Vectors and Matrices, Problem Set 3

Scalar products and determinants

Prof Andre Lukas, MT2013

Un-starred questions indicate standard problems students should be familiar with.

Starred questions refer to more ambitious problems which require a deeper understanding of the material.

1. Calculate the determinants of the matrices

   √ √ √ 
0 −i i 3 − 2 − 3
1 √
(a) A =  i 0 −i  , (b) B = √  1 6 −1  .
8
−i i 0 2 0 2

Are the matrices (i) real, (ii) diagonal, (iii) symmetric, (iv) antisymmetric, (v) singular, (vi) orthogonal,
(vii) Hermitian, (viii) anti-Hermitian, (ix) unitary?

2. Use the Gram-Schmidt procedure to find an ortho-normal basis of R3 (with the standard scalar
product), starting with the basis
     
1 2 0
v 1 =  1  , v2 =  1  , v3 =  2  .
0 2 −1

Check your result.

3. Let A be an n × n matrix.
(a) Why are the statements “A is invertible” and det(A) 6= 0 equivalent?
(b) For which values of the parameters a, b is the matrix
 
a 1 a
A =  1 b −1 
0 −1 a

not invertible?

4. ∗ In Rn , we have n−1 linearly independent vectors v1 , . . . , vn−1 . Define the vector w with components
wi = det(v1 , . . . , vn−1 , ei ), where ei are the standard unit vectors.
(a) Show that w is perpendicular (with respect to the standard scalar product in Rn ) to all vectors
va , where a = 1, . . . , n − 1.
(b) Show that |w| = det(v1 , . . . , vn−1 , n), where n = w/|w|.
(c) Work out the vector w if va = ea , for a = 1, . . . , n − 1.
(d) Show that in three dimensions w can be written in terms of a cross product.

1
 5. Solve the following system of linear equations

x + 2y + 3z = 2
3x + 4y + 5z = 4
x + 3y + 4z = 6

by
(a) calculating the matrix inverse
(b) Cramer’s method
(c) row reduction on the augmented matrix.
If you had to write a computer program solving systems of linear equations (of arbitrary and
possibly large size) which of the above methods would you base it on?

R∞ 2
6. ∗ On the vector space of polynomials f : R → R we define hf, gi = −∞ dx e−x f (x)g(x).
(a) Why does this define a scalar product?
(b) Consider the polynomials p0 (x) = 1/n0 , p1 (x) = 2x/n1 and p2 (x) = (4x2 − 2)/n2 , where the
na are real numbers. Show that these polynomials are orthogonal under the above scalar product.
(c) Determine the numbers na such that the polynomials pa are normalized to one, so hpa , pa i = 1.
R∞ 2 R∞ 2 √ R∞ 2
(Hint: You can use that −∞ dx xn e−x = 0 for n odd (why?) and −∞ dx e−x = π, −∞ dx x2 e−x =
√ R∞ 2 √
π/2, −∞ dx x4 e−x = 3 π/4.)

7. The vector space V is equipped with a hermitian scalar product h·, ·i and an ortho-normal basis
e1 , . . . , en .
(a) Show that non-zero and pairwise orthogonal vectors w1 , . . . , wk are linearly independent.
(b) Show that the coordinates vi of a vector v, relative to the basis e1 , . . . , en , are given by
vi = hei , vi.
(c) Show that the scalar product of two vectors u, v can be written as hu, vi = ni=1 hu, ei ihei , vi.
P

(d) A second ortho-normal basis e01 , . . . , e0n is related to the first one by e0j = i Uij ei , where Uij
P
are complex numbers. Show that Uij = hei , e0j i and that the matrix U with entries Uij is unitary.

8. Consider Rn with the standard scalar product hv, wi = vT w and orthogonal matrices R, that is,
matrices satisfying hRv, Rwi = hv, wi for all vectors v, w.
(a) Show that orthogonal matrices can, alternatively, also be characterized by the equation RT R = 1
and that det(R) = ±1.
(b) Focus on n = 2 and show that two-dimensional orthogonal matrices can be written in the form
 
cos ϕ ∓ sin ϕ
R(ϕ) = .
sin ϕ ± cos ϕ

Which of these matrices correspond to two-dimensional rotations? What is the interpretation of the other
matrices?
(c) For two-dimensional rotations, show that R(ϕ1 )R(ϕ2 ) = R(ϕ1 + ϕ2 ).
(d) The vectors x = (x, y)T and x0 = (x0 , y 0 )T are related by a rotation, so x0 = R(ϕ)x. What is
the relation between the two associated complex numbers z = x + iy and z 0 = x0 + iy 0 ?

2
 9. ∗ (a) A three-dimensional rotation, R3 , around the z-axis can be constructed from the previous
two-dimensional rotations by setting
 
cos ϕ − sin ϕ 0
R3 (ϕ) =  sin ϕ cos ϕ 0  ,
0 0 1

and analogously for three-dimensional rotations R1 and R2 around the x and y axis. Construct a general
three-dimensional rotation by combining three such rotations, that is, work out R = R1 (α1 )R2 (−α2 )R3 (α3 )
for three angles α1 , α2 and α3 .
(b) P
Show that for small angles αi the three-dimensional rotations from part (a) can be written as
R = 13 + 3i=1 αi Ti + · · · , where the dots stand for terms of quadratic or higher order in the angles, and
determine the (angle-independent) matrices T1 , T2 , T3 .
(c) The change of a vector x under a rotation R is given by δx = Rx − x. Show that for small-angle
rotations this can be written as δx = α × x + · · · , where α = (α1 , α2 , α3 )T and the dots stand for
quadratic or higher order terms in the angles.

10.∗ A bi-linear form on R2 is defined by hv, wi = vT ηw, where η = diag(−1, 1).

(a) Why is this bi-linear form not a scalar product?
(b) Consider the 2 × 2 matrices Λ which leave the above bi-linear form invariant, that is, which
satisfy hΛv, Λwi = hv, wi for all vectors v, w. Show that these matrices can also be characterized by
the equation ΛT ηΛ = η and that det(Λ) = ±1.
(c) Show that the matrices Λ with det(Λ) = 1 and Λ11 > 0 can be written in the form
 
cosh ξ sinh ξ
Λ(ξ) = ,
sinh ξ cosh ξ

where ξ is a real parameter. Re-write Λ in terms of the parameter β, defined as β = tanh ξ.

(d) Verify that Λ(ξ1 )Λ(ξ2 ) = Λ(ξ1 + ξ2 ). What does this rule imply for the parameter β, that is, if
Λ(β1 )Λ(β2 ) = Λ(β), how does β depend on β1 and β2 ?