Proving that a Jordan basis is not uniquely determined.

64 Views Asked by At

It seems that a Jordan basis, for, say, an operator $φ:V→V$ with $V$ a vector space can have several Jordan bases. However I don't see how this is true and I didn't achieve to prove it. Indeed I tried assuming the uniquness of a Jordan basis of $\varphi$ to achieve a contradiction but I did not find any.. I also looked up on mathstack but the only thing I found was this:

Jordan basis unique?

And it doesn't realy help because I'm more looking for the general case, which is unclear to me. Could anyone explain why this is true and help me proving it ?

1

There are 1 best solutions below

0
On BEST ANSWER

A simple proof: for any operator $\varphi$, any basis $\mathcal B_1 = \{v_1,\dots,v_n\}$, and any non-zero constant $k$, the basis $$ \mathcal B_2 = \{kv_1,\dots,kv_n\} $$ is such that the matrix of $\varphi$ with respect to $\mathcal B_1$ is equal to the matrix of $\varphi$ with respect to $\mathcal B_2$. Thus, if $\mathcal B_1$ is a Jordan basis for $\varphi$ (over any algebraically closed and therefore infinite field), we can now generate infinitely many other Jordan bases.