SP351 HW6

Nikhil Vasan
September 2021

1a. Given each set of basis vectors use Gram-Schmidt Orthogonalization

to find a set of orthogonal vectors
a. A = (0, 2, 0, 0), B = (3, −4, 0, 0), C = (1, 2, 3, 4)

A
e1 = = (0, 1, 0, 0) (1)
|A|

B − (e1 · B)e1
e2 = = (1, 0, 0, 0) (2)
|B − (e1 · B)e1 |

C − (C ė1 )e1 − (C · e2 )e2 3 4

e3 = = (0, 0, √ , √ ) (3)
|C − (C ė1 )e1 − (C · e2 )e2 | 25 25
3 4
= (0, 0, , ) (4)
5 5
As illustrated above, the Gram-Schmidt Orthogonalization process involves
iteratively subtracting from each vector the component of the previous unit
vectors within the original vector.
b. A = (0, 0, 0, 7), B = (2, 0, 0, 5), C = (3, 1, 1, 4)

A
e1 = = (0, 0, 0, 1) (5)
|A|

B − (e1 · B)e1
e2 = = (1, 0, 0, 0) (6)
|B − (e1 · B)e1 |

1
 C − (C ė1 )e1 − (C · e2 )e2 1 1
e3 = = (0, √ , √ , 0) (7)
|C − (C ė1 )e1 − (C · e2 )e2 | 2 2
c. A = (6, 0, 0, 0), B = (1, 0, 2, 0), C = (4, 1, 9, 2)
A
e1 = = (1, 0, 0, 0) (8)
|A|

B − (e1 · B)e1
e2 = = (0, 0, 1, 0) (9)
|B − (e1 · B)e1 |

C − (C ė1 )e1 − (C · e2 )e2 1 2

e3 = = (0, √ , 0, √ ) (10)
|C − (C ė1 )e1 − (C · e2 )e2 | 5 5
1b. Find the norms of A, B and demonstrate that they obey the Cauchy-
Schwartz Inequality, that is, |A · B| ≤ |A||B|.
a. A = (3 + i, 1, 2 − i, −5i, i + 1), B = (2i, 4 − 3i, 1 + i, 1).
q
|A| = Σ40 A∗i Ai (11)

We take notice that because A, B ∈ C5 , we must use the inner-product as

opposed to the dot product, as such, we take the hermitian conjugates of our
vectors and multiply term-by-term, this is equivalent to taking the modulus
of the elements of our vectors
p √ √
|A| = |3 + i| + |1| + |2 − i| + | − 5i| + |i + 1| = 10 + 1 + 5 + 25 + 2 = 43
(12)
We do the same for B.
p √ √
|B| = |2i| + |4 − 3i| + |1 + i| + |3i| + |1| = 4 + 25 + 2 + 9 + 1 = 41
(13)
Let us check the Cauchy-Schwartz Inequality

|A · B| = |(3 − i)(2i) + (1)(4 − 3i) + (2 + i)(1 + i) + (5i)(3i) + (i − 1)(1)|

(14)
√
= |6i + 2 + 4 − 3i + 1 + 3i − 15 + i − 1| = |7i − 9| = 130 (15)

2
 √ p
As we can see, 130 < (41)(43), and the inequality holds.

b. A = (2, 2i − 3, 1 + i, 5i, i − 2), B = (5i − 2, 1l3 + i, 2i, 4)

p √ √
|A| = |2| + |2i − 3| + |1 + i| + |5i| + |i − 2| = 4 + 13 + 2 + 25 + 5 = 49 = 7
(16)

p √ √
|B| = |5i − 2| + |1| + |3 + i| + |2i| + |4| = 29 + 1 + 10 + 4 + 16 = 60
(17)

|A∗ · B| = |(2)(5i − 2) + (2i + 3)(1) + (1 − i)(3 + i) + (5i)(2i) + (i + 2)(4)|

(18)
√
= |10i − 4 + 2i + 3 + 4 − 2i − 10 + 4i + 8| = |14i − 1| = 170
(19)

Again we see that our inequality holds.

3
 2. Verify that the matrix is orthogonal
" #
√1 −2
√
5 5
C= √2 √1
(20)
5 5
" # " #
√1 −2
√
5 5
and that the eigenvectors A = √2
,B= √1
, are orthogonal. Let us first
5 5
determine whether the eigenvectors are orthogonal.
1 −2 1 2
A · B = ( √ )( √ ) + ( √ )( √ ) = 0 (21)
5 5 5 5
Furthermore, we notice something interesting about the magnitude of these
vectors ...
1
A · A = (12 + 22 ) = 1 (22)
5
Interesting! Let’s see if this holds in B
1
B · B = (12 + (−2)2 ) = 1 (23)
5
We also notice something interesting about the matrix C
 
C= A B (24)

C is the matrix, the columns of which, form an orthonormal basis. It must

be an orthogonal matrix! Let us test that.
   T   
T
  AT AA AB T 1 0
CC = A B = = (25)
BT BAT BB T 0 1

C is an orthogonal matrix, as CC T = I.
2b. Verify the following statements.
a. If C is orthogonal, and M is symmetric, prove that C −1 M C is symmetric.
Proof. We use direct proof. By hypothesis, C is orthogonal, that is, CC T =
CC −1 = I, therefore C T = C −1 . As such, C −1 M C = C T M C. By definition
of matrix transposition, (C T M C)T = C T M T (C T )T = C T M C = C −1 M C.

b. If C is orthogonal and M is anti-symmetric, show that CM C −1 is

anti-symmetric.

4
Proof. We use direct proof. By hypothesis, M is anti-symmetric, that is
M T = −M . Therefore, (C −1 M C)T = (C T M C)T = C T M T C = −C −1 M C.
Therefore, C −1 M C is anti-symmetric.

5
 3.  Find the
 eigenvalues and eigen-vectors of the following matrix. a
3 −2
M=
−2 0
the eigen-vectors of this matrix come in the form
M⃗v = λ⃗v (26)

The eigen-vectors of this matrix come in the form

M⃗v = λ⃗v = λI⃗v (27)
As such,
(M − λI)⃗v = 0 (28)
⃗v is a non-trivial vector, as such ⃗v ∈ N [M − λI], because the null space of
M − λI is non-null, the det(M − λI) = 0. Evaluating our determinant we
obtain a polynomial in λ.
 
3 − λ −2  2
 −2 −λ = −(3 − λ)λ − 4 = λ − 3λ + 4 = (λ + 1)(λ − 4) = 0 (29)
 

Our constraints are satisfied when λ = −1, 4. We can substitute these values
of λ into our earlier equation to obtain our eigenvectors
     
3 − (−1) −2 v1 0
= (30)
−2 0 − (−1) v2 0
Let us create a system of equations and see how v1 and v2 are related.
0 = 4v1 − 2v2 (31)
= −2v1 + v2 (32)
 
1
Therefore any vector of the form α is and eigenvector of this matrix with
2
eigenvalue -1 Let us solve when λ = 4
     
3 − (4) −2 v1 0
= (33)
−2 0 − (4) v2 0
Let us create a system of equations and see how v1 and v2 are related.
0 = −1v1 − 2v2 (34)
= −2v1 − 4v2 (35)

6
As we previously solved, we can solve
 for the eigenvector with eigenvalue 4,
2
which is any vector of the form α
−1
 
2 0 2
b. M = 0 2 0 
2 0 −1
As in 3a. We can solve for eigenvalues of this matrix
 
2 − λ 0 2 
 
 0 2 − λ 0  = (2 − λ)2 (−1 − λ) − 4(2 − λ) = (2 − λ)((2 − λ)(−1 − λ) − 4)
 
 2 0 −1 − λ
(36)
= (2 − λ)(λ2 − λ − 2 − 4) = (2 − λ)(λ2 − λ − 6) = 0
(37)
= (2 − λ)(λ − 3)(λ + 2) = 0 (38)

Therefore our eigenvalues are λ = 2, 3, −2, we follow our previous method in

finding our eigenvectors For λ = 2
    
0 0 2 v1 0
0 0 0  v2  = 0 (39)
2 0 −3 v3 0

We notice that v1 = v3 = 0, however,  v3 is can be any value that we would

0
like it to be. As such, any multiple of 1 is an eigenvector with eigenvalue

0
λ = 2.
Now for λ = 3     
−1 0 2 v1 0
 0 −1 0  v2  = 0 (40)
2 0 −4 v3 0
We solve this system of equations to obtain our eigenvector

0 = −v1 + 2v3 (41)

= −v2 (42)
= 2v1 − 4v3 (43)

7
v2 = 0 as is indicated by the second row. However we notice equations 1 and
3 are multiples
  of each other. As such, v1 = 2v2 , and any multiple of the
2
vector 0 is an eigenvector with eigenvalue λ = 3
1
Finally we solve for our last eigenvalue, λ = −2.
    
4 0 2 v1 0
0 4 0 v2  = 0 (44)
2 0 1 v3 0

Again a similar situation as our previous

 eigenvalue ensues, and we obtain
1
that any multiple of the vector 0 is an eigenvector with λ = −2 as an
2
eigenvalue.

8
  
0 −i
4a. Diagonalize the pauli spin matrix σy = . That is, find hte
i 0
matrix D = C −1 σy C.
We notice that σy is hermitian, as such it’s eigenvalues are real and the
eigenvectors are orthogonal, and if we normalize our eigenvectors we will
produce a unitary matrix with our eigenvectors as our column vectors.
 
−λ −i  2
 i −λ = λ − 1 = 0 (45)
 

We find that λ = ±1. Let us find the eigenvectors. We start with λ = 1

    
−1 −i v1 0
= (46)
i −1 v2 0

We solve this system to obtain our first eigenvector

−v1 − iv2 = 0 (47)

iv1 − 1 = 0 (48)

We see that

v1 = −iv2 (49)
v1 = −i (50)
 
−i
Solving this system we notice that v1 = −i, v2 = 1, our eigenvector
1
with eigenvalue λ = 1. Let us now solve for λ = −1
    
1 −i v1 0
= (51)
i 1 v2 0

Let us solve for our second eigenvector

v1 − iv2 = 0 (52)
iv1 + v2 = 0 (53)

We reduce

v1 = iv2 (54)
v1 = iv2 (55)

9
  
i
Our eigenvector with eigenvalue λ = −1 is . Let us define the matrix
  1
1 i −i
U= 2√ . We note that U is unitary.
1 1
    
† 1 i 1 −i i 1 0
U U= = (56)
2 −i 1 1 1 0 1
     
1 0 −i i −i 1 i −i −1 0
σy U = √ =√ (57)
2 i 0 1 1 2 1 1 0 1
We multiply by U −1 from the left and obtain
   
−1 1 −i 1 0 −i i −i
U σy U = (58)
2 i 1 i 0 1 1

Equation (57) comes from the fact that because the columns of U are eigen-
vectors of σy we know that σy U is equivalent to U times a matrix  that 
−1 0
multiplies each column by a scalar, which is what the matrix D =
0 1
−1 dagger
does. (58) comes after we multiply U by U = U .
−1 −1
b. What are U σx U and U σx U ?
      
−1 1 −i 1 0 1 i −i 1 −i 1 1 1
U σx U = = (59)
2 i 1 1 0 1 1 2 i 1 i −i
   
1 0 −2i 0 −i
= = (60)
2 21 0 i 0

We notice that
U −1 σx U = σy (61)
Let us atttempt σz
      
−1 1 −i 1 1 0 i −i 1 −i 1 i −i
U σz U = = (62)
2 i 1 0 −1 1 1 2 i 1 −1 −1
 
1 0 −2
= = −σx (63)
2 −2 0

U −1 σx U = −σx (64)

10