Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have the following problems when solving a linear equation.

Let $A=(a_{i,j})_{n \times n}$ be a non-negative matrix with $a_{i,j} \in (0,1)$, and let $0<r<1$ be a scalar. Now we define a vector $x=(x_i)$ of length $n$ as follows:

(I) The first component of $x$ is 1, that is $x_1=1$.

(II) The other components of $x$ (except the first entry of $x$) satisfies the following equation:

$$r \cdot Ax=x .$$

Or equivalently, both (I) and (II) tell that $x$ satisfies the following equation: $$max \{r \cdot Ax,e_1\} =x $$ where $max$ is entry-wise maximum operator, and $e_1={(1,0,\cdots,0)}^T$ .

Based on such a defintion of $x$, I want study the relations between $x$ and the vector $y$ that satisfies $r \cdot Ay=y$ (including the first entry of $y$). In other words, can we compute $x$ from $y$ ?

I would really appreciate any suggestions.

share|cite|improve this question
I've retagged as it was in no way related to Combinatorics or Numerical Methods.. For the question: What exactly is the dimension of $x = (x_i)$ when $i \neq 1$? Note that it shouldn't be $(n-1) \times 1$ as you're multiplying it with an $n \times n$ matrix. – Rijul Saini Sep 1 '12 at 10:31
I think you should add "numeric" tag since it is hard to solve $x$ algebrically. – John Smith Sep 1 '12 at 10:44
I still don't understand the question.. Can you give some example describing what's happening here? – Rijul Saini Sep 1 '12 at 10:58
John: Where are we at? If you consider that this question is a lost cause, then erase it. Otherwise, modify it, addressing the concerns raised in the comments. – Did Sep 9 '12 at 15:32

(This answer applies to the original version of the post, which might not reflect the question the OP has in mind. Some heroic tries to reach a proper formulation of the question are occurring right now...)

Both $x$ and $y$ are eigenvectors of $A$ for the eigenvalue $1/r$, hence, if the eigenvalue $1/r$ is simple, then $x$ and $y$ are proportional. Since $x_1=1$, this yields $y=y_1\cdot x$.

share|cite|improve this answer
I don't think $x$ is an eigenvector of $A$. Actually, $x$ satisfies the following equation: $max \{Ax,e_1\} =x $, where $max$ is entry-wise maximum, and $e_1={(1,0,\cdots,0)}^T$ . – John Smith Sep 1 '12 at 10:41
@JohnSmith How on Earth is any entry-wise maximum involved in your question?? You might wish to mention what $Ax$ is to you, for $A$ an $n\times n$ matrix and $x$ a vector of size $n$, if not the vector of size $n$ everybody understands. (And I seem to read eigenvector in your title, or am I dreaming?) – Did Sep 1 '12 at 10:46
$x$ is a vector of $n$ size. The first entry of $x$ is always 1, and other entries of $x$ satisfy the equation: $r\cdot Ax=x$, as clearly described in my post. The title is eigenvector because $y$ is the eigenvector of $A$ that satisfies $r\cdot Ay=y$. So my goal is to compute $x$ from $y$! – John Smith Sep 1 '12 at 10:48
Your post states that $rAx=x$, not $Ax=x$, please make up your mind. Anyway, if $Ax=x$ or if $rAx=x$, then $x$ is an eigenvector of $A$, whether its first coordinate $x_1$ is $1$ or not. Sorry but your clearly described in my post might be a tad too optimistic here. – Did Sep 1 '12 at 10:51
Here is a suggestion: rephrase (II), which at the moment makes no sense. – Did Sep 1 '12 at 10:58

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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