1
$\begingroup$

I have a 3x3 matrix with non-zero entries ONLY along the main diagonal and the diagonal above. There are exactly two non zero diagonals in the matrix like this

\begin{pmatrix} a & 0 & 0 \\ d & b & 0 \\ 0 & e & c \\ \end{pmatrix}

I would like to compute exp(A) or at least a good approximation. While this matrix looks quite simple, I need to compute the exponential several hundreds of time with slightly different matrices, hence I need an efficient algorithm.

$\endgroup$
0
$\begingroup$

Git, your matrices on the RHS do not commute.

With Maple, if $(b-c)(a-b)(a-c)\not=0$ then $e^A=\begin{pmatrix} e^a& 0& 0\\\dfrac{ -d(e^b-e^a)}{a-b}& e^b& 0\\\dfrac{ -ed(e^ba-e^bc-e^ab+e^ac-e^ca+e^cb)}{(-c+b)(a-b)(-c+a)}& \dfrac{e(-e^c+e^b)}{-c+b}& e^c\end{pmatrix}$

$\endgroup$
  • $\begingroup$ Of course they don't, thank you. $\endgroup$ – Git Gud Apr 1 '14 at 15:03
0
$\begingroup$

Basically we are talking about Jordan-Chevalley decomposition. Read the following: http://en.wikipedia.org/wiki/Matrix_exponential#Projection_case . The sections Projection case, Nilpotent case and Generalization. So you could easily implement those formulas into a program such as mathematica / matlab...

$\endgroup$

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.