ∆t ∆t t
Figure 6. Both depicted worldlines 010 and 010’1’2’ satisfy energy conservation,
but only one describes the real motion.

Summarizing we can say that Maupertuis’ principle of least action tells the falling apple to
move so that the product of mass and area under the phase curve has the smallest possible
value (subject to energy conservation). So far as computer modeling is concerned, it is the
problem as before.

Acknowledgments. The work was partly supported from the grant of Slovakian cultural and educa-
tional agency (KEGA), Project No. 3/3005/05.

List of references
Chabay, R.W., Sherwood, B. A. (2002) Matter and Interactions: I. Modern Mechanics. John Wiley &
Sons, New York. <http://www4.ncsu.edu/~rwchabay/mi/>
Euler, L. (1744). Methodus Inveniendi Lineas Curvas Maximi Minimive Proprietate Gaudentes sive…
Laussanne and Geneva; see also chap. 2 in Goldstine H. H. (1980), A History of the Calculus
of Variations from the 17th Through the 19th Century. Springer-Verlag, New York.
Feynman, R.P., Leighton, R.B., Sands, M. (1964) The Feynman Lectures on Physics. Addison-Wesley,
Reading, MA, Vol. I
Goldstein, H., Poole, Ch., Safko, J. (2002). Classical Mechanics. (3rd ed.) Addison–Wesley, Reading,
MA
Hanc, J., Taylor, E.F., Tuleja, S. (2005). Variational mechanics in one and two dimensions. Am. J.
Phys. (73), 603-610 (and references in the article) <http://www.eftaylor.com/leastaction.html>
Hand, L.N., Finch, J. D. (1998). Analytical Mechanics. Cambridge University Press, Cambridge, UK.
Landau, L. D., Lifshitz, E. M. (1976). Mechanics (3rd ed.). Butterworth-Heinemann, Oxford, UK.
Moore, T. A. (2003) Six ideas that Shaped Physics: Unit N - The Laws of Physics are Universal.
McGraw Hill, New York. <http://www.physics.pomona.edu/sixideas/>
Marion, J.B., Thornton, S.T. (2003). Classical Dynamics of Particles and Systems (5th ed.). Thomson
Brooks/Cole.
Newton, I. (1687) Philosophiae Naturalis Principia Mathematica, London; see also chap. 3 in Good-
stein, D. L. et al (1999). Feynman's Lost Lecture …W.W. Norton, New York.

Ogborn, Hanc, Taylor 13

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

A First Introduction to Quantum Behavior

Jon Ogborn Institute of Education University of London

Abstract
The physics curriculum in England and Wales has a requirement to introduce
quantum phenomena to students in the first year of the two-year pre-university
physics course in schools. Usually this means discussing the photoelectric effect,
with a few words about “waves or particles”. In the innovative course “Advancing
Physics” we take a more fundamental approach, following Feynman’s “many paths”
formulation of quantum physics. The experiences of teaching this material for the
past five years are discussed, together with difficulties it has thrown up.

Introducing quantum behavior: a national requirement

In England and Wales, all “A-level” Physics courses – that is, high school physics courses
leading to University entrance are required to introduce the quantum behavior of photons and
electrons. For the innovative course Advancing Physics (Ogborn & Whitehouse 2000) we
decided to base our approach on Richard Feynman’s remarkable small book “QED”
(Feynman 1985), in which he describes in the simplest possible terms his ‘sum over histories’
or ‘many paths’ approach to quantum mechanics. Others have tried something similar (see
Hanc et al 2005).

The Feynman approach in essence

‘…Dick Feynman told me about his … version of quantum mechanics. “The electron
does anything it likes,” he said. “It just goes in any direction at any speed, forward or
backward in time, however it likes, and then you add up ….” I said to him, “You’re
crazy.” But he wasn’t.’
Freeman Dyson
Feynman’s big idea, starting with his 1942 doctoral thesis (Brown 2005), was to track all the
space-time paths available to photons or electrons in a given situation. Every possible path is
associated with a quantum amplitude. To find the probability of an event, you add up (taking
account of phase) the amplitudes for all possible paths leading to that event. The probability is
then given by the square of the amplitude, suitably normalized. The phase of the amplitude is
given by a result first noted by Dirac, namely that the number of rotations of the phase
‘arrow’ along a path is just the classical action S along the path, divided by the Planck
constant h.
This alternative way of setting up quantum mechanics led to enormous simplification
of calculations in quantum field theory, and is still today an essential tool for theoretical
physics. Our concern, however, is with the simplification and clarification it can bring to a
first introduction to the quantum world.

Six steps in introducing quantum behavior

We think of our teaching program in six steps:
1. random arrival
2. photon energy in lumps of size E = hf
3. superposition of amplitudes
4. what is quantum behavior?
5. quantum behavior can explain ....
6. electrons do it too.

Step 1 Random arrival

Perhaps the most important first experience of quantum behavior is to listen to a Geiger
counter detecting gamma photons: “click....click. click........click”. The key point is that the

Ogborn, Hanc, Taylor 14

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

gamma photons arrive at random. The time of arrival of a photon is not predictable: the only
thing we can know is the probability of arrival. Here is the first cornerstone of an
understanding of quantum behavior: only the probability of events is predictable.
A well known set of photographs (Figure 1) illustrates the idea beautifully.
A more careful treatment would want to show that the arrival of photons follows a
Poisson distribution, but this level of discussion is not available to us in this course, though
we are working on it (Ogborn, Collins & Brown 2003a,b).

Figure 1 The same picture taken with progressively more and more photons. The photons arrive
randomly, but with probabilities such that the picture gradually builds up

Step 2 Photon energy in lumps of size E = hf

The next step is to measure the energy and frequency to arrive at an estimate of the Planck
constant h. We suggest the use of a set of light-emitting diodes (LEDs), measuring the
wavelength of the light they emit, and the minimum potential difference needed for light just
to be emitted.
The point is to get across, as simply and directly as possible, that whatever happens in
between emission and arrival, light is always emitted and absorbed in discrete amounts E =
hf.. Now we have to think about “what happens in between”.

Step 3 Superposition of amplitudes

In Advancing Physics, the study of quantum behavior follows a study of the nature of light.
There, the classical story of the development of the wave picture is told, from Huygens and

Ogborn, Hanc, Taylor 15

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

Fermat through to Young and Fresnel. Thus students know that interference effects arise
when there are alternative paths between emission and absorption events. Now we marry up
this wave picture of superposition with the story of quantum behavior.
The big idea of quantum theory can be put very simply. It is just: “steal the wave
calculation but forget about the waves”. That is, associate with each path a phase ‘arrow’, and
add up the ‘arrows’ to get the resultant amplitude from all paths.
This is just Huygens’ wavelet principle. But something new and essential is added. It
is tht the rate of rotation of the quantum arrow along a path is given by f = E/h. In this way,
the results of the wave calculations of interference and diffraction patterns are all taken over.
The novelty is to start with the photon energy E as given and fundamental, and to use h, the
quantum of action, to translate it into a rate of rotation of phase.
Here we arrive at the heart of quantum behavior. In wave theory, the existence of a
phase is a consequence of the nature of wave motion. In quantum thinking, the existence of a
phase is rock-bottom fundamental. It is to be thought of as a given, not as a consequence.

Step 4 Quantum behavior

We can now describe the behavior of quantum objects. For each possible path between initial
and final discrete events, there is an ‘arrow’ (a phasor). For photon paths, the rate of rotation
of the arrow between start and finish is E/h. Add up the arrows for all the possible paths, tip to
tail, to get the resultant ‘arrow’ for the pair of events. The square of the resultant ‘arrow’ is
proportional to the probability of the pair of events.
Although initial and final events are localized, there is no reason to think of a photon
as localized “in between”. Lumpiness in energy does not imply lumpiness in space. To say as
Feynman does that the particles “go everywhere” is not to picture them as trying out all the
possibilities one at a time. Part of the essence of “being a particle”, namely its continuing
existence at a succession of places and times, has been taken away. Waves, of course, do “go
everywhere”, but part of their essence has gone too. Their energy can no longer be divided
into smaller and smaller amounts, without limit.
Not surprisingly, this account gives students difficulties. Like most of us, they
naturally try to form as concrete a picture as possible. So, thinking of photons as particles, and
imagining tracking each along a given path, they wrongly imagine them trying out all possible
paths one by one. They tend to think of the rotating phasor, not as associated with the path,
but as ‘riding on the back of a photon’ as it travels.
However, these difficulties are simply the difficulties of getting used to quantum
thinking. A possible merit of the approach is that it brings them out so clearly, by denying that
photons are some mixed-up approximation to waves and particles. Thus we present quantum
behavior as itself, not as like something else. Its essence is that all possibilities contribute,
superposing taking account of phase.

Step 5 Quantum behavior explains...

We conclude by offering students some comfort, by showing how quantum behavior explains
some familiar things, in particular, the laws of reflection and refraction.
Figure 2 shows a diagram from Feynman’s book, illustrating how the law of
reflection at a plane surface arises directly out of the quantum behavior of photons.

Ogborn, Hanc, Taylor 16

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

Figure 2 Reflection of photons at a plane mirror

Every possible path from source to detector counts. But only in the middle are nearby
paths closely similar in length, so that the ‘arrows’ associated with them are nearly in phase
and thus ‘line up’. Further out, the phase changes rapidly as the path changes, so that the
arrows from these paths ‘curl up’, giving a very small contribution to the final resultant arrow.
Only the paths very close to the path prescribed by the law of reflection make an important
contribution.
This explains Fermat’s principle of least time. The graph shows how the travel-time
varies as the path changes. At the minimum the time does not vary as the path changes. So the
phase associated with these paths does not change either. However, quantum thinking gives
us something more. By thinking about how much the phase changes between nearby paths we
can estimate the limits to the description given by geometrical optics.
The traditional “wave-particle duality” story makes it seem as if wave and particle
behavior are partial explanations of quantum behavior. Something more like the reverse is
true. Particle behavior, in particular all of classical Newtonian mechanics is explained by
quantum behavior (see Ogborn & Taylor 2005). Equally, the wave behavior of light follows
from the quantum behavior of photons, in the limit of low photon energy and many photons.
Quantum physics helps to explain why – and when – classical ideas work.

Step 6 Electrons do it too

Ogborn, Hanc, Taylor 17

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

The description of quantum behavior is appropriate not just for photons, but for all particles,
including electrons. For a free (non-relativistic) electron, the rate of rotation of the quantum
arrow is just K/h, where K is the kinetic energy. Demonstration apparatus available for school
laboratories makes it straightforward to show electron diffraction. Even better is to see
electrons arrive one by one, building up gradually into a two-slit interference pattern. This has
been achieved by a Hitachi team led by Akira Tonomura, who have produced a beautiful film
clip of this most fundamental experiment (Tonomura et al 1999).
The rate of rotation of the phase of an electron is given by L/h, where L is the
Lagrangian, and the total phase rotation is S/h, where S is the action along the path. We don’t
mention this to students in Advancing Physics, but we mention it here to show how the simple
approach can be extended in later work.

Does it work?
The only evaluation we have, so far, of the value of this approach is the experience of it being
taught in about 25% of UK schools teaching physics at this level. The teachers have available
to them an email network, to discuss whatever they please. Every year, when this topic is
taught, there is a flurry of discussion, about what exactly the ideas mean and about how to
respond to students’ questions.
These discussions show that teachers new to the approach are understandably
nervous, and need a good deal of re-assuring. It also shows that they, like their students, are
prone to giving the ideas an over-concrete interpretation, particularly in wanting to associate a
‘traveling phasor’ with a ‘traveling photon’, rather than associating a phasor with a possible
path.
Some ways of teaching the ideas lead very strongly to over-concrete interpretation.
One popular idea is to push a rotating wheel along a number of paths and to note the total
rotation for each path. This very effectively shows how, for example, the results of Figure 2
arise. But students inevitably want to know what role the wheel, and its diameter and speed,
play in the theory.
However, it remains true that there is, in evaluation of the whole Advancing Physics
course, no demand to remove or change this topic. Nor do students complain about the
examination questions set on it. We can say, therefore, that this rather radical innovation has
succeeded at least to the extent of having survived for six years so far, in a course taught on a
national scale in schools of widely differing kinds. And at least a proportion of both teachers
and students find it very interesting and thought-provoking.

References
Ogborn, J & Whitehouse, M. (Editors) (2000) Advancing Physics AS. Bristol: Institute of Physics
Publishing
Feynman, R. (1985) QED: The strange theory of light and matter. London: Penguin Books
Brown, L. M. (Editor) (2005) Feynman’s Thesis: A new approach to quantum theory. Singapore:
World Scientific.
Hanc, J., Tuleja, S., (2005) ‘The Feynman Quantum Mechanics with the help of Java applets and
physlets in Slovakia’, Proceedings of MPTL 10, Freie Universität Berlin, 5 - 7 October,
<pen.physik.uni-kl.de/w_jodl/MPTL/MPTL10/Proceedings.html>
Ogborn, J., Collins, S. & Brown, M. (2003a) ‘Randomness at the root of things 1: random walks’.
Physics Education volume 38, number 5, pages 391-397
Ogborn, J., Collins, S. & Brown, M. (2003b) ‘Randomness at the root of things 2: Poisson sequences’.
Physics Education volume 38, number 5, pages 398-405
Ogborn, J. & Taylor, E.F. (2005) ‘Quantum physics explains Newton’s laws of motion’. Physics
Education Volume 40, number 1, pages 26-34
Tonomura, A. et al (1999) ‘Demonstration of single-electron buildup of an interference pattern’.
American Journal of Physics Volume 57, No 2, pages 117-120 (movie clip downloadable
from www.pnas.org. Search for article under Author: Tonomura, A; Year 2005; Volume 102;
First Page 14952.)

Ogborn, Hanc, Taylor 18

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

Use, Abuse, and Unjustified Neglect of the Action Principle

Edwin F. Taylor, Massachusetts Institute of Technology, [email protected]

Abstract
Traditionally, differential equations dominated physics education; the action
principle was used primarily to derive differential equations such as Lagrange's
equations. Now the computer allows the action principle to be applied directly from
first principles, often bypassing analytic solutions entirely. The action principle can
illuminate and unify physics education and physics research from quantum field
theory to cosmology.

Computer solutions without solving equations

In classical mechanics it is not true that action along a particle worldline is always minimum.
Often it is a minimum, and it is never a maximum. But sometimes action can be a saddle
point, which means that, compared with the true worldline, some adjacent curves have a
greater value of the action, while some have a smaller value of the action. Seems complicated.
However, it is always true that the action along a sufficiently short worldline -- or a
sufficiently short segment of a worldline -- is always a minimum. Therefore we can correctly
say:

A particle follows a worldline such that the action along every small segment is a
minimum compared with that along every nearby segment with the same
endpoints.

Looking at this universal principle tells you immediately that computers are perfect
for applying action. Computers do increments beautifully. If a computer finds, by whatever
means, a worldline along every segment of which action is a minimum, then that is a true
worldline, one that a particle can follow.

Figure 1. Transfer trajectory between parking orbits around earth and moon. Dots on the trajectory are
the events of clock ticks that completely determine the worldline. The operator can add intermediate
clock ticks, drag individual ticks to minimize the action, or have the computer minimize action
automatically.

Figure 1 shows a frame of an interactive program by Slavomir Tuleja (2005). The

goal is to transfer a probe in an unpowered trajectory from a parking orbit around earth to a

Ogborn, Hanc, Taylor 19

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

parking orbit around the moon. We use Hamilton's action S because the fixed initial and final
points must be events; the probe must get to the location of the moon when the moon is there.
The dots along the trajectory are ticks on the clock carried by the probe, so the representation
completely determines the worldline: position vs time. The operator can drag the clock ticks
back and forth one at a time to minimize the total action, can add many intermediate ticks to
increase the accuracy, and can ask the computer to minimize the action automatically, which
it does in a split second. Notice that this analysis moves directly from the action principle to a
visualized solution with no intermediate mathematical analysis. And automatically generated
spread sheet data, the time and location of every clock tick, can be analyzed to any desired
level of detail.
A critic might object that the solution is now too easy; all the student does is push a
few buttons. On the contrary, the student has now been freed to investigate a hugely expanded
world of possible problems. For example, by changing the time lapse between initial and final
events, the student can search for the worldline that minimizes the total rocket impulse
required for the transfer from earth orbit to moon orbit. She can try the same for different
parking orbits around earth and moon. She can apply a similar program to transfers between
earth and mars. The software offers analysis of motion in other potentials as well.

The action principle for special and general relativity and the cosmos
When I was editor of the American Journal of Physics, I despaired about the twin paradox,
which seemed to poison the literature. Every month some engineer or retired doctor would
submit a paper disproving the obviously ridiculous predictions of the twin paradox, thus
invalidating special relativity.
In fact the twin paradox is central to the use of action to describe high speed motion:
In flat spacetime a true particle worldline yields the longest proper time (wristwatch time,
aging) between fixed initial and final events. A general expression for the action uses the
Lagrangian L, which for low speeds is the difference between the kinetic and potential
energies.

S≡ ∫ Ldt
A particle moving at any speed in an electromagnetic field has the Lagrangian

1 q
L= mv • v − v •A
2 c

where φ is the scalar potential and A is the vector potential. Two equations determine all
possible worldlines under electromagnetic influence in flat spacetime!
Now, general relativity is weird, but has the following gorgeous simplification: When
there are no singularities or gravitational waves, then at every event on a particle worldline
you can always find a local inertial frame in which special relativity holds. Since the
worldline is the sum of segments in these local frames, and since proper time is an invariant,
the same for all observers, therefore in general relativity the worldline of a particle between
fixed end events is the one with maximum aging along each small segment. We call this result
the principle of maximal aging. For low speeds and small curvature of spacetime, the
principle of maximal aging reduces to the principle of least action.
How do we find the value dτ of the proper time (aging) along a segment of a
worldline in curved spacetime? From the metric, the solution to the field equations. On the
left of the metric equation is the increment dτ of proper time (wristwatch time, aging)
between a nearby pair of events on the worldline. On the right side of the metric equation are
the corresponding increments of the (arbitrarily chosen) coordinates between that pair of
events.

Ogborn, Hanc, Taylor 20

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

Now, the metric is expressed in increments; manipulating the metric requires only
calculus. This means that if, instead of starting with the field equation, we start with the
metric solutions, we can introduce general relativity to sophomores using only calculus and
the principle of maximal aging. John Archibald Wheeler and I did this in our text Exploring
Black Holes, Introduction to General Relativity, Fig. 2 (Taylor & Wheeler 2000)

Figure 2. Text Exploring Black Holes by Edwin F. Taylor and John Archibald Wheeler that introduces
general relativity to sophomores using the metric and the principle of maximal aging.
The action principle not only tracks particles in curved spacetime. It is also at the root
of Einstein's field equations themselves. Hilbert derived the field equations from an action
principle, some say before Einstein completed his theory. Landau and Lifshitz do the same for
advanced physics students. Thus one can say that action describes the fundamental non-
quantum laws of the cosmos.

Quantum mechanics from the bottom

Recall Feynman's formulation of nonrelativistic quantum mechanics, exemplified by the
electron:

• The electron explores all possible worldlines between fixed end events that we choose.
• Along each trial worldline the total rotation of the quantum phasor is S/h = (Action)/h.
• At the end event, add up phasors for all worldlines to give the resultant quantum amplitude.
• The probability of detecting the electron at the final event is proportional to the squared
magnitude of the resultant quantum amplitude.

This is not a new idea! Sixty four years ago it was summarized in the introduction to
Feynman's 1942 Ph.D. thesis under John Wheeler: A generalization of quantum mechanics is
given in which the central mathematical concept is the analogue of the action in classical
mechanics . . . It is only required that some form of least action principle be available . . . if a

Ogborn, Hanc, Taylor 21

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

Lagrangian exists . . . the generalization reduces to the usual form of quantum mechanics. In
the classical limit, the quantum equations go over into the corresponding classical ones, with
the same action function. (Brown 2003)
Feynman was a co-recipient of a Nobel Prize for expanding these ideas to quantum
field theory, which one can take to be the most fundamental current theory of the very small
(a status that string theory has not yet achieved). And quantum field theories can be derived
from action. A quote from the Web: "Of all possible fields with a given boundary condition
the one that provides an extremum . . . of the action is The Solution."
Thus action carries us seamlessly from the smallest that we know to the largest, the
universe as a whole.

Variational principles: parents of action

. . . once the laws of physical theory are expressed as differential equations, the possibility of
their reduction to a variational principle is evident from purely mathematical reasoning . . .
(Yourgrau & Mandelstam 1968)

Traditionally differential equations have been the analytic tool of choice in physics, both for
education and research. As we have seen, the computer can apply variational principles
directly, often eliminating intermediate mathematical analysis. The use of variational
principles in current physics text is, at best, spotty. A few scattered examples:
Landau and Lifschitz develop the first two Maxwell Equations from experiment and
hand-waving. The last two equations they derive from a variational principle, which they call
action.
A standard method for finding the ground state wave function of an atom is to
minimize the electromagnetic energy.
The relaxation method is a powerful one for determining the electrostatic field
resulting from an array of fixed charges.
Textbooks often miss powerful and simplifying applications of variational principles.
For example, Van Baak (1999) replaces Kirchoff's circuit theorems by requirements that (1)
current is conserved, and (2) the rate of dissipation of energy is minimized. By using this
method, he says, the usual "extravagance of equations is wholly avoided."

Strategies for introducing action

Variational principles -- and action -- are tools, like differential equations, and not
fundamental physical principles (though they are related to conservation laws through
Noether's theorem; see e.g. Hanc et al. 2004). Still, they are unifying tools that the computer
can implement throughout physics education and research. Here are some notes on strategies
for introducing them to physics instruction:

1. Sneak bits of action and variational principles into secondary classes and introductory
undergraduate courses (see particular examples in Hanc et al. 2003 or Hanc & Taylor 2004).

2. Do NOT make action the primary tool in introductory physics; it is not concrete enough.
Introductory students need to feel the pushes and pulls of forces.

3. Use action and variational principles as unifying tools in the remainder of undergraduate
and graduate programs.

References
Hanc, J., Taylor E.F., (2004) ‘From Conservation of Energy to the Principle of Least Action: A Story
Line’, Am. J. Phys. 72 (4), 514 - 521
Hanc, J., Tuleja, S., Hancova, M. (2004). ‘Symmetries and Conservation Laws: Consequences of
Noether's Theorem’, Am. J. Phys. 72 (4), 428 - 435;

Ogborn, Hanc, Taylor 22

 The Girep conference 2006, Modeling in Physics and Physics Education, Universiteit van Amsterdam

Hanc, J., Tuleja, S., Hancova, M., (2003), ‘Simple derivation of Newtonian mechanics from the
principle of least action’, Am. J. Phys. 71 (4), 386-391.
Brown, L. M. (Editor) (2005) Feynman's Thesis -- A New Approach to Quantum Theory, Singapore:
World Scientific Publishing Co. page 1
Taylor, E. F. & Wheeler, J. A. (2000) Exploring Black Holes: Introduction to General Relativity, San
Francisco: Addison Wesley.
Tuleja, S. et al (2005). Use of a computer in advanced mechanics - Principle of least action,
Proceedings of MPTL 10, Freie Universität Berlin, 5 - 7 October, 2005, <pen.physik.uni-
kl.de/w_jodl/MPTL/MPTL10/Proceedings.html>
Van Baak, D. A. (1999) ‘Variational alternatives to Kirchoff's loop theorem in dc circuits’, Am. J.
Phys. 67, 36-44.
Yourgrau, W. and Mandelstam, S. (1968), Variational Principles in Dynamics and Quantum Theory,
New York: Dover Publications, page 175.

Software and background publications are available at www.eftaylor.com/leastaction.html;

Ogborn, Hanc, Taylor 23