PH 217: Quantum Physics

Hilbert space

John Kondoro 1
1. Introducing the concept of space functions, specifically
Hilbert space
(a) Recall that in Cartesian 3-space a vector V is a set of three
numbers called components (𝑉𝑥 , 𝑉𝑦 , 𝑉𝑧 ). Any vector in this
space can be expanded in terms of the three unit vectors
𝒆 𝒙 , 𝒆𝒚 , 𝒆𝒛 .

(b) Under such conditions a vector, V is expressed as follows

𝑉 = 𝑒𝑥 𝑉𝑥 + 𝑒𝑦 𝑉𝑦 + 𝑒𝑧 𝑉𝑧 . The vectors are 𝒆𝒙 , 𝒆𝒚 , 𝒆𝒛 𝒂𝒓𝒆 said
to spun the vector space

(c) The inner (dot) product of two vectors (U and V) in the

space is defined as
𝑉. 𝑈 = 𝑉𝑥 𝑈𝑥 + 𝑉𝑦 𝑈𝑦 + 𝑉𝑧 𝑈𝑧

John Kondoro 2
2. A Hilbert space is much the same type of object.

(a) The elements of Hilbert space are functions instead of

three dimensional vectors.

(b) The similarity between the two is so close that the

functions are sometimes called vectors.

(c) Hilbert space has the following properties

The space is linear. A function space is linear under the

following conditions

John Kondoro 3
(c) The space is linear. A function space is linear under the
following conditions

(i) If a is a constant and ϕ is any element of the space, then aϕ

is also an element of the space.

(ii) If ϕ and ψ are any two elements of a space, then ϕ + ψ is

also an element of the space

(iii) There is an inner product, 𝜓 𝜑 , for any two elements in the

space. For functions defined in the interval 𝑎 ≤ 𝑥 ≤
𝑏 𝑖𝑛 𝑜𝑛𝑒 𝑑𝑖𝑚𝑒𝑛𝑠𝑖𝑜𝑛 , 𝑤𝑒 𝑚𝑎𝑦 𝑡𝑎𝑘𝑒:

𝑏 ∗
𝜑𝜓 = 𝑎
𝜓 𝜓𝑑𝑥
John Kondoro 4
(d) Any element of Hilbert space has a norm (“length”) that is
2 2
related to the inner product as 𝑛𝑜𝑟𝑚 𝑜𝑓 𝜑 = 𝜑 = 𝜑𝜑

An example of a Hilbert space is given by the set of functions

defined on the interval 0 ≤ 𝑥 ≤ 𝐿 with finite norm:

2 𝐿 ∗
𝜑 = 0
𝜑 𝜑 𝑑𝑥 <∞

(e) Hilbert space is closely akin to a vector space. Mathematicians

call it – an infinite-dimensional vector space.

If two vectors U and V in three dimensional vector space are

orthogonal to each other, the inner product vanishes if
𝜑𝜓 =0
John Kondoro 5
(f) similarly, two vectors in Hilbert space ϕ and ψ are said to be
orthogonal if 𝜑 𝜓 = 0

(g) Any function in Hilbert space can be expanded in a series of the

∞
sequence 𝜑𝑛 , 𝜑 𝑥 = 𝑛=1 𝑎𝑛 𝜑𝑛

(h) The basis vectors 𝜑𝑛 comprise an orthogonal set such that

𝜑𝑛 𝜑𝑛, = 0 𝑤ℎ𝑒𝑛 𝑛 ≠ 𝑛,

(i) 𝜑𝑛 Is a unit vector that has a unit “length” 𝜑𝑛 𝜑𝑛, = 𝜑 2 =1

(j) These two latter statements could be combined into a single

equation as follows 𝜑𝑛 𝜑𝑛, = 𝛿𝑛,𝑛,

John Kondoro 6
(k) The symbol 𝛿𝑛,𝑛, is called the Kronecker delta and is defined
as follows:

(i) 𝛿𝑛,𝑛, = 0 𝑓𝑜𝑟 𝑛 ≠ 𝑛, and 𝛿𝑛,𝑛, = 1 𝑓𝑜𝑟 𝑛 = 𝑛,

(ii) Any sequence of functions that obeys 𝜑𝑛 𝜑𝑛, = 𝛿𝑛,𝑛, is

called an orthonormal set

(iii) To show that 𝑎𝑛 is the projection of ϕ into 𝜑𝑛 , we first in

Dirac notation 𝜑 = 𝑛 𝑎𝑛 𝜑𝑛

Then we multiply from the left by 𝜑𝑛, and use the relation
𝜑𝑛 𝜑𝑛, = 𝛿𝑛,𝑛,

John Kondoro 7
 𝜑𝑛, 𝜑 = 𝑛 𝜑 𝑎𝑛 𝜑𝑛 = 𝑛 𝑎𝑛 𝜑𝑛, 𝜑𝑛 =

𝑛 𝑎𝑛 𝛿𝑛,𝑛, = 𝑎𝑛,

(iv) The coefficient 𝑎𝑛, is an inner product between the basis

vectors 𝜑𝑛, 𝑎𝑛𝑑 𝜑𝑛 . Since 𝜑𝑛, is a “unit” vector , 𝑎𝑛, is the
projection of 𝜑 onto 𝜑𝑛,

John Kondoro 8
 3. Delta function orthogonality

(a) The sequence spans the Hilbert space, it is the basis of the
Hilbert space

(b) The eigen functions of the momentum operator 𝑝.

1
𝜑𝑘 (𝑥) = 𝑒 𝑖𝑘𝑥
2𝜋

(c) The inner product is the set of functions is as follows

1 ∞ 𝑖𝑥 𝑘 , −𝑘
𝜑𝑘 𝜑 𝑘, = 𝑒 𝑑𝑥 = 𝛿(𝑘 , − 𝑘)
2𝜋 −∞

(d) It follows that the inner product between any two distinct
eigenvectors of the operator 𝑝 vanishes
John Kondoro 9
 4. FUNCTIONS IN HILBERT SPACE

(a) Any function in the Hibert space ℌ2 , may be expanded in

terms of eigenvectors 𝜑𝑘 . If 𝜑 𝑥 is any element of a Hilbert
space ℌ2 , then since 𝜑𝑘 spans this space, one may write
∞
𝜑 𝑥 = −∞
𝑏(𝑘)𝜑𝑘 (𝑥) 𝑑𝑘

(b) This is the Fourier integral representation of 𝜑 𝑥 and the

coefficient of expansion b(k) is the projection of 𝜑 𝑥 onto 𝝋𝒌

The integral can be re-written in the form

∞
𝜑 = −∞
𝑑𝑘 𝑏(𝑘)𝜑𝑘

If it is multiplied from the left with 𝜑𝑘 , , the result is

John Kondoro 10
 ∞ ∞
𝜑𝑘 , 𝜑 = 𝑑𝑘 𝜑 𝑏(𝑘)𝜑𝑘 = 𝑑𝑘𝑏(𝑘) 𝜑 𝜑𝑘
−∞ −∞

∞
𝜑𝑘 , 𝜑 = −∞
𝑑𝑘𝑏(𝑘) 𝛿 𝑘 , − 𝑘 = 𝑏(𝑘 , )

(c) The coefficient of expansion 𝒃(𝒌, ) is an inner product

between 𝝋 and 𝝋𝒌, hence may be termed a projection of
onto

(d) The vector 𝜑𝑘 is infinitely long

2 1 ∞
𝜑𝑘 = 𝜑𝑘 𝜑𝑘 = 𝛿 0 = 𝑑𝑘 =∞
2𝜋 −∞

John Kondoro 11
 Although this disqualifies the set 𝜑𝑘 for membership in ℌ2
, they nevertheless span the space.
(e) They comprise a valid set of basis vectors and the projection
of any function in ℌ2 onto any member of the basis 𝜑𝑘
gives a finite result

If 𝝋 is any function in ℌ2 then 𝜑𝑘 𝜑 < ∞

The functions 𝜑𝑘 may be renormalised in a form that

allows them to be members of ℌ2

John Kondoro 12
 4. Hermitian operators

(a) An average for a system in the state is given by…In Dirac

notation, this equation in one dimension is written as follows

𝐴 = 𝜓 ∗ (𝑥, 𝑡)𝐴𝜓(𝑥, 𝑡) 𝑑𝑥 = 𝜓 𝐴𝜓

Since t is a fixed parameter in this equation, the formula gives

the expectation of A at the time t.

The object 𝜓 𝐴𝜓 must be real for all in 𝜓 𝑖𝑛 ℌ2

If 𝐻 is the operator in the 1-D box corresponding to energy,

then

John Kondoro 13
 𝐿 ∗ ℏ2 𝜕 2
𝐸 = 𝜓 𝐻𝜓 = − 𝜓 𝜓𝑑𝑥 must be real for any
0 2𝑚 𝜕𝑥 2

function 𝜓 in ℌ1

(b) Quantum mechanics requires that eigenvalues of an operator

corresponding to a physical observable be real numbers.

The class of operators that have this property are called

Hermitian operators.

Consider an operator 𝐴 , the Hermitian Adjoint of is written 𝐴+

which is different from operator 𝐴

John Kondoro 14
(c) The Hermitian Adjoint of a complex number c is a complex
conjugate of c, that it, 𝑐 + = 𝑐 ∗

If 𝐴 is the operator and 𝜓 is any element of, then 𝐴𝜓 is also in

ℌ

For any two elements ψ1 and ψn in ℌ space , they can form an

inner product

𝜓1 𝐴𝜓𝑛

Suppose there is another operator 𝐴+ which is defined in

space ℌ for which

𝐴+ 𝜓1 𝜓𝑛 = 𝜓1 𝐴𝜓𝑛
John Kondoro 15
(d) Consider 𝐴 = a, a complex number, then

𝑎+ 𝜓1 𝜓𝑛 = 𝜓1 𝑎𝜓𝑛 = 𝑎 𝜓1 𝜓𝑛 = 𝑎∗ 𝜓1 𝜓𝑛

Equating the first and last terms, we get 𝑎+ = 𝑎∗

𝝏
Consider the operator 𝑫 =
𝝏𝒙

∞ 𝜕
Then, 𝜓1 𝐷𝜓𝑛 = −∞
𝑑𝑥 𝜓1 𝜕𝑥 𝜓𝑛 = 𝜓1 ∗ 𝜓𝑛 +∞
−∞
−

∞ 𝜕 ∗
𝑑𝑥 𝜓 𝜓𝑛
−∞ 𝜕𝑥 1

𝜓1 𝐷𝜓𝑛 = −𝐷 𝜓1 𝜓𝑛

The “surface” term is zero because 𝜓1 𝑎𝑛𝑑 𝜓𝑛 are elements of

ℌ2 John Kondoro 16
 Therefore 𝐷+ = −𝐷

(e) The operators that have a property such that 𝜓1 𝐴𝜓𝑛 =

𝐴𝜓1 𝜓𝑛 are called Hermitian operators

+
If and are two Hermitian operators, then 𝐴𝐵 = 𝐴+ 𝐵+

+
𝐴𝐵 + 𝐵 𝐴 = 𝐵+ 𝐴+ + 𝐴+ 𝐵+ = 𝐵𝐴 + 𝐴𝐵 = 𝐴𝐵 + 𝐵𝐴

(f) The square of a hermitian operator

2 + + 2
𝐴 = 𝐴𝐴 = 𝐴+ 𝐴+ = 𝐴𝐴 = 𝐴

John Kondoro 17
Example: Consider the momentum operator 𝑝 . For the free
particle case, 𝑝 is Hermitian if far all 𝜓1 𝑎𝑛𝑑 𝜓𝑛 𝑖𝑛 ℌ2 ,
𝜓1 𝑝𝜓𝑛 = 𝑝𝜓1 𝜓𝑛

Developing the left-hand side, we have

∞ ∗ 𝜕
𝜓
−∞ 1
−𝑖ℏ 𝜓𝑛 𝑑𝑥 =
𝜕𝑥

∗ ∞ ∞ 𝜕 ∗
− 𝑖ℏ 𝜓1 𝜓𝑛 −∞ + 𝑖ℏ −∞ 𝜕𝑥
𝜓 1 𝜓𝑛 𝑑𝑥

∞ ∗ 𝜕 ∞ ∗ 𝜕
𝜓
−∞ 1
−𝑖ℏ 𝜓𝑛 𝑑𝑥 = 𝜓
−∞ 1
−𝑖ℏ 𝜓𝑛 𝑑𝑥 =
𝜕𝑥 𝜕𝑥

𝑝𝜓1 𝜓𝑛

This shows that the operator 𝑝 is Hermitian

John Kondoro 18
 𝑝2
Example: Consider a free particle Hamiltonian 𝑝, 𝐻 =
2𝑚

+
+ 𝑝2 𝑝2
𝐻 = = = 𝐻 Therefore 𝐻 is Hermitian
2𝑚 2𝑚

𝑝2
For a particle in a potential field V(x), 𝐻 = + 𝑉(𝑥)
2𝑚

Since V(x) is a real function that merely multiplies, it is

Hermitian
∞ ∗ 𝑉𝜓 𝑑𝑥 ∞ ∗ 𝜓
𝜓1 𝑉𝜓𝑛 = −∞
𝜓 1 𝑛 = −∞
𝑉𝜓 1 𝑛
𝑑𝑥 =

𝑉𝜓1 ∗ 𝜓𝑛 = 𝑉𝜓1 𝜓𝑛

since 𝑝, 𝑉 𝑎𝑟𝑒 𝐻𝑒𝑟𝑚𝑖𝑡𝑖𝑎𝑛 . It follows that 𝐻 is also Hermitian

John Kondoro 19
 5. Properties of Hermitian Operators

(a) Eigenvalues of Hermitian operators are real

Consider a Hermitian operator 𝐴 and let 𝜑𝑛 and 𝑎𝑛

represent eigen functions and eigenvalues respectively.

𝐴𝜑𝑛 = 𝑎𝑛 𝜑𝑛

In Dirac notation 𝐴𝜑𝑛 = 𝑎𝑛 𝜑𝑛 or 𝐴 𝜑𝑛 = 𝑎𝑛 𝜑𝑛

Multiply the left with 𝜑𝑛 to get

𝜑𝑛 𝐴𝜑𝑛 = 𝜑𝑛 𝑎𝑛 𝜑𝑛 = 𝑎𝑛 𝜑𝑛 𝜑𝑛

Since 𝐴 is Hermitian, then

John Kondoro 20
 𝐴𝜑𝑛 𝜑𝑛 = 𝑎𝑛 𝜑𝑛 𝜑𝑛 = 𝑎𝑛 ∗ 𝜑𝑛 𝜑𝑛 ⟹ 𝑎∗ 𝑛 = 𝑎𝑛 and 𝑎𝑛 is
real

(b) Eigenfunctions of Hermitian operators are orthogonal

Consider a Hermitian operator 𝐴 and let 𝜑𝑛 and 𝑎𝑛

represent eigenfunctions and eigenvalues respectively.

In Dirac notation 𝐴𝜑𝑛 = 𝑎𝑛 𝜑𝑛 or 𝐴 𝜑𝑛 = 𝑎𝑛 𝜑𝑛

Multiply from the left with eigenvector of 𝐴, 𝜑𝑙 to get

𝜑𝑙 𝐴𝜑𝑛 = 𝑎𝑛 𝜑𝑙 𝜑𝑛

Since 𝐴 is Hermitian, then the left hand side can be expressed

John Kondoro 21
 𝑨𝝋𝒍 𝝋𝒏 = 𝒂𝒍 ∗ 𝝋𝒍 𝝋𝒏 = 𝒂𝒍 𝝋𝒍 𝝋𝒏

The eigenvalue is real because it is an eigenvalue of a

Hermitian operator.

Recall the equations

(i) 𝜑𝑙 𝐴𝜑𝑛 = 𝑎𝑛 𝜑𝑙 𝜑𝑛

(ii) 𝐴𝜑𝑙 𝜑𝑛 = 𝑎𝑙 ∗ 𝜑𝑙 𝜑𝑛 = 𝑎𝑙 𝜑𝑙 𝜑𝑛

Subtracting equations (i) and (ii) above would give

𝑎𝑙 − 𝑎𝑛 𝜑𝑙 𝜑𝑛 = 0

If 𝑎𝑙 ≠ 𝑎𝑛 , 𝑡ℎ𝑒𝑛 𝜑𝑙 𝜑𝑛 = 0
John Kondoro 22
If 𝑎𝑙 ≠ 𝑎𝑛 , 𝑡ℎ𝑒𝑛 𝜑𝑙 𝜑𝑛 = 0 is an expression of
orthogonality of the set of functions 𝜑𝑛

If the functions are normalized, they may be generalized to

read the following

𝜑𝑙 𝜑𝑛 = 𝛿𝑙𝑛

Therefore the eigenvalues of a Hermitian operator are real

and its eigen functions are orthogonal

John Kondoro 23