Académique Documents
Professionnel Documents
Culture Documents
p. 1/31
Dirac notation
State vectors in quantum mechanics are written in Dirac
notation. The basic object is the ket-vector |i, which
(given a particular basis) can be represented as a
column vector. The adjoint of a ket-vector is a bra-vector
h|, represented as a row vector.
) = |i .
|i = ... , h| = (1 N
N
N
X
|j |2 = 1.
j=1
p. 2/31
..
|i = . , |i =
N
1
.. ,
.
N
1
X
) .. =
h|i = (1 N
j j .
j
N
1
1 1 1 N
..
...
|ih| = ... (1 N
) = ...
.
.
N
N 1
N N
p. 4/31
Linear operators
transforms states to states such that
A linear operator O
O(a|i
+ b|i) = aO|i
+ bO|i
a11 a1N
..
.. [a ].
.
=
.
O
.
.
ij
.
aN 1
aN N
p. 5/31
Expectation values
, we can calculate a
Given a state |i and an operator O
number
h|O|i
hOi
= h| O|i
Matrix elements
p. 7/31
Hermitian Conjugation
One of the most important operations in complex linear
algebra is Hermitian conjugation. The Hermitian conjugate
is the complex conjugate of the transpose of an operator
O
. If in a particular basis O
= [aij ] then O
= [a ].
O
ji
Hermitian conjugation works similarly to transposition in
=B
A . When applied to
real linear algebra: (AB)
state vectors, (|i) = h|. Similarly, for dyads
(|ih|) = |ih|.
Note that Hermitian conjugation is not linear, but rather
is antilinear:
= a O
,
(aO)
(a|i) = a h|.
p. 8/31
N
X
|jihj| = I.
j=1
= 0.
A traceless operator has Tr{O}
p. 9/31
Tr{AB}
Tr{AB
Normal operators
O
=O
O
. Operators are
A normal operator satisfies O
diagonalizable if and only if they are normal. That is, for
we can always find an orthonormal basis
normal O
{|j i} such that
X
X
=
=
O
j |j ihj |, Tr{O}
j ,
j
Hermitian operators
One very useful class of operators are the Hermitian
that satisfy H
=H
. These are the complex
operators H
analogue of symmetric matrices. They are obviously
H
=H
2 = H
H
. The eigenvalues of a
normal: H
Hermitian matrix are always real. We will look at this in
more detail later.
We have already seen an example: the Pauli matrices.
!
!
!
0 1
0 i
1 0
X=
, Y =
, Z=
.
1 0
i 0
0 1
These matrices are obviously Hermitian. It is easy to
see that any 2 2 Hermitian matrix can be written
+ cY + dZ for some real values a, b, c, d.
aI + bX
p. 12/31
The Commutator
Matrix multiplication is noncommutative, in general. That
is, in general AB 6= B A. Given two operators A and B ,
B]
AB
B
A. [A,
B]
= 0 if and
their commutator is [A,
only if A and B commute.
Occasionally, one will encounter matrices that
B
= B
A. For example, the Pauli
anticommute: A
matrices anticommute with each other. In these cases,
it is sometimes helpful to define the anticommutator:
B}
AB
+B
A.
{A,
Orthogonal projectors
An orthogonal projector P is an Hermitian operator that
obeys P 2 = P . All the eigenvalues of P are either 0 or
1. The complement of a projector I P is also a projector.
Note that |ih| is a projector for any normalized state
vector |i. Such a projector is called one-dimensional; a
.
projector more generally has dimension d = Tr{P}
In dimension 2, any projector can be written in the form
/2 = |~n ih~n |,
P = I + ~n ~
M
X
j Pj ,
j=1
Unitary Operators
A unitary operator satisfies U U = U U = I. It is clearly a
normal operator. All of its eigenvalues have unit norm;
that is, |j | = 1 for all j . This means that
j = exp(ij )
U
means shortly.)
(We will clarify what exp(iH)
p. 16/31
p. 17/31
Operator space
The space of all operators on a particular Hilbert space
of dimension N is itself a Hilbert space of dimension
Functions of Operators
(where
It is common to write a function of an operator f (O)
f is ordinarily a function on the complex numbers) which is
itself an operator. Usually f (x) is defined by a Taylor series:
f (x) = c0 + c1 (x x0 ) + c2 (x x0 )2 + . For the operator
version, we write
2 +
+ c2 ( O
x0 I)
x0 I)
= c0 I + c1 (O
f (O)
Polar Decomposition
There are certain special forms in which operators can
always be written. One of these is the polar decomposition. For
, there is a unitary operator U and
any linear operator O
and B
such that
positive operators A
= U A = B
U ,
O
.
which we call the left and right polar decompositions of O
A positive operator is an Hermitian operator with all
nonnegative eigenvalues. In this case, the positive
operators A and B are uniquely given by
p
p
B
O
O,
O
.
A O
p. 22/31
Tensor products
The tensor (or Kronecker) product is a way of combining
two Hilbert spaces to produce a higher dimensional space.
Let |i be a state in a D1 -dimensional Hilbert space H1 and
|i be a state in a D2 -dimensional Hilbert space H2 . Then
we define |i |i to be a state in the D1 D2 -dimensional
space H1 H2 . Such a state is called a product state. Any
state in this larger space can be written as a linear
combination of product states
X
|i =
| i | i
where | i H1 and | i H2 .
p. 23/31
p. 24/31
k=1
j k |ji1 |ki2 .
j,k
p. 25/31
(aA + bA ) B
+ bB
) = aA B
+ bA B
.
A (aB
= A B
.
(A B)
A B
) = AA B
B
.
(A B)(
= Tr{A}Tr{
Tr{A B}
B}.
p. 26/31
(A B)(|i
|i) = A|i
B|i.
= h|A h|B.
(h| h|)(A B)
Matrix representation
What does the matrix representation of a tensor product
look like? If |i has amplitudes (1 , , D1 ) and |i has
amplitudes (1 , , D2 ), the state |i |i in H1 H2 is
represented as a D1 D2 -dimensional column vector:
1 1
1 2
1 |i
.
..
2 |i
|i |i =
=
.
1
D
.
2
..
2 1
D1 |i
..
D1 D2
p. 28/31
a1D B
a11 B
1
..
..
.
=
.
A B
.
.
.
.
aD D B
aD1 1 B
1 1
p. 29/31
|i =
t11
t12
t21
t22
p. 30/31
1 0 0
0
1 0 0 0
0 1 0
0 1 0 0
0
I I =
,
, Z I =
0 0 1 0
0 0 1 0
0 0 0 1
0 0 0 1
=
I X
0
1
0
0
1
0
0
0
0
0
0
1
0
0
1
0
,
X
Y
=
0 0 0 i
0 0 i 0
.
0 i 0 0
i 0 0 0
p. 31/31