Class 12 Maths, Worked in Full Past papersGet updates
Chapter 3 · Matrices

Miscellaneous · Q3: Find x,y,zx, y, z with A′A=IA'A=I

Miscellaneous ExerciseTranspose and multiplication · Sections 3.4.5 and 3.5
The question

Find the values of xx⁠, yy⁠, zz if the matrix

A=[02yzxy−zx−yz]A = \begin{bmatrix} 0 & 2y & z \\ x & y & -z \\ x & -y & z \end{bmatrix}

satisfy the equation A′A = I.

Exam answer A′=[0xx2yy−yz−zz]A' = \begin{bmatrix} 0 & x & x \\ 2y & y & -y \\ z & -z & z \end{bmatrix}

Let A′A=[cij]A'A = [c_{ij}]⁠. Then

c11=0+x2+x2=2x2c12=0+xy−xy=0c13=0−xz+xz=0c21=0+xy−xy=0c22=4y2+y2+y2=6y2c23=2yz−yz−yz=0c31=0−xz+xz=0c32=2yz−yz−yz=0c33=z2+z2+z2=3z2\begin{aligned} c_{11} &= 0 + x^2 + x^2 = 2x^2 \\ c_{12} &= 0 + xy - xy = 0 \\ c_{13} &= 0 - xz + xz = 0 \\ c_{21} &= 0 + xy - xy = 0 \\ c_{22} &= 4y^2 + y^2 + y^2 = 6y^2 \\ c_{23} &= 2yz - yz - yz = 0 \\ c_{31} &= 0 - xz + xz = 0 \\ c_{32} &= 2yz - yz - yz = 0 \\ c_{33} &= z^2 + z^2 + z^2 = 3z^2 \end{aligned} ∴ A′A=[2x20006y20003z2]\therefore\ A'A = \begin{bmatrix} 2x^2 & 0 & 0 \\ 0 & 6y^2 & 0 \\ 0 & 0 & 3z^2 \end{bmatrix}

As A′A=IA'A = I⁠, equating corresponding entries,

2x2=1⇒x=±126y2=1⇒y=±163z2=1⇒z=±13\begin{aligned} 2x^2 = 1 &\Rightarrow x = \pm\frac{1}{\sqrt2} \\ 6y^2 = 1 &\Rightarrow y = \pm\frac{1}{\sqrt6} \\ 3z^2 = 1 &\Rightarrow z = \pm\frac{1}{\sqrt3} \end{aligned}

∴ x=±12\therefore\ x = \pm\dfrac{1}{\sqrt2}⁠, y=±16y = \pm\dfrac{1}{\sqrt6}⁠, z=±13z = \pm\dfrac{1}{\sqrt3}⁠.

Want the reason behind every step, the basics and the checks?

The one idea this problem is built on

Row ii of A′A' is column ii of A. So the (i,j)(i,j) entry of A′AA'A is column ii of A times column jj of A, multiplied entry by entry and added. Here the three columns of A are arranged so that every product of two different columns cancels to 0. Only the diagonal survives, giving A′AA'A as a diagonal matrix with 2x22x^2⁠, 6y26y^2⁠, 3z23z^2 on the diagonal, and setting it equal to I gives three simple equations.

Given, to find, and the plan
Given
The 3×33\times3 matrix A above, with real unknowns x,y,zx, y, z⁠, and the condition A′A=IA'A = I (the 3×33\times3 identity).
To find
All real x,y,zx, y, z for which A′A=IA'A = I⁠.
Plan
(1) Write down A′A' and check the orders. (2) Compute all nine entries of A′AA'A⁠. (3) Equate entry by entry with I. (4) Solve each equation, keeping both signs. (5) Check the answers.

Basics you need

1Transpose

The rows of A′A' are the columns of A, in order. A 3×33\times3 matrix has a 3×33\times3 transpose. Basics: reading a table sideways

2Row times column

The (i,j)(i,j) entry of PQPQ is pi1q1j+pi2q2j+pi3q3jp_{i1}q_{1j} + p_{i2}q_{2j} + p_{i3}q_{3j}⁠. Basics: pair up, multiply, add

3Equal matrices

Two matrices of the same order are equal exactly when every pair of corresponding entries is equal. Lesson: equality of matrices

4Solving kx2=1kx^2 = 1

For k>0k>0⁠, x2=1kx^2 = \frac1k has the two real solutions x=±1kx = \pm\frac{1}{\sqrt k}⁠. Forgetting the minus sign loses half the answers. Basics: solving equations

Solution

  1. Write down A′A' and check the orders

    The columns of A are

    col 1:(0, x, x)col 2:(2y, y, −y)col 3:(z, −z, z)\begin{aligned} \text{col 1} &: (0,\ x,\ x) \\ \text{col 2} &: (2y,\ y,\ -y) \\ \text{col 3} &: (z,\ -z,\ z) \end{aligned}

    and these become the rows of the transpose:

    A′=[0xx2yy−yz−zz].A' = \begin{bmatrix} 0 & x & x \\ 2y & y & -y \\ z & -z & z \end{bmatrix}.

    A′A' is 3×33\times3 and A is 3×33\times3⁠: columns of A′A' =3==3= rows of A, so A′AA'A is defined and is 3×33\times3⁠, the same order as I=I3I = I_3⁠. The equation A′A=IA'A = I therefore makes sense.

    We read the transpose off column by column rather than trying to "flip" the picture in one go, because that avoids sign slips with the entries −z-z and −y-y⁠.
    Definition 3 (Section 3.5): if A=[aij]A = [a_{ij}]⁠, then A′=[aji]A' = [a_{ji}]⁠. Lesson: transpose
  2. Row 1 of A′AA'A

    Row 1 of A′A' is (0,x,x)(0, x, x)⁠. Multiply it by each column of A:

    p11=0⋅0+x⋅x+x⋅x=2x2p12=0⋅2y+x⋅y+x⋅(−y)=0p13=0⋅z+x⋅(−z)+x⋅z=0\begin{aligned} p_{11} &= 0\cdot0 + x\cdot x + x\cdot x = 2x^2 \\ p_{12} &= 0\cdot2y + x\cdot y + x\cdot(-y) = 0 \\ p_{13} &= 0\cdot z + x\cdot(-z) + x\cdot z = 0 \end{aligned}
    In p12p_{12} the terms xyxy and −xy-xy cancel; in p13p_{13} the terms −xz-xz and xzxz cancel. These cancellations happen for every value of x,y,zx, y, z⁠, not just special ones, because of how the signs in A are placed.
    Row ii of the first factor times column jj of the second gives the (i,j)(i,j) entry. Basics: pair up, multiply, add
  3. Row 2 of A′AA'A

    Row 2 of A′A' is (2y,y,−y)(2y, y, -y)⁠:

    p21=2y⋅0+y⋅x+(−y)⋅x=0p22=2y⋅2y+y⋅y+(−y)(−y)=4y2+y2+y2=6y2p23=2y⋅z+y⋅(−z)+(−y)⋅z=2yz−yz−yz=0\begin{aligned} p_{21} &= 2y\cdot0 + y\cdot x + (-y)\cdot x = 0 \\ p_{22} &= 2y\cdot2y + y\cdot y + (-y)(-y) \\ &= 4y^2 + y^2 + y^2 = 6y^2 \\ p_{23} &= 2y\cdot z + y\cdot(-z) + (-y)\cdot z \\ &= 2yz - yz - yz = 0 \end{aligned}
    In p22p_{22} all three terms are squares, so they cannot cancel: (−y)(−y)=+y2(-y)(-y) = +y^2⁠. In p23p_{23} the first term 2yz2yz is cancelled by the two terms −yz-yz⁠. Also p21=p12p_{21} = p_{12}⁠, as it must be, because A′AA'A is always symmetric: (A′A)′=A′(A′)′=A′A(A'A)' = A'(A')' = A'A⁠.
    (−y)(−y)=y2(-y)(-y) = y^2⁠: the product of two negatives is positive. Basics: pair up, multiply, add
  4. Row 3 of A′AA'A⁠, and the whole product

    Row 3 of A′A' is (z,−z,z)(z, -z, z)⁠:

    p31=z⋅0+(−z)⋅x+z⋅x=0p32=z⋅2y+(−z)⋅y+z⋅(−y)=2yz−yz−yz=0p33=z⋅z+(−z)(−z)+z⋅z=3z2\begin{aligned} p_{31} &= z\cdot0 + (-z)\cdot x + z\cdot x = 0 \\ p_{32} &= z\cdot2y + (-z)\cdot y + z\cdot(-y) \\ &= 2yz - yz - yz = 0 \\ p_{33} &= z\cdot z + (-z)(-z) + z\cdot z = 3z^2 \end{aligned}

    Assembling the nine entries:

    A′A=[2x20006y20003z2].A'A = \begin{bmatrix} 2x^2 & 0 & 0 \\ 0 & 6y^2 & 0 \\ 0 & 0 & 3z^2 \end{bmatrix}.
    The product is a diagonal matrix for every choice of x,y,zx, y, z⁠. The diagonal entries are the "lengths squared" of the columns (each column times itself), and the zeros say that different columns are "perpendicular" (their products cancel). This is exactly why the condition A′A=IA'A = I is easy to meet here.
    A diagonal matrix has all its non-diagonal entries zero (Section 3.3 (iv)). Lesson: types of matrices
  5. Equate with I, entry by entry

    [2x20006y20003z2]=[100010001]\begin{bmatrix} 2x^2 & 0 & 0 \\ 0 & 6y^2 & 0 \\ 0 & 0 & 3z^2 \end{bmatrix} = \begin{bmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 1 \end{bmatrix}

    Both matrices are 3×33\times3⁠, so they are equal exactly when all nine pairs of entries agree. The six off-diagonal pairs read 0=00 = 0⁠, which is true whatever x,y,zx, y, z are, so they impose no condition. The three diagonal pairs give

    2x2=1,6y2=1,3z2=1.2x^2 = 1, \qquad 6y^2 = 1, \qquad 3z^2 = 1.
    It is worth saying out loud that the off-diagonal equations are automatically satisfied. If one of them had come out as, say, xy=0xy = 0⁠, it would have been an extra condition and could have ruled out some of the solutions below.
    Definition 2 (Section 3.3.1): equal matrices have equal orders and equal corresponding entries. Lesson: equality of matrices
  6. Solve each equation, keeping both signs

    2x2=1  ⟺  x2=12  ⟺  x=±126y2=1  ⟺  y2=16  ⟺  y=±163z2=1  ⟺  z2=13  ⟺  z=±13\begin{aligned} 2x^2 = 1 &\iff x^2 = \tfrac12 \iff x = \pm\tfrac{1}{\sqrt2} \\ 6y^2 = 1 &\iff y^2 = \tfrac16 \iff y = \pm\tfrac{1}{\sqrt6} \\ 3z^2 = 1 &\iff z^2 = \tfrac13 \iff z = \pm\tfrac{1}{\sqrt3} \end{aligned}

    The three equations are independent of one another (each involves one unknown), so the signs can be chosen freely: there are 2×2×2=82\times2\times2 = 8 solutions (x,y,z)(x, y, z) in all. In rationalised form, x=±22x = \pm\frac{\sqrt2}{2}⁠, y=±66y = \pm\frac{\sqrt6}{6}⁠, z=±33z = \pm\frac{\sqrt3}{3}⁠.

    A positive number has two real square roots. Every sign choice works because only x2x^2⁠, y2y^2⁠, z2z^2 appear in A′AA'A⁠. Changing the sign of xx changes the sign of the first column of A, and that does not change any of the column products in Steps 2 to 4.
    If t2=ct^2 = c with c>0c > 0⁠, then t=ct = \sqrt c or t=−ct = -\sqrt c⁠. Basics: solving equations
Result

A′A=IA'A = I exactly when

x=±12,y=±16,z=±13,\begin{gathered} x = \pm\frac{1}{\sqrt2}, \quad y = \pm\frac{1}{\sqrt6}, \\ z = \pm\frac{1}{\sqrt3}, \end{gathered}

with the three signs chosen independently (8 solutions). (Textbook answer: x=±12x = \pm\frac{1}{\sqrt2}⁠, y=±16y = \pm\frac{1}{\sqrt6}⁠, z=±13z = \pm\frac{1}{\sqrt3}⁠.)

Check it with numbers

Take all plus signs: x≈0.70710678x \approx 0.70710678⁠, y≈0.40824829y \approx 0.40824829⁠, z≈0.57735027z \approx 0.57735027⁠. Then

2x2≈2(0.5000000)=1.00000006y2≈6(0.1666667)=1.00000003z2≈3(0.3333333)=1.0000000\begin{aligned} 2x^2 &\approx 2(0.5000000) = 1.0000000 \\ 6y^2 &\approx 6(0.1666667) = 1.0000000 \\ 3z^2 &\approx 3(0.3333333) = 1.0000000 \end{aligned}

✓. With exact values, x2=12x^2 = \frac12⁠, y2=16y^2 = \frac16⁠, z2=13z^2 = \frac13 give 2⋅12=6⋅16=3⋅13=12\cdot\frac12 = 6\cdot\frac16 = 3\cdot\frac13 = 1⁠. The verification script substitutes all 8 sign choices into A and confirms A′A=IA'A = I exactly for each.

Common mistakes
  1. Computing AA′AA' instead of A′AA'A⁠. The order matters. (Here AA′AA' is not diagonal in general.)
  2. Giving only the positive roots. x2=12x^2 = \frac12 has two solutions.
  3. Sign slips in A′A', especially the −z-z and −y-y entries. Read the columns of A one at a time.
  4. Ignoring the off-diagonal equations without checking them. They happen to be 0=00 = 0 here, but you only know that after computing them.
What to take away
  1. The (i,j)(i,j) entry of A′AA'A is column ii of A times column jj of A. This turns the product into a set of "column products", and A′AA'A is always symmetric.
  2. A′A=IA'A = I means every column times itself gives 1 and different columns give 0. Such a matrix is called orthogonal (a name you will meet later).
  3. When an equation involves only squares, both signs of each root are solutions, and independent choices multiply the number of solutions.

Every fact used, and where it comes from

FactUsed inSource
Transpose: A′=[aji]A' = [a_{ji}]Step 1Definition 3, Section 3.5 (Lesson: transpose)
Product entry formula; order of a productSteps 1–4Section 3.4.5 (Lesson: matrix multiplication)
A′AA'A is symmetricStep 3Section 3.5.1, properties (i) and (iv)
Diagonal and identity matricesSteps 4, 5Section 3.3 (iv) and (vi) (Lesson: types of matrices)
Equality of matricesStep 5Definition 2, Section 3.3.1 (Lesson: equality of matrices)
All eight solutions; decimalsStep 6, CheckSolved and substituted with sympy in the site's verification script
Found an error, or want more?

Tell us if something on this page is wrong or unclear, or if you would like a fuller explanation, another method or a new feature. Send feedback on this problem →