You are on page 1of 9

Quantum Mechanics for Mathematicians:

Two-state systems and spin 1/2


Peter Woit
Department of Mathematics, Columbia University
woit@math.columbia.edu
October 9, 2012
The simplest truly non-trivial quantum systems have state spaces that are
inherently two-complex dimensional, and have signicantly more structure than
the charge structure of the previous section, which could be analyzed by breaking
up the space of states into one-dimensional subspaces of given charge. Well
study these in this section, encountering for the rst time the implications of
working with representations of non-commutative groups. Since they give the
simplest non-trivial realization of quantum phenomena, such systems are the
fundamental objects of quantum information theory (the qubit) and the focus
of attempts to build a quantum computer (which would be built out of copies
of this sort of fundamental object). Many dierent possible two-state quantum
systems could potentially be used as the physical implementation of a qubit.
One of the simplest possibilities to take would be the idealized situation
of a single electron, somehow xed so that its spatial motion could be ignored,
leaving its quantum state described just by its so-called spin degree of freedom,
which takes values in 1 = C
2
. The term spin is supposed to call to mind
the angular momentum of an object spinning about about some axis, but such
classical physics has nothing to do with the qubit, it is a purely quantum system.
We will see later that quantum systems can be dened with spin quantum
number N/2, for
N = 0, 1, 2, 3, . . .
with dimension of the state space given by N +1. The N = 0 case is the trivial
case, and the case at hand is N = 1, the spin 1/2 case. In the limit N
one can make contact with classical notions of spinning objects and angular
momentum, but the spin 1/2 case is at the other limit, where the behavior is
purely quantum.
1
1 The Two-state quantum system
1.1 The Pauli matrices: observables of the two-state quan-
tum system
For a quantum system with two-dimensional state space, observables are self-
adjoint linear operators on C
2
. With respect to a chosen basis of C
2
, these are 2
by 2 complex matrices M satisfying the condition M = M

(M

is the conjugate
transpose matrix). Any such matrix will be a (real) linear combination of four
matrices:
M = c
0
1 + c
1

1
+ c
2

2
+ c
3

3
with c
a
R and the standard choice of the basis elements given by
1 =

1 0
0 1

,
1
=

0 1
1 0

,
2
=

0 i
i 0

,
3
=

1 0
0 1

The
a
are called the Pauli matrices and are a pretty universal choice of basis
in this subject. They reect an arbitary convention, in particular the convention
of diagonalizing the basis element in the 3 or z-direction of R
3
. One could just
as easily decided to choose a basis diagonal in some other direction.
Recall that the basic principle of how measurements are supposed to work
in quantum theory says that the only states that have well-dened values for
these four observables are the eigenvectors for these matrices. The rst matrix
gives a trivial observable, whereas the last one,
3
, has the two eigenvectors

1
0

1
0

and

0
1

0
1

with eigenvalues +1 and 1. In quantum information theory, where this is


the qubit system, these two eigenstates are labeled [0 and [1 because of the
analogy with a classical bit of information. Later on when we get to the theory of
spin, we will see that
1
2

3
is the observable corresponding to the SO(2) = U(1)
symmetry group of rotations about the third spatial axis, and the eigenvalues

1
2
, +
1
2
of this operator will be used to label the two eigenstates
[ +
1
2
=

1
0

and [
1
2
=

0
1

A crucial and easy to check fact is that [+


1
2
and [
1
2
are NOT eigenvectors
for the operators
1
and
2
. An exercise in the rst problem set is to compute
the eigenvectors and eigenvalues for these operators. It can be checked that no
pair of the three
a
commute, so one cannot nd vectors that are simultaneous
eigenvectors for more than one
a
. This non-commutativity of the operators
is responsible for the characteristic classically paradoxical property of quantum
2
observables: one can nd states with a well dened number for the measured
value of one observable
a
, but such states will not have a well-dened num-
ber for the measured value of the other two non-commuting observables. The
physics language for this phenomenon is that if we prepare states with a well-
dened spin component in the a-direction, the two other components of the
spin cant be assigned a numerical value in such a state. Any attempt to
prepare states that simultaneously have specic chosen numerical values for the
3 observables corresponding to the
a
is doomed. So is any attempt to si-
multaneously measure such values: if you thought you knew the value for one
observable, measuring another would ensure that this rst measurement was no
longer valid.
This non-classical behavior of observables takes some getting used to, and
there are many subtleties in the theory of measurement, but the basic phe-
nomenon is captured accurately by the very simple mathematics of the eigenvec-
tors and eigenvalues of the three 2 by 2 matrices
a
. Many quantum mechanics
textbooks (a good example is Volume III of the Feynman lectures[1]) contain
an extensive discussion of the physical implications of this specic example.
The choice we have made for the
a
corresponds to a choice of basis such
that the basis vectors are eigenvectors of
3
.
1
and
2
take these basis vectors
to non-trivial linear combinations of basis vectors. It turns out that there are
two specic linear combinations of
1
and
2
that do something very simple to
the basis vectors, since
(
1
+ i
2
) =

0 2
0 0

and (
1
i
2
) =

0 0
2 0

we have
(
1
+ i
2
)

0
1

= 2

1
0

(
1
+ i
2
)

1
0

0
0

and
(
1
i
2
)

1
0

= 2

0
1

(
1
i
2
)

0
1

0
0

(
1
+ i
2
) is called a raising operator: on eigenvectors of
3
it either
increases the eigenvalue by 2, or annihilates the vector. (
1
i
2
) is called
a lowering operator: on eigenvectors of
3
it either decreases the eigenvalue
by 2, or annihilates the vector. Note that these linear combinations are not
self-adjoint, (
1
+ i
2
) is the adjoint of (
1
i
2
) and vice-versa.
Well recall soon in our review of linear algebra that an arbitrary self-adjoint
2 by 2 complex matrix can be diagonalized by a unitary change of basis, i.e.
there is a unitary matrix U such that
UMU
1
=

1
0
0
2

where
1
,
2
R are the eigenvalues of the matrix. If c
0
= 0, so we just have a
linear combination of the three Pauli matrices, we must have
2
=
1
, since
the Pauli matrices all have trace zero.
3
1.2 Exponentials of Pauli matrices: unitary transforma-
tions of the two-state system
Exponentiating the identity matrix gives the diagonal unitary matrix
e
i1
=

e
i
0
0 e
i

This matrix commutes with any other unitary 2 by 2 matrix and just acts on our
state space C
2
by multiplication by a phase, as studied in the previous section.
If we exponentiate Pauli matrices, it turns out that since all the
a
satisfy

2
= 1, their exponentials also take a simple form:
e
ia
= 1 + i
a
+
1
2
(i)
2

2
a
+
1
3!
(i)
3

3
a
+
= 1 + i
a

1
2

2
1 i
1
3!

a
+
= (1
1
2

2
+ )1 + i(
1
3!

3
+ )
a
= (cos )1 + i
a
(sin )
As goes from = 0 to = 2, this exponential traces out a circle in
the space of unitary 2 by 2 matrices, starting and ending at the unit matrix.
This circle is a group, isomorphic to U(1). So, we have found three dierent,
non-commuting, U(1) subgroups inside the unitary 2 by 2 matrices.
To exponentiate linear combinations of Pauli matrices, one can check that
these matrices satisfy the following relations, useful in general for doing calcu-
lations with them instead of multiplying out explicitly the 2 by 2 matrices:

a
,
b
=
a

b
+
b

a
= 2
ab
1
Here , is the anti-commutator. This relation says that all
a
satisfy
2
a
= 1
and distinct
a
anti-commute (e.g.
a

b
=
b

a
for a ,= b).
Notice that the anti-commutation relations imply that, if we take a vector
v = (v
1
, v
2
, v
3
) R
3
and dene a 2 by 2 matrix by
v = v
1

1
+ v
2

2
+ v
3

3
=

v
3
v
1
iv
2
v
1
+ iv
2
v
3

and start taking powers of this matrix we nd


(v )
2
= (v
2
1
+ v
2
2
+ v
2
3
)1 = [v[
2
1
If v is a unit vector, we have
(v )
n
=

1 n even
(v ) n odd
4
We can use this to compute the exponential of the matrix iv ( R, v
a unit vector) as follows, using the Taylor series formula for the exponential:
e
iv
= 1 + iv +
(iv )
2
2!
+
(iv )
3
3!
+
= (1

2
2
+ )1 + i(

3
3!
+ )v
= (cos )1 + i(sin )v
Notice that you can easily compute the inverse of this matrix:
(e
iv
)
1
= (cos )1 i(sin )v
since
((cos )1 + i(sin )v )((cos )1 i(sin )v ) = (cos
2
+ sin
2
)1 = 1
Well review linear algebra and the notion of a unitary matrix in one of the next
classes, but one form of the condition for a matrix M to be unitary is
M

= M
1
so the self-adjointness of the
a
implies unitarity of e
iv
since
(e
iv
)

= ((cos )1 + i(sin )v )

= ((cos )1 i(sin )v

)
= ((cos )1 i(sin )v )
= (e
iv
)
1
You can also easily compute the determinant of e
iv
, nding
det(e
iv
) = det((cos )1 + i(sin )v )
= det

cos + i sin v
3
i sin (v
1
iv
2
)
i sin (v
1
+ iv
2
) cos i sin v
3

= cos
2
+ sin
2
(v
2
1
+ v
2
2
+ v
2
3
)
= 1
So, we see that by exponentiating i times linear combinations of the self-
adjoint Pauli matrices (which all have trace zero), we get unitary matrices of
determinant one. These are invertible, and form the group named SU(2), the
group of unitary 2 by 2 matrices of determinant one. If we exponentiated not
just iv , but i(1+v ) for some real constant (which do not have trace
zero unless = 0), we would get a unitary matrix with determinant e
i2
. The
group of unitary 2 by 2 matrices with arbitrary determinant is called U(2). In
our review of linear algebra to come we will encounter the groups SU(n) and
U(n), groups of unitary n by n complex matrices.
5
To get some more insight into the structure of the group SU(2), consider an
arbitrary 2 by 2 complex matrix

Unitary implies that the rows are orthonormal. One can see this explicitly from
the condition that the matrix times its conjugate-transpose is the identity

1 0
0 1

Orthogonality of the two rows gives the relation


+ = 0 = =

The condition that the rst row has length one gives
+ = [[
2
+[[
2
= 1
Using these two relations and computing the determinant (which has to be 1)
gives
=

( + ) =

= 1
so one must have
= , =
and an SU(2) matrix will have the form

where (, ) C
2
and
[[
2
+[[
2
= 1
So the elements of SU(2) are parametrized by two complex numbers, with
the sum of their length-squareds equal to one. Identifying C
2
= R
4
, these are
just vectors of length one in R
4
. Just as U(1) could be identied as a space with
the unit circle S
1
in C = R
2
, SU(2) can be identied with the three-sphere S
3
in R
4
.
2 Commutation relations for Pauli matrices
The anti-commutation relations for Pauli matrices show that when we multiply a
linear combinations of them by i and exponentiate, we get elements of SU(2),
6
in particular a U(1) subgroup of SU(2). Another important set of relations
satised by Pauli matrices are their commutation relations:
[
a
,
b
] =
a

b

b

a
= 2i
3

c=1

abc

c
where
abc
satises
123
= 1, is antisymmetric under permutation of two of its
subscripts, and vanishes if two of the subscripts take the same value. More
explicitly, this says:
[
1
,
2
] = 2i
3
, [
2
,
3
] = 2i
1
, [
3
,
1
] = 2i
2
One can easily check these relations by explicitly computing with the matrices.
Putting together the anticommutation and commutation relations, one gets a
formula for the product of two Pauli matrices:

b
=
ab
1 + i
3

c=1

abc

c
While physicists prefer to work with self-adjoint Pauli matrices and their
real eigenvalues, one can work instead with the following skew-adjoint matrices
s
a
= i

a
2
which satisfy the slightly simpler commutation relations
[s
a
, s
b
] =
3

c=1

abc
s
c
or more explicitly
[s
1
, s
2
] = s
3
, [s
2
, s
3
] = s
1
, [s
3
, s
1
] = s
2
If these commutators were zero, the SU(2) elements one gets by exponentiat-
ing linear combinations of the s
a
would be commuting group elements. The
non-triviality of the commutators reects the non-commutativity of the group.
Group elements U SU(2) near the identity satisfy
U 1 +
1
s
1
+
2
s
2
+
3
s
2
for
a
small and real, just as group elements z U(1) near the identity satisfy
z 1 + i
One can think of the s
a
and their commutation relations as an innites-
imal version of the full group and its group multiplication law, valid near
the identity. In terms of the geometry of manifolds, recall that SU(2) is
the space S
3
. The s
a
give a basis of the tangent space R
3
to the identity
7
of SU(2), just as i gives a basis of the tanget space to the identity of U(1).
3 Dynamics of a two-state system
Recall that the time dependence of states in quantum mechanics is given by the
Schrodinger equation
d
dt
[(t) =
i

H[(t)
where H is a particular self-adjoint linear operator on 1, the Hamiltonian op-
erator. The most general such operator on C
2
will be given by
H = h
0
1 + h
1

1
+ h
2

2
+ h
3

3
for four real parameters h
0
, h
1
, h
2
, h
3
. The solution to the Schrodinger equation
is just given by exponentiation:
[(t) = U(t)[(0)
where
U(t) = e

itH

The h
0
1 term in H just contributes an overall phase factor e

ih
0
t

, with the
remaining factor of U(t) an element of the group SU(2) rather than the larger
group U(2) of all 2 by 2 unitaries.
Using our earlier equation
e
iv
= (cos )1 + i(sin )v
8
valid for a unit vector v, our U(t) is given by taking h = (h
1
, h
2
, h
3
), v =
h
|h|
and =
t|h|

, so we nd
U(t) =e

ih
0
t

((cos(
t[h[

))1 + i(sin(
t[h[

))
h
1

1
+ h
2

2
+ h
3

3
[h[
)
=e

ih
0
t

((cos(
t[h[

))1 i(sin(
t[h[

))
h
1

1
+ h
2

2
+ h
3

3
[h[
)
=

ih
0
t

(cos(
t|h|

) i
h3
|h|
sin(
t|h|

)) i sin(
t|h|

)
h1ih2
|h|
i sin(
t|h|

)
h1+ih2
|h|
e

ih
0
t

(cos(
t|h|

) + i
h3
|h|
sin(
t|h|

))

In the special case h = (0, 0, h


3
) we have
U(t) =

e
i
t(h
0
+h
3
)

0
0 e
i
t(h
0
h
3
)

so if our initial state is


[(0) = [ +
1
2
+ [
1
2

for , C, at later times the state will be


[(t) = e
i
t(h
0
+h
3
)

[ +
1
2
+ e
i
t(h
0
h
3
)

[
1
2

In this special case, one can see that the eigenvalues of the Hamiltonian are
h
0
h
3
. A bit later on in this class we will study the behavior of the observables
of this system as a function of time, for this Hamiltonian.
In the physical realization of this system by a spin 1/2 particle (ignoring its
spatial motion), the Hamiltonian is given by
H =
ge
4mc
(B
1

1
+ B
2

2
+ B
3

3
)
where the B
a
are the components of the magnetic eld, and the physical con-
stants are the gyromagnetic ratio (g), the electric charge (e), the mass (m) and
the speed of light (c), so we have solved the problem of the time evolution of
such a system, setting h
a
=
ge
4mc
B
a
. For magnetic elds of size [B[ in the 3-
direction, we see that the two dierent states with well-dened energy ([ +
1
2

and [
1
2
) will have an energy dierence between them of
ge
mc
[B[
This is known as the Zeeman eect and is readily visible in the spectra of atoms
subjected to a magnetic eld.
References
[1] Feynman, R. Feynman Lectures on Physics, Volume 3.
9

You might also like