3
$\begingroup$

I have read about the Thomas algorithm. Right now, I am trying to use it to solve the following linear system

$$\begin{bmatrix} b_1 & c_1 & 0 & \dots & 0 & a_1\\ a_2 & b_ 2 & c_2 & 0 & \dots & 0\\ 0 & a_3 & b_3 & c_3 & 0 & \dots\\ \vdots & \vdots & \vdots & \vdots & \ddots & \vdots\\ c_n & 0 & \dots & 0 & a_n & b_n\end{bmatrix} \begin{bmatrix}x_1\\x_2\\x_3\\ \vdots\\x_n\end{bmatrix} = \begin{bmatrix}r_1\\r_2\\r_3\\ \vdots\\r_n\end{bmatrix}$$ I have not been able to work out an algorithm for this matrix. Can someone point me to some reference that I can read about this? Thank you.

$\endgroup$
6
  • $\begingroup$ Does your matrix have any special properties? Is (strictly) diagonally dominant by rows or symmetric positive definite? $\endgroup$ – Carl Christian May 29 '16 at 20:56
  • $\begingroup$ No. It's pretty much just a tridiagonal system of equations matrix except that the first and last row have an extra element like above. $\endgroup$ – Kane Billiot May 29 '16 at 20:58
  • $\begingroup$ Can you guarantee that the $a_i$, $b_i$, $c_i$ are not zero? $\endgroup$ – mvw May 29 '16 at 21:08
  • $\begingroup$ Yes. All three are non-zeroes. $\endgroup$ – Kane Billiot May 29 '16 at 21:12
  • $\begingroup$ My opinion (maybe I am wrong) is that you have too few zeros (25%) to expect a "miracle" solution compared to standard methods for solving a linear system. $\endgroup$ – Jean Marie May 29 '16 at 22:06
1
$\begingroup$

The Thomas algorithm can be modified to solve this "periodic" case, as explained here.

$\endgroup$
1
  • 1
    $\begingroup$ The link does not work anymore. Could you provide a different one? $\endgroup$ – Kane Billiot Feb 22 '17 at 15:46

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.