Question from Jesse about Gaussian Elimination and LU factorisation.

Join the discussion
Ask a follow-up here, or get your own question answered by working scientists, mathematicians and engineers — people, not an autocomplete.
Real named experts · corrections over time · the nuance an AI answer skips
1 reply · 4K views
Prove It
Gold Member
MHB
Messages
1,434
Reaction score
20
View attachment 5370

This system can be written as a matrix equation $\displaystyle \begin{align*} A\,\mathbf{x} = \mathbf{b} \end{align*}$ as

$\displaystyle \begin{align*} \left[ \begin{matrix} \phantom{-}2 & 4 & \phantom{-}0 & 1 \\ \phantom{-}2 & 8 & -2 & 7 \\ -2 & 2 & -2 & 7 \\ \phantom{-}0 & 8 & -5 & 11 \end{matrix} \right] \left[ \begin{matrix} x_1 \\ x_2 \\ x_3 \\ x_4 \end{matrix} \right] = \left[ \begin{matrix} \phantom{-}7 \\ \phantom{-}3 \\ -7 \\ -12 \end{matrix} \right] \end{align*}$

To get the LU factorisation we need to use Gaussian Elimination on the coefficient matrix...

$\displaystyle \begin{align*} A = \left[ \begin{matrix} \phantom{-}2 & 4 & \phantom{-}0 & 1 \\ \phantom{-}2 & 8 & -2 & 7 \\ -2 & 2 & -2 & 7 \\ \phantom{-}0 & 8 & -5 & 11 \end{matrix} \right] \end{align*}$

To eliminate the terms under the main diagonal in the first column, we will apply Row 2 - Row 1 to Row 3 and Row 3 - (-1)Row 1 to Row 3. As we are eliminating the elements $\displaystyle \begin{align*} a_{2,1} \end{align*}$ and $\displaystyle \begin{align*} a_{3, 1} \end{align*}$, that means in our lower triangular matrix, which has 1 as all the elements on the main diagonal and the multipliers of the rows as its coefficients under the main diagonal, will have $\displaystyle \begin{align*} \mathcal{l}_{2,1} = 1 \end{align*}$ and $\displaystyle \begin{align*} \mathcal{l}_{3,1} = -1 \end{align*}$. Since we didn't have to do anything with element $\displaystyle \begin{align*} a_{4,1} \end{align*}$ that means the multiplier is 0, and thus $\displaystyle \begin{align*} \mathcal{l}_{4,1} = 0 \end{align*}$. A has now become

$\displaystyle \begin{align*} \left[ \begin{matrix} 2 & 4 & \phantom{-}0 & 1 \\ 0 & 4 & -2 & 6 \\ 0 & 6 & -2 & 8 \\ 0 & 8 & -5 & 11 \end{matrix} \right] \end{align*}$

To eliminate the terms under the main diagonal in the second column, we will apply Row 3 - (3/2)Row 2 to Row 3 and Row 4 - 2 Row 2 to Row 4. As we are eliminating the elements $\displaystyle \begin{align*} a_{3,2} \end{align*}$ and $\displaystyle \begin{align*} a_{4,2} \end{align*}$, this means that $\displaystyle \begin{align*} \mathcal{l}_{3,2} = \frac{3}{2} \end{align*}$ and $\displaystyle \begin{align*} \mathcal{l}_{4,2} = 2 \end{align*}$. A has now become

$\displaystyle \begin{align*} \left[ \begin{matrix} 2 & 4 & \phantom{-}0 & \phantom{-}1 \\ 0 & 4 & -2 & \phantom{-}6 \\ 0 & 0 & \phantom{-}1 & -1 \\ 0 & 0 & -1 & \phantom{-}1 \end{matrix} \right] \end{align*}$

To eliminate the term under the main diagonal in the third column, we will apply Row 4 - (-1)Row 3 to Row 4. As we are eliminating the element $\displaystyle \begin{align*} a_{4,3} \end{align*}$ that means that $\displaystyle \begin{align*} \mathcal{l}_{4,3} = -1 \end{align*}$. A has now become

$\displaystyle \begin{align*} \left[ \begin{matrix} 2 & 4 & \phantom{-}0 & \phantom{-}1 \\ 0 & 4 & -2 & \phantom{-}6 \\ 0 & 0 & \phantom{-}1 & -1 \\ 0 & 0 & \phantom{-}0 & -2 \end{matrix} \right] \end{align*}$

So we have found that our lower triangular matrix $\displaystyle \begin{align*} L = \left[ \begin{matrix} \phantom{-}1 & 0 & \phantom{-}0 & 0 \\ \phantom{-}1 & 1 & \phantom{-}0 & 0 \\ -1 & \frac{3}{2} & \phantom{-}1 & 0 \\ \phantom{-}0 & 2 & -1 & 1 \end{matrix} \right] \end{align*}$ and our upper triangular matrix $\displaystyle \begin{align*} U = \left[ \begin{matrix} 2 & 4 & \phantom{-}0 & \phantom{-}1 \\ 0 & 4 & -2 & \phantom{-}6 \\ 0 & 0 & \phantom{-}1 & -1 \\ 0 & 0 & \phantom{-}0 & -2 \end{matrix} \right] \end{align*}$.

This reduces the system to $\displaystyle \begin{align*} L\,U\,\mathbf{x} = \mathbf{b} \end{align*}$. Notice that $\displaystyle \begin{align*} U\,\mathbf{x} \end{align*}$ is a column matrix when multiplied out, which we can call $\displaystyle \begin{align*} \mathbf{g} \end{align*}$. This now reduces the system to two simpler matrix equations, as the coefficient matrices are diagonal. They are $\displaystyle \begin{align*} L\,\mathbf{g} = \mathbf{b} \end{align*}$ and $\displaystyle \begin{align*} U\,\mathbf{x} = \mathbf{g} \end{align*}$.

Solving for $\displaystyle \begin{align*} \mathbf{g} \end{align*}$ we have

$\displaystyle \begin{align*} \left[ \begin{matrix} \phantom{-}1 & 0 & \phantom{-}0 & 0 \\ \phantom{-}1 & 1 & \phantom{-}0 & 0 \\ -1 & \frac{3}{2} & \phantom{-}1 & 0 \\ \phantom{-}0 & 2 & -1 & 1 \end{matrix} \right] \left[ \begin{matrix} g_1 \\ g_2 \\ g_3 \\ g_4 \end{matrix} \right] = \left[ \begin{matrix} \phantom{-}7 \\ \phantom{-}3 \\ -7 \\ -12 \end{matrix} \right] \end{align*}$

We can see $\displaystyle \begin{align*} g_1 = 7 \end{align*}$. Forward substituting gives

$\displaystyle \begin{align*} g_1 + g_2 &= 3 \\ 7 + g_2 &= 3 \\ g_2 &= -4 \end{align*}$

Forward substituting gives

$\displaystyle \begin{align*} -g_1 + \frac{3}{2}\,g_2 + g_3 &= -7 \\ -7 - 6 + g_3 &= -7 \\ g_3 &= 6 \end{align*}$

Forward substituting gives

$\displaystyle \begin{align*} 2\,g_2 - g_3 + g_4 &= -12 \\ -8 - 6 + g_4 &= -12 \\ -14 + g_4 &= -12 \\ g_4 &= 2 \end{align*}$

Now solving $\displaystyle \begin{align*} U\,\mathbf{x} = \mathbf{g} \end{align*}$ for $\displaystyle \begin{align*} \mathbf{x} \end{align*}$ gives

$\displaystyle \begin{align*} \left[ \begin{matrix} 2 & 4 & \phantom{-}0 & \phantom{-}1 \\ 0 & 4 & -2 & \phantom{-}6 \\ 0 & 0 & \phantom{-}1 & -1 \\ 0 & 0 & \phantom{-}0 & -2 \end{matrix} \right] \left[ \begin{matrix} x_1 \\ x_2 \\ x_3 \\ x_4 \end{matrix} \right] = \left[ \begin{matrix} \phantom{-}7 \\ -4 \\ \phantom{-}6 \\ \phantom{-}2 \end{matrix} \right] \end{align*}$

We can see that

$\displaystyle \begin{align*} -2x_4 &= 2 \\ x_4 &= -1 \end{align*}$

Back substituting gives

$\displaystyle \begin{align*} x_3 - x_4 &= 6 \\ x_3 + 1 &= 6 \\ x_3 &= 5 \end{align*}$

Back substituting gives

$\displaystyle \begin{align*} 4\,x_2 - 2\,x_3 + 6\,x_4 &= -4 \\ 4\,x_2 - 10 - 6 &= -4 \\ 4\,x_2 &= 12 \\ x_2 &= 3 \end{align*}$

Back substituting gives

$\displaystyle \begin{align*} 2\,x_1 + 4\,x_2 + x_4 &= 7 \\ 2\,x_1 + 12 - 1 &= 7 \\ 2\,x_1 &= -4 \\ x_1 &= -2 \end{align*}$

So to answer your question we have $\displaystyle \begin{align*} g_1 = 7 , \, g_2 = -4 , \, g_3 = 6 , \, g_4 = 2, \, x_1 = -2 , \, x_2 = 3 , \, x_3 = 5 \end{align*}$ and $\displaystyle \begin{align*} x_4 = -1 \end{align*}$.
 

Attachments

  • gelu.jpg
    gelu.jpg
    67 KB · Views: 133
Physics news on Phys.org


Thank you for the detailed explanation, it was very helpful! It's interesting to see how the Gaussian Elimination process can be broken down into smaller, simpler equations.