# Jordan basis

1. Jun 29, 2011

### estro

In my book I see that the author finds the jordan basis for matrix A={(-3,9),(1,3)} by finding nullspace for (A-3I) and (A-3I)^2 without any justification, do I miss something trivial here?

2. Jun 29, 2011

### micromass

Staff Emeritus
Hi estro!

Your book just tries to find the generalized eigenvectors of the eigenvalue 3. Your book should mention what generalized eigenvectors are and why they are important though...
What book are you using?

3. Jun 29, 2011

### estro

Thank you for the quick reply!

I'm an Israeli student so I'm pretty sure you won't be familiar with the book. [It's called Linear Algebra 2 from the open university]
Can you refer me to a source where I can read and clear my concerns regarding this idea?

4. Jun 29, 2011

### micromass

Staff Emeritus

Check the section on "Complex matrices" which explains what generalized eigenvectors are and why they are important.
Also, check out the "Example" section which gives an example on how to calculate the Jordan normal form.

5. Jun 29, 2011

### estro

Thank you! I should have been looking there in the first place=...)
If you don't mind I'll ask here again if wikipedia won't be enough for me.

6. Jun 29, 2011

### micromass

Staff Emeritus
Feel free to ask anything you don't understand!

7. Jul 1, 2011

### estro

After reading the wiki and the proof for the existence of the jordan form I think that I'm getting into the idea, however I was able to think about the following example:

Lets choose matrix A={(2,0,0),(0,0,1),(0,0,0)} so the characteristic polynomial is also the minimal: p(t)=(t-2)t^2.

Now this is how I find the jordan basis:
1. NullSpace(A-2I)=Sp{(1,0,0)}
2. NullSpace(A-0I)=SP{(0,1,0)}
3. NullSpace(A-0I)^2={(0,0,1),(0,1,0)}

So the jordan basis is {(1,0,0),(0,1,0),(0,0,1)}
But how can I in what order to write these column vectors in my matrix P? [to satisfy P^{-1}AP]

8. Jul 1, 2011

### micromass

Staff Emeritus
The order in which to write the column vectors doesn't matter, you just need to group the vectors from the same eigenspace together. So you could write

$$P=[(1,0,0),(0,1,0),(0,0,1)]$$

or

$$P=[(0,0,1),(0,1,0),(1,0,0)]$$

the order of the vectors will only induce a permutation of the Jordan blocks, and that won't matter.

9. Jul 1, 2011

### estro

Thanks!
But did I understand the concept of finding jordan basis? []
Will I be able to find the jordan basis for every possible matrix with this technique?

10. Jul 1, 2011

### micromass

Staff Emeritus
Yes, you understood the technique. Take a look at http://www.google.be/url?sa=t&sourc...g=AFQjCNEs8yAwLNlJ4PCC-tbXajtjSyupdw&cad=rja" for more examples of the Jordan normal form. If you can do all those examples, then you understand the concept well!!

But finding Jordan basisses is a very time-consuming thing for large matrices

Last edited by a moderator: Apr 26, 2017