Matrices of different dimensions in a matrix equation

564 Views Asked by At

Given two matrices A, B, respectively, and an equation XA=B; how can we obtain X and discuss the system?

\begin{pmatrix} 1 & a\\ a & 1 \\ \end{pmatrix} \begin{pmatrix} a & 1 \\ a^2 & 0 \\ 1 & 1 \\ \end{pmatrix}

I am not familiar with this structure and I can't even picture what the dimensions of X are. Any general matrix I write down looks uncomplete.

1

There are 1 best solutions below

0
On BEST ANSWER

$X$ must be a $3 \times 2$ matrix; we may write

$X = \begin{pmatrix} x_{11} & x_{12} \\ x_{21} & x_{22} \\ x_{31} & x_{32} \end{pmatrix}; \tag 1$

we may thus write out the system

$XA = B \tag 2$

as

$\begin{pmatrix} x_{11} & x_{12} \\ x_{21} & x_{22} \\ x_{31} & x_{32} \end{pmatrix} \begin{pmatrix} 1 & a \\ a & 1 \end{pmatrix} = \begin{pmatrix} a & 1 \\ a^2 & 0 \\ 1 & 1 \end{pmatrix}; \tag 3$

here

$A = \begin{pmatrix} 1 & a \\ a & 1 \end{pmatrix}, \tag 4$

and

$B = \begin{pmatrix} a & 1 \\ a^2 & 0 \\ 1 & 1 \end{pmatrix}; \tag 5$

we may write out the system (3) as a set of $6$ linear equations in two variables each:

$x_{11} + ax_{12} = a, \tag 6$

$ax_{11} + x_{12} = 1; \tag 7$

$x_{21} + ax_{22} = a^2, \tag 8$

$ax_{21} + x_{22} = 0, \tag 9$

$x_{31} + ax_{32} = 1, \tag{10}$

$ax_{31} + x_{32} = 1; \tag{11}$

the system of equations (6)-(11) is in fact easy to solve, one pair $(x_{i1}, x_{i2})$ at a time; for example, we multiply (6) by $a$:

$ax_{11} + a^2x_{12} = a^2, \tag{12}$

and subtract (7) from (12):

$(a^2 - 1) x_{12} = a^2 - 1, \tag{13}$

whence

$x_{12} = 1, \tag{14}$

provided of course that

$a \ne \pm 1; \tag{15}$

having $x_{12}$ we find via (6) that

$x_{11} = a(1 - x_{12}) = 0; \tag{16}$

the remaining equations (8)-(11) may be similarly solved.

The system (2), (3) may also be addressed in a somewhat more matrix-oriented manner by setting

$Y = \begin{pmatrix} x_{11} & x_{12} \\ x_{21} & x_{22} \end{pmatrix}, \; \mathbf y = (x_{31}, x_{32}), \; C = \begin{pmatrix} a & 1 \\ a^2 & 0 \end{pmatrix}; \mathbf d = (1, 1); \tag {17}$

then (3) may be written in the from

$\begin{pmatrix} Y \\ \mathbf y \end{pmatrix} A = \begin{pmatrix} C \\ \mathbf d \end{pmatrix}, \tag{18}$

leading to

$YA = C, \; \mathbf y A = \mathbf d, \tag{19}$

whence

$Y = CA^{-1}, \; \mathbf y = \mathbf d A^{-1}; \tag{20}$

where it is easily seen that

$A^{-1} = \dfrac{1}{1 - a^2} \begin{pmatrix} 1 & -a \\ -a & 1 \end{pmatrix}. \tag{21}$

We assume $a = \pm 1$ in deriving (17)-(21); in the event that $a = \pm 1$ we see that (8)-(9) become

$x_{21} \pm x_{22} = 1, \; \pm x_{21} + x_{22} = 0, \tag{22}$

which lead to the contradicion

$\pm 1 = 0; \tag{23}$

thus we see there is no solution to (2) when $a = \pm 1$.