Sei sulla pagina 1di 13

Why is theoretical physics hard to learn no matter

how much you already know?


Sergei Winitzki
December 26, 2004
Abstract
This essay is an outgrowth of my ruminations on the problem of learning and
teaching theoretical physics. I try to explain why it seems to be very hard to learn
a recently discovered branch of physics, while it is comparatively easy to learn
and to teach a long-established branch. Another issue is that, among students,
even well-known areas of physics have a reputation of being mysterious and in-
comprehensible. It seems that one source of the problem is the widespread atti-
tude that performing calculations and obtaining specic results is more important
than achieving a logical and conceptual understanding of the material. I explore
the interdependence of mathematics and physics which is inevitably involved in a
learning process and show that this principle of delayed understanding has detri-
mental consequences. Another problem is the mixing of mathematical methods
and their application in physics, which frequently confuses students. To illustrate
these problems, I quote some examples from widely used physics textbooks (the
reader who is not familiar with theoretical physics may skip the examples). I also
comment on the basic differences between the contemporary practices of teach-
ing mathematics and physics, in an attempt to propose a better way of presenting
physics to students.
What one needs to learn when studying theoretical physics
The nature of physical theories is such that a student of physics needs to learn four
conceptually separate steps.
1. A set of mathematical relations between some abstract quantities is postulated,
and other relations are derived from the postulated ones. (For example, in classi-
cal mechanics one postulates the Lagrange action principle and derives the equa-
tions of motion.) The student needs to understand the denition and properties
of the mathematical objects; for this, one needs a certain limited facility with
abstract mathematics.
2. The equations are solved to obtain numerical values of some of the quantities
(note that not all mathematical entities have numerical values; for example, a su-
persymmetric multiplet is an element of a noncommutative algebra rather than a
number-valued function). The student needs to learn some practical methods for
1
solving the relevant mathematical equations. Since comparatively very fewequa-
tions can be solved exactly, one also learns to derive asymptotic or approximate
solutions. One thus needs experience in solving computational mathematical
problems (but not in proving general theorems).
3. The numerical results are interpreted as approximate descriptions of certain phys-
ical experiments that can be in principle performed in real life. The student needs
to remember how the known physical phenomena (e.g. the velocity of a ow of
water, or the amount and color of light emitted by a heated body) are related to
the numerical values obtained by the calculations in step (2).
4. By denition, a physics problem is a description of an imaginary experiment
and a request to predict some of the results of hypothetical measurements with
adequate precision. When a newphysics problemis presented, the student should
be able to recall or construct the relevant mathematical structures that describe
the proposed experiment with sufcient accuracy. (This is of course followed by
solving the mathematical equations that follow from the general theory in that
particular case.) Building mathematical models of physical processes is a highly
nontrivial task which lies at the core of the physical science.
Students cannot achieve a good command of any branch of physics unless they mas-
ter the above steps. However, there is usually no time to learn all four steps when
studying more advanced areas of physics. Only the basic areas of physics (mechanics,
electromagnetism, thermodynamics) are allotted enough time to be studied thoroughly.
It is common to give the students just enough mathematical background so that
they can follow calculations. This means that the step (1) is being skipped: Instead of
seeing a consistent and logical description of mathematical structures used in a physical
theory, students face an array of strange-looking equations in which unfamiliar notation
is used. Students do not understand the mathematical properties of these equations and
instead resort to memorizing the various methods of their transformation and solution.
When the step (2) is being glossed over, students are presented merely with some
tricks that help to solve certain equations. Often there exist general mathematical meth-
ods of solving a class of problems, but limited time does not permit to rst introduce
these general methods and then show their application to a particular problem. Some-
times, a mathematical method is motivated by physical considerations or appeals to
physical intuition (this quantity is always very small, so we can disregard it in this
equation). This confuses students into thinking that they are learning a method that
applies only to a particular physical situation.
For example, such purely mathematical subjects as the perturbation theory of lin-
ear operators, the phase integral (WKB) approximation, or the theory of represen-
tations of the rotation group, are usually taught as part of quantum mechanics courses
in the form of particular methods of solution of certain quantum-mechanical prob-
lems. Variational calculus and perturbation theory for differential equations are taught
within classical mechanics, while vector analysis and Greens functions are lumped
with electrodynamics. This practice is perfectly admissible as long as students have
a clear understanding that, for instance, innite-dimensional linear algebra in Hilbert
2
spaces is not a theory of quantum-mechanical wave functions but a separate mathemat-
ical theory which is used in quantum mechanics as well as in other areas of physics
and mathematics. Likewise, it always helps students when perturbation theory is rst
briey presented as a certain purely mathematical tool yielding approximate solutions
to equations. After that it is straightforward to understand the application of perturba-
tive techniques to particular problems in physics.
(The steps 3 and 4 cannot be allotted insufcient time, unless very few problems
are given out to students. However, the skill of solving physics problems is incomplete
without the mastery of concepts.)
In addition to the lack of time, there are certain common teaching practices which
routinely impede the learning of all areas of physics.
The principle of delayed understanding
It seems to be a common approach in teaching physics to require that students be able
to perform certain calculations before they grasp the necessary mathematical and phys-
ical concepts. It is often stated that an understanding of the material will be achieved
gradually, only after many more calculations are performed and many more problems
are solved. The mathematical and logical foundations of such calculations are de-
emphasized in favor of teaching more tricks for solving various difcult cases. I call
this idea the principle of delayed understanding. Below I present a specic example
where this principle is applied, namely the calculation of causal Greens function using
Fourier transforms and contour integration as found in almost all textbooks. This ex-
ample demonstrates how the details of calculations are brought to the forefront while
at the same time the underlying logic is obscured.
Detailed example: Greens functions and contours
This is a somewhat long example which may be incomprehensible to non-scientists,
and yet will probably evoke a familiar shudder of delayed understanding in those who
have studied theoretical physics.
I quote (with some inessential abbreviations) from a widely used standard text-
book by J. D. Jackson, Classical electrodynamics (Wiley, 1962), section 6.6, Greens
function for the time-dependent wave equation:
1
... the Greens function ... will satisfy the equation,
_

2
x

1
c
2

2
t
2
_
G
_
x, t; x

, t

_
= 4
_
xx

_
t t

_
(6.55)
... Of course, the Greens function will have to satisfy certain boundary
conditions demanded by physical considerations.
1
In the latest (third) available edition of that textbook, the presentation of this issue was reworked and
made somewhat clearer, but in the mean time the original argument has found its way into many other
textbooks.
3
The basic Greens function satisfying (6.55) is a function only of the
differences in coordinates (xx

) and times (t t

). To nd G we con-
sider the Fourier transform of both sides of (6.55). ... The Fourier trans-
form g(k, ) is to be determined. ... it turns out that g(k, ) is
g(k, ) =
1
4
3
1
_
k
2


2
c
2
_. (6.59)
When g(k, ) is substituted into (6.58) and the integrations over k and
are begun, there appears a singularity in the integrand at k =
2
/c
2
.
Consequently solution (6.59) is meaningless without some rule as to how
to handle the singularities. The rule cannot come from the mathematics. It
must come from physical considerations ... [based on which] we demand
that our solution for G have the following properties:
(a) G = 0 everywhere for t <t

.
(b) G represent outgoing waves for t >t

.
If we think of the integration in (6.58), the singularities in g(k, ) occur
at = ck. We can do the integration as a Cauchy integral in the com-
plex plane. For t >t

the integral along the real axis in (6.58) is equiva-


lent to the contour integral around a path C closed in the lower half-plane,
since the contribution on the semicircle at innity vanishes exponentially.
On the other hand, for t

>t, the contour must be closed in the upper half-


plane, as shown in Fig. 6.4 by path C

. In order to make G vanish for t <t

we must imagine that the poles at = ck are displaced below the real
axis, as in Fig. 6.4. Then the integral over C for t > t

will give a nonva-


nishing contribution, while the integral over C

for t < t

will vanish. The


displacement of the poles can be accomplished mathematically by writing
(+i) in place of in (6.59). Then the Greens function is given by
G
_
x, t; x

, t

_
=
1
4
3

d
3
k

d
e
ikRi
k
2

1
c
2
(+i)
2
(6.60)
where R = xx

, =t t

, and is a positive innitesimal.


Students usually can neither understand this derivation nor identify the confusing points,
which is certainly one of the worst ways of being confused. The pedantic objection that
an innitesimal is undened can be easily answered: in equation (6.60), the param-
eter is actually an arbitrary real number such that > 0. The reference to an in-
nitesimal certainly makes deciphering the presentation more difcult (students have
heard lectures in standard calculus where innitesimals are completely absent). But the
main obstacle to understanding is the sacramental phrase the solution is meaningless
without a rule to handle the singularities ... this rule must come from physical consider-
ations. This phrase suggests that physical considerations provide a magic wand that
transforms mathematically meaningless solutions into correct results. (A normal train
of thought would be: if the solution is meaningless, we should discard it!) Likewise,
it is unclear why one is allowed to introduce contours in the complex plane at all; the
4
frequency is certainly a purely real quantity. However, most of the text is concerned
with the technical details of choosing and deforming the contours, and the students
attention is diverted from these issues. As a result, students can follow the individual
steps of the calculation but they do not understand why the calculation proceeds in this
way.
A straightforward presentation of this calculation can be given at cost of just a little
more mathematics. The discussion of boundary conditions should not be postponed
until the point when the calculation is stymied for an obscure reason (the singularity
in the integrand that appears when the integrations are begun). Instead one should
begin the discussion by specifying the boundary conditions for G that follow from
the relevant physical considerations. This makes the task of computing the function
G a well-dened and standard mathematical problem (to solve a differential equation
with boundary conditions) which can be tackled by a plethora of standard methods.
For the problem at hand, the Fourier transform G g(k, ) is the most convenient
method. Then it should be explained that g(k, ) is not a function but a distribution (a
generalized function) since G satises an equation involving a function and so must
be itself understood as a distribution. In the space of distributions, the solution of the
equation
g(k, )
_
k
2


2
c
2
_
=
1
4
3
is not the expression (6.59) but
g(k, ) =P
1
4
3
1
_
k
2


2
c
2
_ +A(k)(ck) +B(k)(+ck),
where A and B are arbitrary functions of k and P denotes the principal value inte-
gral. This expression is well-dened as a distribution. The boundary conditions for G
uniquely determine the unknown functions A and B. Now that the problem is math-
ematically well formulated, these functions can be computed straightforwardly (the
principal value integral is evaluated using residues). The trick with the choice of con-
tours (or equivalently with adding i) is merely a shortcut to determine the functions
A(k) and B(k) with the required boundary conditions without having to write the prin-
cipal value integral and the delta functions explicitly.
How we cope
I think that the principle of delayed understanding is one of the reasons that students
perceive physics a hard subject. It is of course true that one cannot master a branch of
physics without gaining experience in solving problems. However, even after tackling
a long and varied array of problems involving Greens functions, the gained result will
often be merely a facility in solving these problems. When students taught in this
way become practicing scientists, they usually forget that they never really understood
the reasoning behind the contour integration trick for Greens functions. Thus our
practicing scientists will be unable to confer understanding to a new generation of
students. The professors can only say, Solve more problems and understanding will
5
come. I would translate that sentence as: solve more problems and you will forget that
you still do not understand the material.
It seems that for most people the principle of delayed understanding does not
work: rather than being merely delayed, the understanding does not arrive at all. Un-
derstanding comes not as a result of performing many calculations, but as a result of
conceptual thinking. It is a professors task to help students gain such thinking, and a
researchers task to help other researchers in the conceptual understanding of new re-
sults. I think it is unnecessary and harmful to substitute an extensive problem-solving
experience for an explanation of the logical development of the material. Physics is
a logical and mathematical science, and in almost all cases a physics problem can be
solved by a completely straightforward application of a xed set of concepts, resulting
in sequence of logically connected steps.
Many students compensate for the lack of conceptual thinking by memorization
or by developing specic kinds of intuition for ill-explained facts. However, older re-
searchers frequently nd it impossible to adjust their thinking in this way, since their
intuition has already been shaped to guess the missing logical links in older theories.
Since new theories are often explained according to the principle of delayed under-
standing, those researchers nd that they cannot follow the new developments. (This
is of course not the only reason for the problem of generation gap in physics.)
Historical presentation
Another obstacle to understanding is the frequent practice of presentation of a physi-
cal theory in the way it was originally developed, with some of the wrong steps and
misconceptions that accompanied its discovery. The motivation for this method of
historical presentation is that since a completely logical and self-contained exposi-
tion of the theory is not available (or unknown to the teacher), the best one can do for
students is to expose to them all the steps of the historical development of the ideas,
hoping that the students will understand those ideas at least as well as the creators of
the theory. However, it seems to be much more efcient to present a physical theory in
a fully contemporary and logical formulation, made as straightforward as possible, and
stripped of the historical baggage. If a fully satisfactory formulation is not available,
the students should be given the best available formulation. There is nothing wrong in
telling the students that physics is not yet a nished science and that even some of its
foundational issues are still subjects of current research.
The meaning of physical meaning
The term physical meaning (or physical interpretation) is used in two ways. The
rst and more straightforward usage is in the spirit of the item (3) above: namely,
a mathematical quantity describes a certain physical measurement. For example, one
says In the formula E =
_
p
2
+m
2
, the physical meaning of the parameter mis the rest
mass of the particle. The second usage is indirect: one rst performs some calculations
and derives an expression such as
1
2
, and then one remarks, e.g., The parameter
1
2

has the physical meaning of effective mass of the particle. It is implied that a physicist
should feel the signicance of the quantity
1
2
.
6
The second usage of the term physical meaning may sound a bit philosophical
and contribute to the reputation of physics as a mysterious science whose adepts need to
develop a peculiar physical thinking. It is true that physics implies a certain peciliar
way of thinking, as outlined in the steps (1)-(4) above. However, physicists do not
(or need not) imbue the mathematical expression
1
2
with meaning. Physics is not
philosophy and is concerned not with meaning but only with approximate prediction
of experimental results. The reference to the physical meaning of a quantity X is
intended to help the reader choose the correct equation into which the quantity X will
be substituted after it is calculated. It is difcult to remember all the relevant equations
that are connected with the physical world; the association of a certain mathematical
expression with an intuitive concept such as mass or temperature helps one to
reinforce ones memory.
Therefore I would say that physical meaning is basically a mnemonic device
designed to facilitate the memorization of the numerous mathematical structures and
relations used in physics. When a physics book is being praised for a clear exposition
of physical meaning, it indicates the the authors have found good mnemonic rules to
help the readers memory. The physical meaning has more to do with human memory
than with physics, and students should be aware of that.
Why most seminar talks are incomprehensible
Everyone has had the experience of listening to completely incomprehensible talks.
Typically during such talks the speaker shows many equations and tries to explain
something while the audience has long lost track of the presentation. I think the reason
for this unfortunate phenomenon is that scientic communication in physics is often
about details of calculations which are only intelligible to those already doing similar
calculations, and not about concepts which would be more widely understandable. In
other words, the speaker thinks that it is more important to communicate technical
details of calculations, e.g. that the variable X in this equation can sometimes be
negative, than to explain the idea in general terms and let the audience pick up the
details later. Thus it becomes very hard to learn new physics from seminar talks.
It is difcult to explain complicated material, and it will not be always possible
for the speaker to adjust to an audience. I would like to propose a simple signaling
system: every person in the audience raises a small red ag when they feel that they
cannot follow the talk any more and would rather do something else. When the speaker
nds that all people in the audience have raised their red ags, it is time to stop the
presentation or to switch to another topic.
Correct results from wrong calculations
There exists in physics literature a pervasive practice of obtaining correct results through
mathematically unjustied or even plainly illogical calculations. (I am not referring to
unintended errors or misprints.) There is a peculiar notion of a physical level of rigor
that admits a wrong calculation if it supercially appears to be correct or is super-
cially analogous to other correct calculations. In physics literature, such calculations
7
are called formal and are considered acceptable as long as the results are plausi-
bly correct. I believe that this practice is another major obstacle to understanding for
students of physics at all levels.
There are two cases when wrong calculations are used. In the rst case, there exists
an equivalent but completely rigorous calculation which is longer or more complicated
but yields the same result. In the second case, there is no known rigorous method of
derivation (but the result is known to be valid and agrees with experimental data).
In the rst case, students should certainly be warned that the calculations are not
completely rigorous and that a more satisfactory approach exists but is not being pre-
sented for lack of time.
An example of the second case is the Feynman path integral which is openly admit-
ted to be an undened expression.
2
In quantum eld theory, the Feynman integral is
manipulated as if it were an ordinary convergent integral, even though a mathematical
denition of such an integral is not available in the general case. Presently no compu-
tational methods for calculating such path integrals, either analytically or numerically,
are known. Finding an adequate mathematical denition of Feynman path integrals is
a subject of current research in mathematical physics. However, physicists ignore this
difculty and proceed to perform manipulations with such undened integrals mostly
by analogy with ordinary integrals. As a result, one obtains certain other expressions,
e.g. various perturbative expansions in coupling constants, that can be transformed into
well-dened mathematical objects and computed numerically. In such cases, physicists
are usually aware that they are using calculations without proper mathematical justi-
cation. (Of course, the results of such calculations are accepted only insofar they agree
with experiments.)
There seems to be no escape from the practice of using the physical level of rigor
because there is no time to wait until all mathematical details are worked out. The only
help to a student is an honest admission that some mathematical details are skipped
because they are too cumbersome, or because a rigorous mathematical justication is
not yet known.
Why isnt there a good cosmology textbook?
It is widely recognized that one cannot nd usable self-contained textbooks to teach
recently developed branches of physics (such as advanced quantum eld theory, cos-
mology, superstring theory, and advanced solid state physics) even though all major
results in these branches have been established for some decades. There are many
books and monographs but they are all unreadable to most students. Correspondingly,
these areas of physics are not among the standard lecture courses at universities. On the
other hand, there is an abundance of excellent textbooks on such areas as classical me-
chanics, electrodynamics, thermodynamics, and special relativity, and these subjects
are taught already in secondary schools.
2
To quote Yu. I. Manin (from the preface to an edited volume Geometric ideas in physics, Mir publish-
ers, 1983, in Russian): Such an integral is not an object introduced by a mathematical denition, but rather
a hieroglyphic representation of a wide range of physical and mathematical ideas; this representation is to be
deciphered depending on the context.
8
It seems that a good textbook (such that an ordinarily prepared student can read and
understand most of the material without help) appears no less than 50 but no more than
100 years after the discovery of a new fundamental physical theory. For example, as
of 2000 there are excellent books on classical mechanics and electrodynamics (areas
that were essentially complete by 1880s), on special relativity (developed c. 1880-
1910) and on classical statistical physics (late 19th century), good books on general
relativity (1915-1920), adequate books on quantum mechanics (1910-1930), satisfac-
tory books on basic quantum eld theory and elementary particles (1930-1940), and
no adequate textbooks on advanced solid state physics (1950-1970), nonperturbative
quantum methods (1950-1970s), string theory (1970s-...), modern cosmology (1960-
1990), or quantum gravity (1960s-...).
These 50 to 100 years are spent trying to nd an adequate logical and mathematical
formulation of the new physical theory. I would like to contrast this situation with
mathematics where excellent self-contained textbooks often appear even in relatively
new branches of mathematics, as soon as the major new achievements are recognized.
Again, there seems to be no solution to this problem except to keep writing better
physics textbooks and discarding old ones.
It seems that during any given historical epoch, there exist physical theories based
on a mixture of standard calculations, inspired guesses, analogies with previously
known physics, and wrongly applied mathematical formulae. After the major results
of such theories have been accepted as correct by the physics community, it takes quite
some time to develop adequate mathematical structures and logical arguments that al-
low one to reformulate the physical theory in a completely straightforward, logical and
consistent manner. Until this is done, however, no complete understanding or effective
teaching of the physical theory is possible.
For instance, the development of calculus by Newton and Leibnitz resulted in its
wide use by c. 1700, but it was not until the beginning of the 19th century that a rigorous
foundation of calculus was developed by Cauchy and others. (Even in 1840s it was
still widely believed that any series converges if its terms tend to zero.) Nowadays
the notion of limit has been streamlined and simplied to such an extent that it can be
taught in secondary schools.
The discovery of Maxwells equations for electrodynamics in 1840s was followed
by almost a century of application until an adequate mathematical language (tensor
analysis, gauge eld theory) was invented that allowed to fully realize the potential
of that theory. After the theory was reformulated in the tensor language in 1930s, it
became possible to widely and effectively teach it to students.
Of course, not all new physics is based on unknown mathematics that needs to be
guessed and only gradually rigorized. Most physical results are derived by a straight-
forward application of known mathematics. Since mathematicians cannot guess which
areas of mathematics will gain currency in new physical theories, it is inevitable that
there is a certain delay between the discovery of new physics and a rigorous formula-
tion of the requisite new mathematics.
However, the 20th century has seen a development of a somewhat unhealthy rela-
tion between physics and mathematics. Leading physicists reinvented certain branches
of mathematics that were largely unknown to physicists although not novel to math-
ematicians of the day. For instance, it is documented that Heisenberg did not know
9
much linear algebra when he invented the matrix formulation of quantum mechan-
ics. Another example is the Wentzel-Kramers-Brillouin (WKB) approximation that
had to be rediscovered in 1930s in connection with the Schrdinger equation in quan-
tum mechanics, although it was known to mathematicians since at least 1830s as the
Liouville-Green phase integral approximation. Of course, the physicists were primar-
ily interested in getting their results rather than in achieving full mathematical rigor
or in the development of mathematics per se. Therefore, rather than adapting known
mathematics to their needs, physicists developed their own informal versions of those
mathematical theories and proceeded to teach them under the guise of physics, without
a fully logical development of the underlying mathematics. To this day, students are
impeded by the insufcient development of the mathematical and logical framework in
those branches of physics.
As an example, consider the presentation of the Wigner-Eckart theorem in the text-
book by J. J. Sakurai, Modern quantum mechanics (revised edition, Addison-Wesley,
1994), section 3.10:
We are going to prove one of the most important theorems in quantum
mechanics, the Wigner-Eckart theorem.
The Wigner-Eckart Theorem. The matrix elements of tensor operators
with respect to angular-momentum eigenstates satisfy

, j

T
(k)
q
|, jm =

jk; mq| jk; j

_
_

||T
(k)
||j
_

2 j +1
, (3.10.31)
where the double-bar matrix element is independent of m and m

, and q.
Before we present a proof of this theorem, let us look at its signi-
cance. First, we see that the matrix element is written as the product of
two factors. The rst factor is a Clebsch-Gordan coefcient for adding j
and k to get j

. It depends only on the geometry, that is, the way the system
is oriented with respect to the z-axis. There is no reference whatsoever to
the particular nature of the tensor operator. The second factor does depend
on the dynamics, for instance, may stand for the radial quantum number
and its evaluation may involve, for example, evaluation of radial integrals.
On the other hand, it is completely independent of the magnetic quantum
numbers m, m

, and q, which specify the orientation of the physical system.


To evaluate

, j

| T
(k)
q
|, jm with various combinations of m, m

, and
q, it is sufcient to know just one of them; all others can be related geo-
metrically because they are proportional to Clebsch-Gordan coefcients,
which are known. The common proportionality factor is
_

||T
(k)
||j
_
,
which makes no reference whatsoever to the geometric features.
The formulation of the theoremis made completely unclear by introducing the double-
bar matrix element. The notation ||A|| has not been used elsewhere in the book.
The boldface words indicate to students that a new concept is being introduced, but no
denition is presented for the mysterious double bar. The comments on the signi-
cance of the theorem do not clarify howone could compute the quantity
_

||T
(k)
||j
_
10
or what it means to omit the subscript q from T
(k)
q
. The text fails to explain that the
quantity
_

||T
(k)
||j
_
is not a matrix element of the operator T but simply an un-
known function about which we know that it depends only on (,

, j, j

, k) but not on
(m, m

, q). It would be better to denote that function e.g. by f (,

, j, j

, k; T). Then it
becomes self-evident that the function f will be determined if one compute sufciently
many different matrix elements in the LHS of equation (3.10.31) with a particular op-
erator T and various (,

, j, j

, k). In fact, the Wigner-Eckart theorem has little to do


with quantum mechanics per se but is a purely mathematical property of irreducible
representations of the rotation group SO(3). Basis vectors in the representation spaces,
labeled by the indices , j, etc., can be chosen so that matrix elements of certain op-
erators T are not arbitrary functions of (,

, j, j

, m, m

, k, q) but are always expressed


as products of the Clebsch-Gordan coefcient jk; mq| jk; j

and a suitable function


of (,

, j, j

, k).
The development of quantum mechanics in 1920s proceeded without much refer-
ence to linear algebra, representation theory, or functional analysis,the established
branches of mathematics that are in fact foundational to quantum mechanics. Physi-
cists invented new notations and rediscovered old theorems, but the mathematical side
of the theory was developed only to the extent necessary for performing calculations.
For this reason, the contemporary students of quantum mechanics are often presented
a mysterious new set of mathematical notations and computational tools which are not
clearly explained. Together with a set of genuinely difcult and counter-intuitive phys-
ical concepts, this creates an impression that quantum mechanics is unapproachably
complicated.
Why contemporary mathematics is also difcult to learn
Such long-studied areas of mathematics as representation theory, Lie group theory, al-
gebraic topology, classical functional analysis, and algebraic geometry have remained
unappreciated by the physics community until very recent times, despite being required
for the formulation of physical theories. These mathematical theories are not yet in-
cluded in the standard physics curriculum. Therefore a student of modern theoretical
physics needs to spend a fair amount of time studying mathematics on the side.
A student of mathematics faces somewhat different challenges. Roughly, the pro-
cess of learning mathematics can be split into two steps:
1. Understanding of the known mathematical theories and of the ways to dene
new objects. The student is required to develop an ability to quickly reason
about mathematical structures in an abstract way. (For example, after learning
the concept of tensor product of two vector spaces, one should be able to derive
the properties of the tensor product of three spaces without much difculty; once
one understands the construction of the tangent bundle to a manifold, one should
also understand how to dene the tangent bundle to the tangent bundle.)
2. Developing a certain facility in the manipulation of abstract concepts, applying
them to specic problems (i.e. to proving theorems) or relating them to other
concepts. The experience of performing abstract (i.e. not numerical) calculations
11
is needed to develop an intuition about the new mathematical concepts. Without
such intuition it is difcult for a human to envisage the possible applications of
those concepts and to create new mathematical theories.
The step 1 certainly presents a great challenge in itself because mathematics involves
great levels of abstraction where the ordinary human intuition cannot help. However,
there is another source of problems. The contemporary mathematical literature has
been heavily inuenced by the Burbakist tradition of presenting new material with-
out motivation (or with a perfunctory motivation such as this will allowus to formulate
further theorems), and with few examples. A mathematical theory presented in this
way appears as a list of new denitions and theorems about new mathematical objects.
A student reading such texts does not usually have an intuitive picture that could illu-
minate the new material, which therefore appears as a mass of notation that needs to
be deciphered and interpreted. Without a picture to keep in mind, it is hard to remem-
ber all details of new denitions (the human memory is associative rather than linear).
When one does not have even a rough intuitive idea of what one is doing, it becomes
almost impossible to manipulate the symbols correctly. Of course, a student of mathe-
matics must out of necessity develop the ability to build the intuition on the go, while
reading new mathematical texts.
This difculty can be mitigated if students are rst presented by a few motivational
examples based on already known material, and then shown how these examples are
generalized or abstracted in the new denition. This presentation should however be
mathematically rigorous and logical, showing to students that the new concepts are
necessarily (and not arbitrarily) dened in this or that particular fashion and that oth-
erwise they could not serve as tools for solving particular problems. An unmotivated
denition is a major obstacle to understanding (both in physics and in mathematics).
Those students who would rather memorize the rules of calculation than have a
mental map of the ideas will certainly be disappointed at the increased amount of
empty theorizing. Presenting a summary of such computational rules would certainly
be helpful for everyone.
The step 2 is hard because one cannot quickly develop intuition about a new ab-
stract concept. To overcome this difculty, one needs to spend considerable time ma-
nipulating the new concept and performing calculations. However, there is frequently
not enough time for solving problems and even not enough practice problems available,
especially in new areas of mathematics. It is crucial to give students enough motivation
so that they will study these mathematical areas more thoroughly on their own.
Finally, I would like to note that physicists require a different approach to teaching
mathematics than mathematicians, mainly because physicists are focused on computa-
tional issues rather than on creating and proving theorems about abstract mathematical
constructions. Therefore there seems to be no solution except to offer lecture courses
of mathematics specially taylored for physicists.
Conclusion
I have tried to describe my view of the obstacles that hamper the progress of students of
theoretical physics. The main problem seems to be the under-appreciation of the math-
12
ematical and logical foundations of physics. I have cited specic examples where an
insufcient mathematical basis is a direct cause of difculty for students, and suggested
ways to overcome these problems.
13

Potrebbero piacerti anche