18
$\begingroup$

I've been looking at some LU Decomposition problems and I understand that making a matrix A reduced to the form A=LU , where L is a lower triangular matrix and U is a upper triangular matrix, however I am having trouble understanding the steps to get to these matrices. Could someone please explain the method for LU Decomposition in detail, preferably excluding the concept of permutation matrices? ( We haven't talked about permutation matrices in class yet so our professor forbids us to use them for the decomposition).

$\endgroup$
27
$\begingroup$

$LU$ decomposition is really just another way to say Gaussian elimination. If you're familiar with that, putting the pieces together is easy. Here is an example. Let $$ A=A^{\left(0\right)}=\left[\begin{array}{ccc} 8 & 1 & 6\\ 4 & 9 & 2\\ 0 & 5 & 7 \end{array}\right]. $$ Proceed by Gaussian elimination. The first multiplier is $\ell_{2,1}=4/8=0.5$ (this is the multiplier that allows us to cancel $a_{2,1}=4$ using the first row) and the second is $\ell_{3,1}=0/8=0$. We arrive at $$ A^{\left(1\right)}=\left[\begin{array}{ccc} 8 & 1 & 6\\ 0 & 8.5 & -1\\ 0 & 5 & 7 \end{array}\right]. $$ To cancel out $a_{3,2}^{\left(1\right)}=5$, we use the multiplier $\ell_{3,2}=5/8.5\approx0.5882$ to yield $$ A^{\left(2\right)}\approx\left[\begin{array}{ccc} 8 & 1 & 6\\ 0 & 8.5 & -1\\ 0 & 0 & 7.5882 \end{array}\right] $$ which yields the $LU$ decomposition $$ A=LU\approx\left[\begin{array}{ccc} 1 & 0 & 0\\ 0.5 & 1 & 0\\ 0 & 0.5882 & 1 \end{array}\right]\left[\begin{array}{ccc} 8 & 1 & 6\\ 0 & 8.5 & -1\\ 0 & 0 & 7.5882 \end{array}\right]. $$ Note that $L$ is just made up of the multipliers we used in Gaussian elimination with $1$s on the diagonal, while $U$ is just $A^{\left(2\right)}$.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Very interesting. Thank you for explaining it so well $\endgroup$ – Ishan Jain Apr 10 '19 at 15:08

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.