Вы находитесь на странице: 1из 22


After matrix mechanics and wave mechanics approach of quantum mechanics Richard
Feynman came up with a new approach for dealing with quantum mechanical systems.
Feynman path integral approach determines how the wave function evolves with time by
using propagator. Idea behind this approach recognizing that the propagator is directly related
to the sum of all possible paths a particle can take between two points.


The Feynman path integral formulation depends upon finding the propagator for a given
system. The path integral for the propagator offers a way of obtaining the classical limit of
quantum mechanics. However, there is no way to distinguish between classical and the
quantum mechanics paths at initial stage. As the propagator is a function that is independent
of the initial conditions and describes how a system evolves in time. The propagator is as
follows in quantum mechanics.


we may write the wavefunction in the direrent eigenstates evaluated on the basis of
convenience. To obtain this result. How the wavefunction evolves in time, recalled by
following equation.:

when given some initial wavefunction at some time t0, chosen arbitrarily to be zero, the
propagator, U(t), determines that how the wavefunction evolves in time. After the analysis of
the above mentioned equation it lead us to the Schrdinger Equation. Now we will derive
equation (1) following Shankar. We will study Schrdinger equation for a state [as
formulated by Schrdinger, Heisenberg and Dirac

By evaluating above equation we can obtain the time independent Schrdinger equation as
By using Schrdinger equation we can expand the original wavefunction
out in terms of the eigenkets jni; j(t)i =Pan(t)jni wherean(t) = an(0) exp(iEt=h) are
obtained from the separation of variables and a0(0) = hnj(0)i. Now substituting back into
the expansion for j(t)i we obtain j(t)i =Xjnihnj(0)ieiEt=h=XjnihnjeiEt=hj(0)i=
U(t)j(0)I and hence we can obtain the expression for the propagator. We can now completely
describe the system since this propagator is independent of the initial position of the
wavefunction and can describe how the wavefunction evolves in time. This is entirely
equivalent to another form of the propagator which we write as U(t) = eiHt=h and we may
also consult to find a different derivation of (1). This derivation assumes discrete and non-
degenerate energy levels but if we have a continuous spectrum, such as the free particle, we
replace the sum with an integration over jni. If we have degeneracies we have to double
sum over the degeneracies.


If we are able to work out the propagator we can completely describe the system at any time.
There is a great difficulty in working out these propagators due to the complicated nature of
the summations. So we turn to the path integral method due to Feynman to evaluate these
propagators. The actual derivation of the propagator is too long to be derived in this short
introduction. So I am discussing the consequences of the path integral formulation in this
paper. The major result of the derivation is that between two points (x; t) ! (x0; t0)

U(x; t; x0; t0) = A Xall paths eiS[x(t)]=h (2)

where A is a normalisation factor and S[x(t)] is the classical action corresponding to each path
and is denoted by

S =Ldt

An alternate way to write this formulation that there are an number of paths is

U(x; t; x0; t0) =Z x0xeiS[x(t)]=hD[x(t)] (3)

where it is implied by the D[x(t)] that we sum over all the paths connecting (x; t); (x0; t0).
The consequence of this formula is that all paths are equally weighted. So we face a serious
question that by what mechanism is the classical path chosen over the other paths if they are
equally weighted and how the classical path is favoured in the classical limit? We may opt a
way to think of the above formula is for each path to represent a vector in the complex plane.
Assume, without loss of generality, that each vector is of unit length. Here the action
determines the phase of the vector. And if we move away from the classical path there is
destructive interference as all the different non-classical paths tend to cancel each other out.
And if we move towards the classical path the picture changes. If we talk about Hamilton's
principle, it is as follows :

S = ZLdt = 0

The above mentioned principle serves as the basis of analytical classical mechanics as both
the Hamiltonian and Lagrangian formulation follow from it. The variation along the classical
path is at an extremum, stated by above formula. So when we start moving towards the
particles classical path it will lead us towards constructive interference as the variation in the
path is zero and hence it will further lead us towards the classical path. It means that the
destructive interference effectively cancels the contribution of all paths that are not near the
classical path and constructive interference near the classical path allows us to regain the
classical path. All being equally weighted, the classical path is very important. Above the
term 'moving towards the particles classical path' was used, but what exactly do we mean by
'moving towards'?How far away from the particle's classical path must we go before we
encounter destructive interference? Effectively we may make an elementary guess that
S[xcl(t)]=h or otherwise if the action is h away from the classical path we have
destructive interference. The classical action of a macroscopic particle is on the order of 1 erg
sec 1027h[3] which is much greater then h. So in the above case we are well beyond the
h and hence the classical path completely dominates the sum. However for an electron,
which has a very tiny mass, has an approximate action of h=6 which is well within the h
needed and thus we cannot think of the electron moving along it's classical path since many
of the non-classical paths contribute to the summation.

According to Feynmans formulation we conclude that the propagator is equivalent to

summing over the action of all the possible paths between two points. But there arise a
question that if this give us the correct result as predicted by the Schrdinger equation so we
imagine a system evolving from (x ; t ) ! (x; t) ?, to show the equivalences. As we are
already know to the fact that the propagator tells us exactly how this system evolves and we
can approximate this as a straight line and constant velocity between the two points. Thus we
are only summing over a single path

(x; t) AZ 11deiS=h(x ; t )

where A is a normalisation constant and the action is determined by S = [1=2m(=)2 V

(x)] which is the action to the first order of epsilon. Now we have that

(x; t) AZ 11deim22h e ih V (x)(x ; t )

with the intention of taking the limit ! 0. Now after taking the said limit the first exponential
oscillates very rapidly unless the is smaller then the . If we pick a small neighbourhood
about such that it is smaller then we can expand out terms into Taylor Series about; and

(x; t) AZ 11deim22h [ @@t ihV + @@x +12


@x2 2 + ]

Now after expanding the terms, the is independent of and hence all the integrals can easily
be performed as they are all Gaussian integrals.

(x; t) A



V + ih




Now to make the terms in the order of vanish we setA = (m=2ih)1=2 and hence the
terms in front of the are






@x2 + V (4)

which is the time-dependent Schrdinger equation. Hence, the Feynman path integral
formulation is entirely equivalent to the Schrdinger wave equation.

Free Particle

Starting with the simplest example we will now do an actual computation of a path integral
the free particle. It is one of such cases where the calculation of path integral is quite
cumbersome compared to that of using common Quantum Machanics formalism. In the Path
Integral calculation we do the same step an infinite number of times as compared to ordinary
quantum mechanics.
As we calculated the propagator for one "path" is a product of these
Upath =
Uxj+1;xj =
ei(m_ x2j
=2V (xj ))=h

V (xj ); (3.13)
dx1 dxN1Upath =

N=2 Z NY1
dxk e
V (xj ) =

N=2 Z NY1
dxk e
L(x_ j ;xj ) (3.14)
where we have identified the Lagrangian L = mx_ 2
2 V (x).
In the limit N ! 1, ! 0 the sum goes over to an integral as
R j=0 L(x_ j ; xj) !
L(x_ j ; xj)dt = S[x(t)]. Hence, we may write
Dx(t)eiS[x(t)]=h; (3.15)
where the so called measure
Dx(t) =

N=2 R QN1
k=1 dxk. We have thus
derived equation (2.18).
When calculating an explicit path integral, it is implicit that we are to use equation (3.14) and
then take the limit N ! 1, ! 0.
With V (x) = 0, we get from equation (3.14):

There is a pattern to discover if one computes one integral at a time. Starting with the
integration over dy1 we get:

Completing the square with respect to y1 in the exponent yields

where we used that

where we used that

eax2dx =

a . Inserting this into the integral over dy2 and performing the integral in the same way as
above, we get

dy2 e

2 (y2y0)2+i(y3y2)2

dy2 e


2 [y2


3 (y0+2y3)y2+






dy2 e


2 [(y21

3 (y0+2y3))2+




9 (y0+2y3)2] =








9 (y2


3+4y0y3)] =


3 (y3y0)2

: (4.5)

So we may conclude that if we compute the N 1 integrals in this way, we will get


dy1dy2 dyN1 ei



(yj+1yj )2





N (yNy0)2







ei(yNy0)2=N =



ei(yNy0)2=N: (4.6)

Changing back from yi to xi we finally arrive at

U(xN; T; x0; 0) =




2Nh (xNx0)2



Important point to be noted that the phase of the total propagator here is equal to the phase of
the classical trajectory only, which in the case of a free particle corresponds to the action for a
straight line.
Note also that the limit N ! 1, ! 0 was very easy taking in this example, since N = T.


We take the general potential V = a + bx + cx2 + dx_ + exx_ , to demonstrate the Feynman
path integral formulation, of which the well known harmonic oscillator is a special case. I
will only be discussing the results here. It can be found that for the above potential the
propagator is

U(x; t; x0; t0) = eiScl=hA(t) (5)

The result says that the propagator depends only on the classical path and some function A(t)
which has to be determined through other means. Consider the case of the free particle. As
we have explained above it makes sense that the classical path plays an important role in the
evaluation of the path integral. It turns out that it completely dominates the summation i.e
calculations and we can easily work out the propagator for the free particle. Now consider the
harmonic oscillator. Suppose we wanted to write it down in position space in the form (1). To
obtain the propagator in position space we may derive all the wave functions and energy
levels and then plug them in U(x; t; x0; t0) =X jihjeiEt=h



An exp





An exp





exp[i(n + 1=2)!(t t0)]

where Hn(x) are the Hermite polynomials. This complicated sum, once evaluated can
completely describe the system. Evaluating the sum is no easy task and no attempt to do so
will be made here. However if we look at the propagator from Feynman's point of view, all
we need to do is obtain the classical action of the oscillator and we are within a function A(t)
from the sum. The classical action is easily derived as

Scl = m!2 sin !t[(x20+ x02) cos !t 2x0x0]

In this case the function A(t) can be evaluated using Fourier series [4] and works out to be

A(t) = m2hit1=2

Thus the propagator for the harmonic oscillator is

U = m




2h sin !t


+ x02) cos !t 2x0x0]


This expression for the propagator can be used to find the energy levels and the wave
functions and we are able to completely describe the harmonic oscillator without resorting to
solving a complicated PDE but instead the equation of motion for a harmonic
oscillator,which is much easier.

The Classical Limit

A nice way of obtaining the classical limit of quantum mechanics is using the path integral
expression for the propagator, although at first glance there is no way to distinguish between
the classical and the quantum mechanical paths. If we have a look at equation (3.15) again:



Since jei j = 1 for all , every path essentially gives the same contribution to the probability
amplitude thus the classical path is no more or no less important than any other path of the
particle. However, when integrating, the phases turn out to be crucial. The phase S[x(t)]=h
comes with the very small number h in the denominator, so even a quite small change in
S[x(t)] for different paths will cause great variations in the phase. When the amplitudes for
such paths are added, they tend to cancel each other and cause destructive interference. For
macroscopic (classical) particles, S is very large compared to h. So even a small (but still
macroscopic) change in the action for such a particle will cause great changes in the phase.
For such particles, many path amplitudes cancel each-other, making them un important for
the particles way of motion. We actually discovered that only one certain path followed the
classical trajectory. A question arises that why is this the only path in which amplitude is not
cancelled? The answer to the question is found in the classical theory. From classical
mechanics, we know the "principle of least action". The principle states that the action S[x(t)]

R t2

t1 L(x; x_ )dt has a minimum (or sometimes a maximum) for the classical path xcl(t), i.e.
S[xcl(t)] = 0. Thus, a small deviation from the classical path does only cause infinitesimal

Variations in S[x(t)] and therefore in the phase. The probability amplitudes for the paths
extremely close to the classical trajectory tends to add up and cause constructive interference!
For the paths deviating from the classical one, the change in the phase will be so large in
terms of h that the interference will be destructive instead. It follows that the classical
trajectory will contribute to the motion for macroscopic particles only.

Figure 5.1: Emergence of the classical limit. There is constructive interference between paths
close to the classical trajectory (thick line) whereas the trajectories further away tend to
cancel because of the rapidly changing sign of eiS[x(t)]=h.
For microscopic (quantum mechanical) particles on the other hand, the action is typically of
the order of h. Thus it takes a comparably large change in the action to achieve a significant
change in the phase S[x(t)]=h. We may also say that for very light particles even paths that
deviate much from the classical will be of importance. we may no longer talk about the
trajectory of the particle, but rather of a superposition of different paths, in this case. In the
limit h ! 0, even very small particles will obviously behave like classical particles, since their
action will then become large compared to h. The limit h ! 0 is also the limit where the
quantization of energy etc disappears, which is another effect of considering the classical
limit of quantum mechanics.

eax2dx =

a . Inserting this into the integral over dy2 and performing the integral in the same way as
above, we get


While this method looks much more complicated, the path integral formulation has a much
deeper application in quantum mechanics. One such example is that it allows us to formulate
driven quantum harmonic oscillators which play an important role in quantum
electrodynamics since we can represent the E-M eld as quantum driven harmonic oscillators

Finally the path integral method leads to perturbation calculations through Feynman
diagrams and plays an important role in relativistic quantum mechanics alongside the Dirac


R.P. Feynman, Rev. Mod. Phys 20, 367 (1948).

J.J Sakurai, Modern Quantum Mechanics (Pearson Edu-

cation, 1994)

R. Shankar, Principles of Quantum Mechanics (Plenum

Press, 1980)

R.P Feynman and A. R. Hibbs, Path Integrals and Quan-

tum Mechanics (McGraw-Hill Book Company, 1965)

M.G. Calkin, Lagrangian and Hamiltonian Mechanics

(World Scientic Publishing, 1998)

D. Griths, Introduction to Quantum Mechanics (Pearson

Education, 2005)