CO?

NUTING OBSERVABLES AND SIMULTANEOUS EIGENFUNCTIONS

Consider two operators A and B which represent observables and arc

therefore Herinitian. If there is a ih such that

A* = a4’ and 13’4.’ = b4

then 4’ is a simultaneous eigenfunction of A and B, belonging to eigenvalues

a: .1 and b, respectively Hence,

BAiji B(a4) = aB = abiji

AB = A (b) = bA4’ = ba4

and
(AB - BA)iJ = 0.

Thus 4i is also an eigenfunction of [A,B], belonging to the eigenvalue zero.

If A and B commute, then [A,B]* = 0 holds for any ii’.

Now we can show that the eigenfwictions of two commuting operators cin
always be constructed in such a way that they are simultaneous eigenfunctions.

Proof: Suppose D is an eigenfunction of A, so

Ai = a4 and suppose A and B commute.

Then
BAi AB’IJ Bu aB

or
A(B4) = a(ThI’).

Thus, if we assume B4 t 0, the fi.mction B is an eigenfunction of A belong

ing to the same eigenvalue, a. Two cases to consider:

(1) Nondegenerate case. If there is only one state 1 corresponding to the

eigenvalue a, then B can differ from only by a constant factor, or
BiJ = b*.

Hence, in this case, [A,B]ip 0 is both a necessary and sufficient

condition for i to be a simultaneous eigenfuncticn of A and B.

(2) Degenerate case. Suppose, for example, that there are two states corres
ponding to the cigenvalue a, and 2. These must be linearly inde
pendent and may for convenience be assumed to be orthonormal. Any
linear combi

11) i
1
c + 1’
2
c

is also an eigenfunction of A but need not be an eigenfunction of B.

However, we can try to determine a particular linear combination(s)
COMMUTING OBSERVABLES - 2

which will be an elgenfunction of B, i such that

P
1
B(c + )
i
2
c = Ji
1
b(c + ).
2
c 2

Taking scalar products with and respectively, we obtain

2’
11
(
1
cB - b) + c
1
B
2 2 = 0 with 1
B 2 = 2
B
1
(p
) , etc.
p

c
2
B
1 1 + 22
(
2
c B - b) = 0

Nontrivial solution exists if

-b
1
B1 B
12

B
2 1 =b
2
B 2

If the two roots bi and 2

b of the scalar equation are distinct, the
corresponding eigenfunctions of B are linearly independent. Hence the
degeneracy has been resolved via the commuting operator B.

If the roots b
1 and b2 are the same, then if B is Hermitian, B 11 = B22
and 1
B 2 = *
21 = 0 are necessary and sufficient conditions. Accordingly any
B
Ci and cz can be used; i.e., IIJ1 and ii
2 are simultaneous, degenerate eigen
functions of B as well as A.

The analysis can be generalized immediately to cases of N-fold degen

eracy. If the secular equation in b has N distinct roots, the degeneracy
is completely resolved. If throots are not all diinct .,,it may be possi
ble to find a third operator &‘which commutes with ‘ and and which will
further resolve the remaining degeneracy, and so on. Ultimately, an
unambiguous description of the system can be obtained in terms of a
complete set of commuting observables,the eigçlues of which provide a
unique, nondegenerate specification of each state.
 NONCa.NIJTING OBSERVABLES AND THE UNCERTAINTY PRINCIPLE

The dynamical quantities corresponding to two Hermitian operators 4 und

= 0.
can be arbitrarily we]l defined in the same State if and only if
If [4] 0, then an eigenfunction of is not in general an eienfunction
4
of but rather a linear combination of several eigenfunctions. The results
of measurement of observable will then give a probability distribution
and cannot both
rather than a unicue ei2envalue. Thus the observables 4
be sharply defined simultaneously.

The tmcertaintv relation can be stated as

AB

where

2
(f) = (A _(A)
2

2
(B) = 2 _(B
B 2

and
define the widths or variance of the probability distributions in 4
and
=

where is a Hermitian onerator with expectation value (c) = KIcki) in the

normalized state of interest.

) uOUc *‘
Proof: ‘ Introduc&the operator
A A
=A+ (a+iB)B

in which and 8 are arbitrary real numbers. The scalar product DID) is
non-negative. Hence we find f
+ 2
(a2+B 2)KB + aKc’ +a(c)o

where
C’=AB+BA

is the anticommutator of A and B.

If Bi 1 0, the above expression can be rearranged as

(c))2 + ())2 - i(c)2

(c’)
(A2) + (2) +
()
NONCOr.tUTING OBSERVABLES - 2

This inequality holds for any values of a and ; in particular we can choose
a and ) vanish. Then the inequality be
such that the coefficients of (B
2
comes
(A2)(B2) (c’) +(c)2) .c)2

Thus, if
(A) = <B) = 0, we have

(B)
2
(A) . -(c) or MB c) Q.E.D.

and
If (A) and (B) are not zero, then apply this analysis to = -

= - for which again [,,‘J = i and KA = (B’) = 0.

The result of carrying this out and then expressing the inequality in
terms of and rather than ‘ and ‘ is

(B)
2
(A) -(c’ - 2
2A)(B ) +

so again
(M)(B) -(c) Q.E.D.

Comments:

1) Note that the most general form of the uncertainty relation is

2
(M)B) 2
x + 2
Y (see Bohm, Chap. 10)

where
= cc’ - 2
2cA)(B)) = -((AB + BA) - 2
2A)(B )

or - (A)(B - (p)) + (B - B)(A

V(A -

In probability theory x (A (A))(B (B))) = CAB) (A(B is called the

- -
-

if A and B
correlation coefficient of two random v riables A and B. Only
Also,
are independent (uncrrelated) does (ABs = (A(B) and X = 0.

2; BA))2..

or
I ((A - (A)) (B - (B)- (B - (B)) (A - (A))>j

This quantity, Y -I([A,B)I has no ana1o’ in ordinary probability theory.

2) Even when = i 0, it may happen that for some particular

occurs
State the expectation value (c = 0. mis is a rare case and if it
_______

NONCaIMUTING OBSERVABLES - 3

the more general form

tMB.(X
T +Y)
21/2

should be used.

3) The quantity MtB becomes a minimum when the equal sign holds in the
tmcertainty relation and the anticommutator term vanishes. Then. AB is as
small as the noncommutivity allows:

= 4 J(c)J.
This situation requires D4 = 0 or

A = -(a+IS)B4i.

As before, choose

and
K)
=
- 4 • Thus we have A
=

Example: Consider A B = x
= p,
Then =
(A) = (B) 0; M = 2 tB
(p)”

[x,p) = icso c =ic (c)

For the mimimum uncertainty state we have

iT
Px 2
2 (tx)

with i— , solution of the differential equation gives

p = X
2

- 2
4(Ax)
1
[2ir(tx) ]
and for this state tp xx =