I was reading about complex eigen vectors from here.

The given matrix is $$A= \begin{pmatrix} 3 & -2\\ 4 & -1\\ \end{pmatrix} $$

The roots are found as $1\pm2i$. The article then uses the eigen value $1+2i$ and rewrites the system as

$$(2-2i)x-2y=0 $$ $$4x-(2+2i)y=0$$

Then it goes on saying:

In fact the two equations are identical since $(2+2i)(2-2i)=8$.So the system reduces to one equation $(1-i)x-y=0$

I didnt get this.

  • $\begingroup$ The system of linear equations stems from inputting the value of the eigenvalue in the equation $\;\det (tI-A)\;$ , which makes the matrix $\;\lambda I-A\;$ singular. Thus, that homogeneous system has a non-trivial solution and thus one of the equations must be a linear combination of the other one(s) ... $\endgroup$ – Timbuc May 10 '15 at 19:43
  • $\begingroup$ For any $2 \times 2$ matrix $A$, if it is singular, then you must be able to write one column (row) in terms of the other. $\endgroup$ – copper.hat May 10 '15 at 19:44
  • $\begingroup$ Ok I got this: $2[(1-i)x-y]=(2-2i)x-2y$ and $[(1-i)x-y](2+2i) = 4x-(2+2i)y$, but still didnt get from where $(2+2i)(2-2i)=8$ has come. Must be doing something really stupid here. $\endgroup$ – Maha May 10 '15 at 19:54


from the system: $$ \begin{cases} (2-2i)x-2y=0\\ 4x-(2+2i)y=0 \end{cases} $$ multipling the first equation by $4$ and the second by $(2-2i)$, you have:

$$ \begin{cases} 4(2-2i)x-8y=0\\ 4(2-2i)x-(2+2i)(2-2i)y=0 \end{cases} $$

$$ \begin{cases} 4(2-2i)x-8y=0\\ 4(2-2i)x-8y=0 \end{cases} $$


I think that the remark about $(2+2i)(2-2i)=8$ is completely useless.

If $\lambda$ is an eigenvalue of the matrix $A$, then, by definition, $A-\lambda I$ has rank $<n$ (where $n$ is the order of the matrix $A$).

Thus the matrix of the system you're looking at has rank $<2$. Since the rank is at least $1$, you conclude that one of the equations is redundant, without any need to check. Just choose a non trivial one, in this case either of the two.

In case $n>3$ one can't predict which of the equations will be redundant, because the only information is that the rank is less than $n$, but it could be anything from $0$ to $n-1$, depending on the matrix and the eigenvalue.

  • $\begingroup$ OK if that is useless then how the reduction to $(1-i)x-y$ is done? $\endgroup$ – Maha May 10 '15 at 20:12
  • $\begingroup$ @Mahesha999 Just take the first equation and remove the common factor $2$. $\endgroup$ – egreg May 10 '15 at 20:13
  • $\begingroup$ hey damn thanks...just want to know if there is any fact behind: If $\lambda$ is an eigenvalue of the matrix $A$, then, by definition, $A-\lambda I$ has rank $<n$ (where $n$ is the order of the matrix $A$). I am missing something simple intuitive stuff? $\endgroup$ – Maha May 11 '15 at 10:29
  • $\begingroup$ @Mahesha999 There is $v\ne0$ with $Av=\lambda v$, which is the same as saying that $v\ne0$ is in the null space of $A-\lambda I$; having non zero null space and rank $<n$ are equivalent conditions for an $n\times n$ matrix. Thus one can think to the definition of eigenvalue as “$A-\lambda I$ is not invertible”. $\endgroup$ – egreg May 11 '15 at 11:35

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.