-2

Let $A$ be a $m \times n$ matrix and $B$ be a $m \times k$ matrix. Let $N(A),N(B),R(A),R(B)$ be the nullspace of $A$, the nullspace of $B$, the range of $A$, and the range of $B$, respectively. Prove that $$\dim(N(A|B)) = \dim(N(A)) + \dim(N(B)) + \dim(R(A) \cap R(B))$$

Any idea to solve it? Thanks for help in advanced.

Akira
  • 17,367
lap lapan
  • 2,188

2 Answers2

1

Whenever you see something like $\dim N(A)$, you know that “rank-nullity” is lurking.

What does this say in your case? You have three matrices, so you can say

  1. $\dim R(A)+\dim N(A)=n$
  2. $\dim R(B)+\dim N(B)=k$
  3. $\dim R(A|B)+\dim N(A|B)=n+k$

Thus what you want to prove can be rewritten as $$ n+k-\dim R(A|B)=n-\dim R(A)+k-\dim R(B)+\dim\bigl(R(A)\cap R(B)\bigr) $$ which becomes $$ \dim R(A|B)=\dim R(A)+\dim R(B)-\dim\bigl(R(A)\cap R(B)\bigr) $$ Can you link $R(A|B)$ with $R(A)$ and $R(B)$ and apply another well-known result?

egreg
  • 238,574
0

Treat $A,B, A|B$ as linear map form $R^n, R^k, R^{n+k}$ respectively to $R^m$.

By rank-nullity, the equality is

$(n+k)-dim(R(A|B)) = n-dim(R(A)) + k-dim(R(B)) + dim(R(A) \cap R(B))$

$dim(R(A|B)) = dim(R(A)) + dim(R(B)) - dim(R(A) \cap R(B))$

If we prove $R(A|B)=R(A)+R(B)$, the equality is equvalent to dimension fomula of $R(A)+R(B)$: $dim(R(A)+R(B)) = dim(R(A)) + dim(R(B)) - dim(R(A) \cap R(B))$

This formula is true, you can see:Dimension of the sum of two vector subspaces

Here is the rest of the proof($R(A|B)=R(A)+R(B)$):

For $x \in R(A)+R(B)$, there exist $v_a,v_b$ s.t. $x=Av_a+Bv_b$.

Let $v$ be (n+k)-vector with first n slots $v_a$ and last k slot $v_b$.

Then $x=(A|B)v \in R(A|B)$

Similarly, $x\in R(A|B)$, then $x=(A|B)v $ for some v.

Let $v_a$ be first n slots of v, $v_b$ be last k slots of v. Then $x=Av_a+Bv_b\in R(A)+R(B)$.