Documente Academic
Documente Profesional
Documente Cultură
change in time. This means that these variables must be real numbers. If
The connection between operators and numerical values that we can mea-
^ =
O (3.1)
^ In this equation O
number is called an eigenvalue of O. ^ is known, and
QM 3: Eigenvalue problem May 30, 2005 2
chapter I will explain what an eigenvalue problem is and state some of the
the eigenvalue equation requires more time and more mathematics than is
examples.
x 2. The eigenvalue problem for p^x . Following the general denition (3.1),
a number (an eigenvalue of p^x ). The operator p^x was dened in Chapter 2,
h
@
= p (3.4)
i @x
QM 3: Eigenvalue problem May 30, 2005 3
Here p and are unknown and we need to nd them by solving this equation.
= exp[ikx]
h
@ ikx h
p^x eikx = e = (ik)eikx = h
keikx (3.5)
i @x i
If you compare Eq. 3.2 with Eq. 3.5, you see that (x) exp[ikx] is an
x 3. p^x has an innite number of eigenvalues. The eigenvalue equation Eq. 3.4
h
k is the corresponding eigenvalue for any number k, be it real or complex.
There are as many eigenvalues and eigenfunctions of p^x as there are complex
numbers.
^ (i.e. if O
Moreover, if is an eigenfunction of a linear operator O ^ = o,
^ = o and O
that O ^ ^
= O.
QM 3: Eigenvalue problem May 30, 2005 4
nomenon has more than one solution and that not all of them have a physical
(where the vector k has the Cartesian components fkx ; ky ; kz g and r has the
hkx and an eigenstate of p^y with the eigenvalue hky and an eigenstate of
value problem for angular momentum to give another simple example, which
QM 3: Eigenvalue problem May 30, 2005 5
Mechanics.
I will discuss later the physical interpretation of these operators. For now I
^ z is
denition (Eq. 3.1), the eigenvalue problem for L
^ z (x; y; z) = `z (x; y; z)
L (3.9)
this equation.
^ 2 (x; y; z) = (x; y; z)
L (3.11)
QM 3: Eigenvalue problem May 30, 2005 6
^ 2 from
You can calculate L
^ 2 = L
L ^ xL
^ x + L
^y L
^y + L
^zL
^z
^ x, L
and Eqs. 3.6{3.8,which give expressions for L ^ y , and L
^z .
^ 2 L
L ^ L
^ =
leads to
"
2 @ @2 @ @2 @2
h 2z + (x2 + y 2 ) 2 2y 2yz + (x2 + z 2 ) 2
@z @z @y @y@z @y
#
2 2 2
@ @ @ @
2x 2xz + 2xy + (y 2 + z 2 ) 2 = `
@x @x@z @x@y @x
3):
x + iy
1 (x; y; z) = p 2 (3.12)
x + y2 + z 2
x iy
2 (x; y; z) = p 2 (3.13)
x + y2 + z 2
and
z(x + iy)
3 (x; y; z) = (3.14)
x2+ y2 + z 2
QM 3: Eigenvalue problem May 30, 2005 7
^ z 1 (x; y; z) = h
L 1 (x; y; z)
^ z 2 (x; y; z) = h2 (x; y; z)
L
^ z 3 (x; y; z) = h
L 3 (x; y; z)
and
^ 2 (x; y; z) = h
L 2 (x; y; z) for = 1; 2; 3 (3.15)
^ z and L
This means that 1 , 2 , and 3 are eigenfunctions of both L ^ 2 . While
the two operators have common eigenfunctions they have dierent eigenval-
ues.
One can prove the following general theorem: If two linear operators A^
Conversely if two linear operators have the same eigenfunctions then they
commute.
(AB BA) = 0
for any function that is acceptable in Quantum Mechanics (see Section 3.2
As you'll see later, this rule is very important because if two dynamical
variables (e.g. p^x and V^ ) do not commute, their values cannot be measured
Principle, which says that position and momentum cannot be measured si-
multaneously. One can show that this happens because the coordinate and
Exercise 3.6 Show that if two operators have the same eigenvalues, then
^ B]
and [A; ^ = 0, then B
^ is also an eigenfunction of A,
^ with the eigenvalue
a.
^ z and L
and 3 (x; y; z) = z(x + iy)=r2 are eigenfunctions of both L ^ 2 where
p 2
r = x + y 2 + z 2 . Calculate the corresponding eigenvalues (see Work-
x + iy
1 (x; y; z) = p 2 (3.16)
x + y2 + z 2
QM 3: Eigenvalue problem May 30, 2005 9
x iy
2 (x; y; z) = p 2 (3.17)
x + y2 + z 2
^ 2 1 (x; y; z) = h
L 2 1 (x; y; z) (3.18)
and
^ 2 2 (x; y; z) = h2 2 (x; y; z)
L (3.19)
You can test that this is true by inserting the expressions Eqs. 3.16 and 3.17
Eqs. 3.18 and 3.19 indicate that 1 , 2 are eigenfunctions of the operator
Webster the adjective degenerate means \a: having declined (as in nature,
imagine why the person who named this property found it so oensive and
QM 3: Eigenvalue problem May 30, 2005 10
morally repulsive. These poor functions are stuck with the demeaning name,
and the prospect of naming them dierently to clear their reputation are very
dim.
You can also show easily (see Cell 3 of WorkBookQM.3) that 1 and 2
^ z 1 (x; y; z) = h
L 1 (x; y; z) (3.20)
and
^ z 2 (x; y; z) = h2 (x; y; z)
L (3.21)
^ z , corresponding
which indicate that 1 and 2 are also eigenfunctions of L
and h, respectively.
to the eigenvalues h
^ 2 and L
functions 1 (x; y; z) and 2 (x; y; z) are eigenvalues of both L ^ z . These
two operators are related: one gives the square of the length of the angular
momentum vector and the other gives the projection of the angular momen-
tum vector on the OZ axis. From classical mechanics we know that the length
the particle; the projection of the angular momentum vector on the OZ axis
tells us the orientation of the axis of rotation. Since the eigenstates 1 (x; y; z)
QM 3: Eigenvalue problem May 30, 2005 11
in which the particle has the same rotational energy. If the particle in the
states 1 (x; y; z) has the same rotational energy as a particle in the state
2 (x; y; z), why are these states dierent? The answer is given by Eqs. 3.20
and 3.21, which tell us that the two states correspond to particles having
dierent orientation.
^ are
This behavior is general. The degenerate states of an operator O
cases studied so far have revealed behavior that is general and must be re-
membered.
An eigenvalue problem does not have a unique solution. As you have seen
^ z , and L
in x3 and x5 the operators p^x , L ^ 2 all have more than one eigenfunc-
Two dierent operators can have common eigenfunctions, but with dif-
ferent eigenvalues. This means that given two linear operators A^ and B,
^ it
QM 3: Eigenvalue problem May 30, 2005 12
^ = a and B
is sometimes possible to nd a function such that A ^ = b.
rule is that two operators have the same eigenfunctions if and only if they
commute.
tions that correspond to the same eigenvalue. For example, exp[ikx] and
exp[ikx] are eigenfunctions of p^2x =2m; they both correspond to the eigenvalue
2 k 2 =2m (show that this is true). When this happens, the eigenvalue is called
h
degenerate.
Z +1 Z +1
hji dr(1) dr(N ) (r(1); : : : ; r(N ))(r(1); : : : ; r(N ))
1 1
(3.22)
which is called the scalar product of with . Here dr(k) dx(k) dy(k) dz(k)
and x(k); y(k); z(k) are the coordinates of particle k, located at r(k). (r(1); : : : ; r(N ))
Z +1 Z +1
hji dr(1) dr(N ) (r(1); : : : ; r(N ))(r(1); : : : ; r(N )) = 1
1 1
(3.23)
The condition in Eq. 3.23 follows from the physical interpretation of eigen-
is the probability that the particle labeled 1 is in the volume element dr(1),
centered around the point r(1), the particle labeled 2 is in the volume element
QM 3: Eigenvalue problem May 30, 2005 14
dr(2), centered around the point r(2), . . . . The integral in Eq. 3.23 is the
probability that the particles are located somewhere in space, and this must
be equal to one.
^
operator O:
^ =
O
^ = O(C)
O ^ ^ = C() = (C) =
= C(O)
Why is this of any help to us? Let us assume that we have found an
satisfy the normalization condition (3.23). This does not mean that we need
to throw this function away and look for one that is normalized. We can use
Z +1 Z +1
hji = dr(1) dr(N ) (r(1); : : : ; r(N ))(r(1); : : : ; r(N)) = M
1 1
(3.24)
for what follows; what counts is that h; i is not innite (it is easy to show
p (3.25)
M
^ and hji.
This statement is very easy to check by evaluating O
We can now replace the normalization condition (3.23) with the weaker
Eq. 3.25. You will see soon an example using this idea.
QM 3: Eigenvalue problem May 30, 2005 16
x 11. The fact that mathematics provides more solutions than are needed by
physics is not unusual. This is not a great nuisance, as long as we have a cri-
There are cases in which additional conditions are necessary. These are
not general, but follow from an examination of the physics of the specic
of this sort.
how they are used, and to indicate how they aect the eigenvalues and the
eigenfunctions.
bounded space. In other words, this particle is alone in the universe. Since
no force acts on the particle, its total energy is equal to its kinetic energy.
QM 3: Eigenvalue problem May 30, 2005 17
x 13. The eigenfunctions of the kinetic energy operator. The eigenvalue prob-
^
K(x) = K(x) (3.26)
it: exp[kx]; exp[kx]; sin(kx); cos(kx); exp[ikx]; exp[ikx]. I will verify this
2 k 2 kx
h
e = Kekx
2m
Cell 4, I show that all the functions listed above are eigenfunctions. Their
2 k 2 =2m. If two
are degenerate: they correspond to the same eigenvalue h
QM 3: Eigenvalue problem May 30, 2005 18
^ h2
k2 h2
k2 h2
k2 h2
k2 2 2 2 2
K= 2m 2m 2m 2m
k2m
h
k2mh
^
constrain k to be a real number) for the kinetic energy operator K.
^ having the
functions 1 and 2 are eigenfunctions of a linear operator O
same eigenvalue , then for any complex (or real) numbers A and B, the
^ with the
function (x) A1 (x) + B2 (x) is also an eigenfunction of O,
^ 1 = 1
O
and
^ 2 = 2 ;
O
and
then
^ =
O
^
You can easily see that this follows from the linearity of O:
^ = O(A
O ^ ^ ^
1 +B2 ) = AO1 +B O2 = A1 +B2 = (A1 +B2 ) =
QM 3: Eigenvalue problem May 30, 2005 19
use it for the eigenfunctions sin(kx) and cos(kx) of the kinetic energy oper-
Similarly
Exercise 3.9 Show that 1 and 2 dened by Eqs. 3.28 and 3.29 are not
ceptable. This means that we will test which one satises the condition in
Eq. 3.24.
QM 3: Eigenvalue problem May 30, 2005 20
quires that
Z +1 Z +1
hji = dx (x)(x) = dx sin2 (kx) = M < 1
1 1
It is easy to see that this integral is not nite. Indeed (see your calculus book
or use Mathematica)
Z +z
sin(2kz)
sin2 (kx)dx = z
z 2k
Z +1 Z +1
hji = dx exp[ikx] exp[ikx] = dx
1 1
away.
Z +1 Z +1
exp[2k1] exp[2k1]
hji = dx exp[kx] exp[kx] = e2kx dx = =1
1 1 2k 2k
If sin(kx), exp[ikx], and exp[kx] do not satisfy Eq. 3.24, neither will
x 16. Boundary conditions: the particle in a box. We have run into trouble
conned inside an apparatus. This means that the probability of nding the
satises
0xL
the particle is inside the machine. Since the particle cannot escape from the
machine, the probability (x)(x) that the particle is outside the \box" is
zero. This means that we must look for a solution of the eigenvalue equation
Eq. 3.24.
must be continuous. You can see that this is a reasonable condition: the
Because of the continuity requirement, the function (x) must also satisfy
(x = 0) = 0 (3.32)
and
(x = L) = 0 (3.33)
In the theory of dierential equations these are called the boundary con-
ditions, because they force the eigenfunction to take certain values at the
boundaries x = 0 and x = L.
x 17. Force the eigenfunctions to satisfy the boundary conditions. Let us see
and for x = L,
B=0
A sin(kL) = 0 (3.36)
kn = n; n = 0; 1; 2; : : : (3.37)
L
and (3.33), sin(kx) satised the eigenvalue equation and the eigenvalue was
impose the boundary conditions, k can no longer take arbitrary values; the
2 kn2
h 22 2
h
Kn = = n ; n = 0; 1; 2; : : : (3.38)
2m 2mL2
QM 3: Eigenvalue problem May 30, 2005 24
Note that we use the subscript n to label the eigenfunctions and the eigen-
values.
x 18. Quantization. As you will learn soon, if I measure the kinetic energy of
a particle I will obtain one of the eigenvalues of the kinetic energy operator.
For the example of the particle in an one-dimensional box this means that
h2 2 =2mL2 ) 12 or (
the result of the measurement can be ( h2 2 =2mL2 ) 22
or any other discrete value given by Eq. 3.38. No other value is allowed by
particle in a box. If the box size L or the mass m is large, the allowed values
of the kinetic energy are so close to each other that our instruments are not
good enough to resolve the dierence. In such a case it appears to us that the
x 19. The particle cannot have zero kinetic energy. Another curiosity is pro-
0
0 (x) = A sin x =0
L
We must discard this because j0 (x)j2 is identically zero; for this eigenfunc-
tion, the probability that the particle is in the box is zero and this conclusion
as impossible. The smallest kinetic energy the particle can have is K1 . If you
remember that in classical physics the kinetic energy is mv 2 =2, the fact that
the lowest kinetic energy is not zero means that the particle cannot have zero
velocity. This is very strange: the particle in a box can never be at rest! On
the other hand we have all seen that any large, classical object put in a box
does not move if we don't act on it with a force. It appears that Quantum
Mechanics is wrong if the object is large. How can we have two mechanics,
one for small object and one for the large ones? And what happens to the
objects in between, that are neither very large nor very small?
that our instruments cannot determine that it is dierent from zero. This is
Exercise 3.11 Calculate the value of K1 for m = 1 gram and L = 1 cm. Try
to explain why none of our instruments could measure such a small kinetic
energy. Assume that you can dene a velocity v by equating K1 with mv 2 =2.
Calculate how large this velocity is and explain why we could not measure
it.
Exercise 3.12 Show that the function 2 (x) dened by Eq. 3.29 leads to
the same eigenfunctions and eigenvalues as does the function 1 (x) dened
by Eq. 3.28.
Exercise 3.13 Assume that the particle is in a box whose boundaries are
located at
L L
x
2 2
Show that the eigenfunctions look dierent from those derived above (where
the position of the box was dened by 0 x L) but the eigenvalues are
the same. The two boxes are the same except that the second box is the
QM 3: Eigenvalue problem May 30, 2005 27
rst box translated a distance L=2 to the left. Should any of the physically
(3.32) and (3.33). (Hint: the boundary conditions can be satised only if k is
x 20. Imposing the boundary conditions removes the trouble we had with nor-
^ obtained by impos-
malization. It is easy to see that the eigenfunctions of K
Cell 6)
Z +1 Z L " #
2 2 2 nx L L sin(2n)
jn (x)j dx = A sin dx = A2
1 0 L 2 4n
The rst equality follows because n (x) = 0 for x < 0 or x > L and n (x) =
A sin(nx=L) for x 2 [0; L] (see Eq. 3.39). Because sin(2n) = 0, the integral
We can normalize the eigenfunction in Eq. 3.39 by using Eq. 3.25. This
QM 3: Eigenvalue problem May 30, 2005 28
2
2 kn2
h 2 n
h
Kn = = ; n = 1; 2; : : : (3.41)
2m 2m L
tor 1 (i.e. n (x) = n (x)) and therefore they correspond to the same
physical state. For this reason, we consider only positive values of the index
n.
We will come back in a future chapter to this example, to bring out the
physics contained in this result. Its purpose, so far, has been to illustrate
how one solves an eigenvalue problem and to point out how important the
always real numbers. This is important because | as you will learn in the
QM 3: Eigenvalue problem May 30, 2005 29
(3.42)
^
1 (r(1); : : : ; r(N )); 2 (r(1); : : : ; r(N)); : : : of eigenfunctions of an operator O
satisfying Eqs. 3.23 and 3.42 are said to form an orthonormal set. Unlike the
comes free: the operators representing dynamical variables are such that
orthogonal.
Let us check whether this is true for the eigenfunctions (Eq. 3.40) of the
q
kinetic energy operator. We have n (x) = 2=L sin(nx=L) for 0 x L
Z L
2 nx mx
= dx sin sin
0 L L L
sin((m n)) sin((m + n))
=
mn m+n
sin((mn)) sin((m+n))
Since m and n are positive integers and m 6= n, mn
= m+n
=0
^ = ;
O
For the operators that appear in Quantum Mechanics, this equation has a
^ represents
large number of solutions, labeled n (r(1); : : : ; r(N)) and n . If O
have physical meaning. However, we can impose the less stringent condition
Z +1
hji = dr(1) dr(N )n (r(1); : : : ; r(N))n (r(1); : : : ; r(N )) = M < 1
1
QM 3: Eigenvalue problem May 30, 2005 31
by
= p = q
M hji
same eigenvalue as .