0

Let $M$ be a $m\times n$ matrix with entries in a field $F$ and $R_{1}, R_{2}, \ldots, R_{m}$ be its rows. We also assume that null space of $M$ contains a non-zero element.

Let $v = \begin{bmatrix} v_{1} & v_{2} & \ldots & v_{n} \end{bmatrix}$ be a non-zero element of null space of the matrix $M$. If the row space of $M$ contains $\begin{bmatrix} 1 & 1 & \ldots & 1 \end{bmatrix}$, i.e., $\sum_{i=1}^{k} \mu_{i}R_{i} = \begin{bmatrix} 1 & 1 & \ldots & 1 \end{bmatrix} $ then it easy to show that

$$ \sum_{j=1}^{n} v_{j} = \begin{bmatrix} 1 & 1 & \ldots & 1 \end{bmatrix} \times v = \sum_{i=1}^{k} \mu_{i} R_{i} \times F = \sum_{i=1}^{k} \mu_{i} \big( R_{i} \times F \big) = 0.$$

I want to show that the converse: if the row space of the matrix $M$ does not have $\begin{bmatrix} 1 & 1 & \ldots & 1 \end{bmatrix}$, then there exists a non-zero element $v = \begin{bmatrix} v_{1} & v_{2} & \ldots & v_{n} \end{bmatrix}$ in the null space of $M$ such that $\sum_{j=1}^{n} v_{j} \neq 0$. Since the null space of a matrix is the intersection (or linear combination) of null spaces of its rows $R_{i}$s, I thought this would not be hard to prove. However, my attempts have proved futile so far.

  • I don’t believe it does since the matrix is not necessarily square. – HumbleStudent Oct 24 '22 at 23:06
  • This doesn't matter at all: in that post, the number $n$ of linear forms is not supposed to be the dimension of the vector space. – Anne Bauval Oct 25 '22 at 09:26
  • See.. in that link $L$ is a linear form and here our matrix $M$ only induces a linear map from $\mathbb{R}^{n}$ to $\mathbb{R}^{m}$ but not a linear functional on $\mathbb{R}^{n}$. – HumbleStudent Oct 25 '22 at 19:20
  • Each line does. – Anne Bauval Oct 25 '22 at 20:51
  • @AnneBauval I see.. my linear operators are those induced by the rows and the last one is by $\begin{bmatrix} 1 & 1 & \ldots & 1\end{bmatrix}$. I do not want to quote a result from Functional Analysis for something related to finite dimensional algebra. Any chance you know of a reference just for finite dimensional vector spaces? – HumbleStudent Oct 26 '22 at 00:19
  • 1
    As you can see from the accepted answer in the proposed duplicate, this is not a result of functional analysis but of linear algebra (not restricted to finite dimension). Do you mean you would prefer to quote some book on algebra in a paper you are writing? – Anne Bauval Oct 26 '22 at 05:42
  • Yes, please - that would be great! – HumbleStudent Oct 26 '22 at 20:07
  • @AnneBauval Actually, I found a reference. I think this answer https://math.stackexchange.com/questions/2358365/hoffman-and-kunze-linear-algebra-chapter-3-theorem-20?noredirect=1&lq=1 gives the same. – HumbleStudent Oct 26 '22 at 20:52
  • Fine! Thanks for having told me. It is a relief. – Anne Bauval Oct 26 '22 at 20:56

0 Answers0