$$A =\left(\begin{array}{cc}5 & -4 \\9 & -7\end{array}\right)$$

I found that the eigenvalues are $-1$ (algebriac multiplicity 2)

Therefore, the jordan form looks like this:

$$J =\left(\begin{array}{cc}-1 & 1 \\0 & -1\end{array}\right)$$

Also solving $(A-\lambda I$)$=\vec{0}$, when $\lambda = -1$ gives us the nullspace $(3,2)^T$.

However I need one more basis vector, and unsure how to find it?


from Jordan theorem we know that a matrix M exists such that $$M^{-1}AM=J$$

let $$M=[v,w]$$

then M has to satisfy the following system: $$AM=MJ$$ that is in your case $$Av=-v$$ $$Aw=v-w$$ once you find $v$ from the first equation you can find also $w$ from the second.

  • $\begingroup$ Thank you I understand $\endgroup$ – K Split X Nov 28 '17 at 23:11
  • $\begingroup$ that's good, I think it's a clear way to see how it works! $\endgroup$ – gimusi Nov 28 '17 at 23:13
  • $\begingroup$ Can you explain why (A-cI)w = v where c is some eigenvalue? What do you do when there is higher multiplicity? eg) mult. 3 in this case $\endgroup$ – skim Dec 5 '17 at 4:12
  • 1
    $\begingroup$ To find M you have already to know as J Matrix is done (by Jordan's Theorem). Once you have J you only have to solve the above system AM=MJ. $\endgroup$ – gimusi Dec 5 '17 at 14:08

You can take any vector which is linearly independent of the eigenvector, except that you need to adapt its length in order to get exactly $1$ (and not some other constant) in the upper right corner of $J$. For example, take something simple like $(k,0)^T$, see what matrix you get in the new basis, and then choose $k$ so that you get exactly $J$.


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.