1
$\begingroup$

The problem is given below:

In the following $X, A, B$ and $C$ are all $n \times n$ matrices. Solve for $X$ and account for any assumptions made for each of the following.

(a) $AX + B = C$

(b) $AXA - B = C$

(c) $AX + B = X$

(d) $A^2 X + BX + C = A$

I solved (a) by $$ X = A^{-1} (C-B). $$

I'm confused how to solve the others, can someone explain the proces?

$\endgroup$
4
  • 1
    $\begingroup$ What fact are you using about $A$ to be able to write that as an answer? The remainder are similar -- $AXA = C+B \Rightarrow X = A^{-1}(C+B)A^{-1}$, and so on. You use the same property as the previous one. $\endgroup$
    – walkar
    Apr 3, 2016 at 22:24
  • $\begingroup$ @walkar what about c) X = X*(A-B)^-1 ? $\endgroup$
    – AdiT
    Apr 3, 2016 at 22:30
  • $\begingroup$ When you are solving for $X$, you cannot have $X$ on both sides of the equation. The solution for c) is $AX-X = -B \Rightarrow (A-Id)X = -B \Rightarrow X = (A-Id)^{-1}(-B)$. But again, you have to use a a particular assumption here. Can you identify it? $\endgroup$
    – walkar
    Apr 3, 2016 at 22:32
  • $\begingroup$ I'm still a little confused, but yes. $\endgroup$
    – AdiT
    Apr 3, 2016 at 22:44

2 Answers 2

1
$\begingroup$

For completeness, let me spell out what I've said in the comments.

For part a), we have $AX + B = C$, so to solve for $X$, we move $B$ over: $AX=C-B$, and then multiply by the inverse of $A$, i.e. $A^{-1}AX = A^{-1}(C-B) \Rightarrow X = A^{-1}(C-B)$. To do this, we implicitly assume $A^{-1}$ is a well-defined object -- that is, that $A$ is invertible.

In part b), we use the same property. $AXA-B=C \Rightarrow AXA = C+B \Rightarrow X = A^{-1}(C+B)A^{-1}$. Again, for this to make sense, we need $A$ is invertible.

For part c), we do the same procedure, more or less. Isolate $X$: $AX+B=X \Rightarrow B = X-AX$. Now we factor out $X$: $B = X-AX \Rightarrow B = (Id-A)X$.

The $Id$ appears in place of a $1$ as you would do in polynomial factoring -- for example, $x-ax$ as polynomials, we get $(1-a)x$. In a matrix space, the identity matrix $Id$ is the $1$. Further, we must also keep $X$ on the left, because in general, $AX \neq XA$ for any matrices $X$ and $A$. This is another way in which matrix spaces differ from polynomials, with which you might be more familiar.

So then $B = (Id-A)X \Rightarrow (Id-A)^{-1}B = X$. So we use the fact that $Id-A$ is invertible so that $(Id-A)^{-1}$ exists.

Part d) is the same. Isolate everything with $X$ on one side, factor out $X$ (being sure to respect the non-commutativity of the matrix multiplication), and multiply by the inverse of the coefficient on $X$. You must use that the coefficient on $X$ is invertible.

$\endgroup$
1
$\begingroup$

You should also account for assumptions you have made. What assumptions have you made in solving for $X$ in part (a)?

Regarding the other questions, follow the same process. Recall, however, that matrix multiplication is non-commutative in general. That is to say, $AX \neq XA$. This should be of use to you in part (b). Part (c) and (d) make use of the same assumptions and properties.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.