Given matrices $A, B \in \Bbb R^{3 \times 3}$ whose ranks satisfy $\mbox{rank} (A), \mbox{rank} (B) \geq 2$, I would like to prove that for large (or small) enough scalar $\alpha \in \mathbb{R} \setminus \{0\}$ the following does hold.

$$\mbox{rank} (A+\alpha B) \geq 2$$

This seems to be true by hand waving argument, but I would like to find some short proof or reference that formally proves it. Thanks.


Since rank$(A)\geq 2$, the matrix $A$ has a $2\times 2$ submatrix $a$ with nonzero determinant; the determinant is a continuous function of the matrix elements, so adding a sufficiently small perturbation $\alpha B$ to $A$ will leave $\det a\neq 0$ and hence the rank of $A+\alpha B$ remains $\geq 2$.

If instead of small $\alpha$ you wish to take large $\alpha$, define $B'=\alpha B$, $A'=\alpha A$ and work with $B'+(1/\alpha)A'$, where $B'$ has rank $\geq 2$.

| cite | improve this answer | |
  • 4
    $\begingroup$ By the way, this argument shows that the rank function is lower-semicontinuous. $\endgroup$ – HenrikRüping Jul 29 at 11:24

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.