Jacobi method · Lablypedia

Description

Let $A x = b$ be a square system of n linear equations, where: $A = [\begin{matrix} a_{11} & a_{12} & \dots & a_{1 n} \\ a_{21} & a_{22} & \dots & a_{2 n} \\ ⋮ & ⋮ & ⋱ & ⋮ \\ a_{n 1} & a_{n 2} & \dots & a_{n n} \end{matrix}], x = [\begin{matrix} x_{1} \\ x_{2} \\ ⋮ \\ x_{n} \end{matrix}], b = [\begin{matrix} b_{1} \\ b_{2} \\ ⋮ \\ b_{n} \end{matrix}] .$

When $A$ and $b$ are known, and $x$ is unknown, we can use the Jacobi method to approximate $x$ . The vector $x^{(0)}$ denotes our initial guess for $x$ (often $x_{i}^{(0)} = 0$ for $i = 1, 2, . . ., n$ ). We denote $x^{(k)}$ as the k-th approximation or iteration of $x$ , and $x^{(k + 1)}$ is the next (or k+1) iteration of $x$ .

Matrix-based formula

Then A can be decomposed into a diagonal component D, a lower triangular part L and an upper triangular part U: $A = D + L + U where D = [\begin{matrix} a_{11} & 0 & \dots & 0 \\ 0 & a_{22} & \dots & 0 \\ ⋮ & ⋮ & ⋱ & ⋮ \\ 0 & 0 & \dots & a_{n n} \end{matrix}] and L + U = [\begin{matrix} 0 & a_{12} & \dots & a_{1 n} \\ a_{21} & 0 & \dots & a_{2 n} \\ ⋮ & ⋮ & ⋱ & ⋮ \\ a_{n 1} & a_{n 2} & \dots & 0 \end{matrix}] .$ The solution is then obtained iteratively via

x^{(k + 1)} = D^{- 1} (b - (L + U) x^{(k)}) .

Element-based formula

The element-based formula for each row $i$ is thus: $x_{i}^{(k + 1)} = \frac{1}{a_{i i}} (b_{i} - \sum_{j \neq i} a_{i j} x_{j}^{(k)}), i = 1, 2, \dots, n .$ The computation of $x_{i}^{(k + 1)}$ requires each element in $x^{(k)}$ except itself. Unlike the Gauss–Seidel method, we cannot overwrite $x_{i}^{(k)}$ with $x_{i}^{(k + 1)}$ , as that value will be needed by the rest of the computation. The minimum amount of storage is two vectors of size n.

4 more chapters are free with a Lably account.

Algorithm

<b id="mwOw">Input:</b> <span class="nowrap" about="#mwt28" typeof="mw:Transclusion" data-mw='{"parts":[{"template":{"target":{"wt":"nowrap","href":"./Template:Nowrap"},"params":{"1":{"wt":"initial guess ''x''<sup>(0)</sup> to the solution"}},"i":0}}]}' id="mwPA">initial guess <i>x</i><sup>(0)</sup> to the solution</span>, (diagonal dominant) matrix <i id="mwPQ">A</i>, right-hand side vector <i id="mwPg">b</i>, convergence criterion <b id="mwPw">Output:</b> <span class="nowrap" about="#mwt29" typeof="mw:Transclusion" data-mw='{"parts":[{"template":{"target":{"wt":"nowrap","href":"./Template:Nowrap"},"params":{"1":{"wt":"solution when convergence is reached"}},"i":0}}]}' id="mwQA">solution when convergence is reached</span> <b id="mwQQ">Comments:</b> pseudocode based on the element-based formula above <span class="nowrap" about="#mwt30" typeof="mw:Transclusion" data-mw='{"parts":[{"template":{"target":{"wt":"nowrap","href":"./Template:Nowrap"},"params":{"1":{"wt":"''k'' = 0"}},"i":0}}]}' id="mwQg"><i>k</i> = 0</span> <b id="mwQw">while</b> convergence not reached <b id="mwRA">do</b> <b id="mwRQ">for</b> <i id="mwRg">i</i> := 1 <b id="mwRw">step until</b> n <b id="mwSA">do</b> <span class="nowrap" about="#mwt31" typeof="mw:Transclusion" data-mw='{"parts":[{"template":{"target":{"wt":"nowrap","href":"./Template:Nowrap"},"params":{"1":{"wt":"''σ'' = 0"}},"i":0}}]}' id="mwSQ"><i>σ</i> = 0</span> <b id="mwSg">for</b> <i id="mwSw">j</i> := 1 <b id="mwTA">step until</b> n <b id="mwTQ">do</b> <b id="mwTg">if</b> <i id="mwTw">j</i> <span typeof="mw:Entity" id="mwUA">≠</span> <i id="mwUQ">i</i> <b id="mwUg">then</b> <span class="nowrap" about="#mwt32" typeof="mw:Transclusion" data-mw="{"parts":[{"template":{"target":{"wt":"nowrap","href":"./Template:Nowrap"},"params":{"1":{"wt":"''σ'' = ''σ'' + ''a''<sub>''ij''</sub> ''x''<sub>''j''</sub><sup>(''k'')</sup>"}},"i":0}}]}" id="mwUw"><i>σ</i> = <i>σ</i> + <i>a</i><sub><i>ij</i></sub> <i>x</i><sub><i>j</i></sub><sup>(<i>k</i>)</sup></span> <b id="mwVA">end</b> <b id="mwVQ">end</b> <span class="nowrap" about="#mwt33" typeof="mw:Transclusion" data-mw="{"parts":[{"template":{"target":{"wt":"nowrap","href":"./Template:Nowrap"},"params":{"1":{"wt":"''x''<sub>''i''</sub><sup>(''k''+1)</sup> = (''b''<sub>''i''</sub> − ''σ'') / ''a''<sub>''ii''</sub>"}},"i":0}}]}" id="mwVg"><i>x</i><sub><i>i</i></sub><sup>(<i>k</i>+1)</sup> = (<i>b</i><sub><i>i</i></sub> − <i>σ</i>) / <i>a</i><sub><i>ii</i></sub></span> <b id="mwVw">end</b> increment <i id="mwWA">k</i> <b id="mwWQ">end</b>

Convergence

The standard convergence condition (for any iterative method) is when the spectral radius of the iteration matrix is less than 1:

ρ (D^{- 1} (L + U)) < 1.

A sufficient (but not necessary) condition for the method to converge is that the matrix A is strictly or irreducibly diagonally dominant. Strict row diagonal dominance means that for each row, the absolute value of the diagonal term is greater than the sum of absolute values of other terms:

| a_{i i} | > \sum_{j \neq i} | a_{i j} | .

The Jacobi method sometimes converges even if these conditions are not satisfied.

Note that the Jacobi method does not converge for every symmetric positive-definite matrix. For example, $A = (\begin{matrix} 29 & 2 & 1 \\ 2 & 6 & 1 \\ 1 & 1 & \frac{1}{5} \end{matrix}) \Rightarrow D^{- 1} (L + U) = (\begin{matrix} 0 & \frac{2}{29} & \frac{1}{29} \\ \frac{1}{3} & 0 & \frac{1}{6} \\ 5 & 5 & 0 \end{matrix}) \Rightarrow ρ (D^{- 1} (L + U)) \approx 1.0661 .$

Examples

Example question

A linear system of the form $A x = b$ with initial estimate $x^{(0)}$ is given by

A = [\begin{matrix} 2 & 1 \\ 5 & 7 \end{matrix}], b = [\begin{matrix} 11 \\ 13 \end{matrix}] and x^{(0)} = [\begin{matrix} 1 \\ 1 \end{matrix}] .

We use the equation $x^{(k + 1)} = D^{- 1} (b - (L + U) x^{(k)})$ , described above, to estimate $x$ . First, we rewrite the equation in a more convenient form $D^{- 1} (b - (L + U) x^{(k)}) = T x^{(k)} + C$ , where $T = - D^{- 1} (L + U)$ and $C = D^{- 1} b$ . From the known values $D^{- 1} = [\begin{matrix} 1 / 2 & 0 \\ 0 & 1 / 7 \end{matrix}], L = [\begin{matrix} 0 & 0 \\ 5 & 0 \end{matrix}] and U = [\begin{matrix} 0 & 1 \\ 0 & 0 \end{matrix}] .$ we determine $T = - D^{- 1} (L + U)$ as $T = [\begin{matrix} 1 / 2 & 0 \\ 0 & 1 / 7 \end{matrix}] {[\begin{matrix} 0 & 0 \\ - 5 & 0 \end{matrix}] + [\begin{matrix} 0 & - 1 \\ 0 & 0 \end{matrix}]} = [\begin{matrix} 0 & - 1 / 2 \\ - 5 / 7 & 0 \end{matrix}] .$ Further, $C$ is found as $C = [\begin{matrix} 1 / 2 & 0 \\ 0 & 1 / 7 \end{matrix}] [\begin{matrix} 11 \\ 13 \end{matrix}] = [\begin{matrix} 11 / 2 \\ 13 / 7 \end{matrix}] .$ With $T$ and $C$ calculated, we estimate $x$ as $x^{(1)} = T x^{(0)} + C$ : $x^{(1)} = [\begin{matrix} 0 & - 1 / 2 \\ - 5 / 7 & 0 \end{matrix}] [\begin{matrix} 1 \\ 1 \end{matrix}] + [\begin{matrix} 11 / 2 \\ 13 / 7 \end{matrix}] = [\begin{matrix} 5.0 \\ 8 / 7 \end{matrix}] \approx [\begin{matrix} 5 \\ 1.143 \end{matrix}] .$ The next iteration yields $x^{(2)} = [\begin{matrix} 0 & - 1 / 2 \\ - 5 / 7 & 0 \end{matrix}] [\begin{matrix} 5.0 \\ 8 / 7 \end{matrix}] + [\begin{matrix} 11 / 2 \\ 13 / 7 \end{matrix}] = [\begin{matrix} 69 / 14 \\ - 12 / 7 \end{matrix}] \approx [\begin{matrix} 4.929 \\ - 1.714 \end{matrix}] .$ This process is repeated until convergence (i.e., until $‖ A x^{(n)} - b ‖$ is small). The solution after 25 iterations is

x = [\begin{matrix} 7.111 \\ - 3.222 \end{matrix}] .

Example question 2

Suppose we are given the following linear system:

\begin{matrix} 10 x_{1} - x_{2} + 2 x_{3} & = 6, \\ - x_{1} + 11 x_{2} - x_{3} + 3 x_{4} & = 25, \\ 2 x_{1} - x_{2} + 10 x_{3} - x_{4} & = - 11, \\ 3 x_{2} - x_{3} + 8 x_{4} & = 15. \end{matrix}

If we choose (0, 0, 0, 0) as the initial approximation, then the first approximate solution is given by $\begin{matrix} x_{1} & = (6 + 0 - (2 * 0)) / 10 = 0.6, \\ x_{2} & = (25 + 0 + 0 - (3 * 0)) / 11 = 25 / 11 = 2.2727, \\ x_{3} & = (- 11 - (2 * 0) + 0 + 0) / 10 = - 1.1, \\ x_{4} & = (15 - (3 * 0) + 0) / 8 = 1.875. \end{matrix}$ Using the approximations obtained, the iterative procedure is repeated until the desired accuracy has been reached. The following are the approximated solutions after five iterations.

The exact solution of the system is (1, 2, −1, 1).

Python example

<span id="mwuQ"></span><span class="linenos" data-line="1" id="mwug"></span><span class="kn" id="mwuw">import</span><span class="w" id="mwvA"> </span><span class="nn" id="mwvQ">numpy</span><span class="w" id="mwvg"> </span><span class="k" id="mwvw">as</span><span class="w" id="mwwA"> </span><span class="nn" id="mwwQ">np</span> <span class="linenos" data-line="2" id="mwwg"></span> <span class="linenos" data-line="3" id="mwww"></span><span class="n" id="mwxA">ITERATION_LIMIT</span> <span class="o" id="mwxQ">=</span> <span class="mi" id="mwxg">1000</span> <span class="linenos" data-line="4" id="mwxw"></span> <span class="linenos" data-line="5" id="mwyA"></span><span class="c1" id="mwyQ"># initialize the matrix</span> <span class="linenos" data-line="6" id="mwyg"></span><span class="n" id="mwyw">A</span> <span class="o" id="mwzA">=</span> <span class="n" id="mwzQ">np</span><span class="o" id="mwzg">.</span><span class="kp" id="mwzw">array</span><span class="p" id="mw0A">([[</span><span class="mf" id="mw0Q">10.</span><span class="p" id="mw0g">,</span> <span class="o" id="mw0w">-</span><span class="mf" id="mw1A">1.</span><span class="p" id="mw1Q">,</span> <span class="mf" id="mw1g">2.</span><span class="p" id="mw1w">,</span> <span class="mf" id="mw2A">0.</span><span class="p" id="mw2Q">],</span> <span class="linenos" data-line="7" id="mw2g"></span> <span class="p" id="mw2w">[</span><span class="o" id="mw3A">-</span><span class="mf" id="mw3Q">1.</span><span class="p" id="mw3g">,</span> <span class="mf" id="mw3w">11.</span><span class="p" id="mw4A">,</span> <span class="o" id="mw4Q">-</span><span class="mf" id="mw4g">1.</span><span class="p" id="mw4w">,</span> <span class="mf" id="mw5A">3.</span><span class="p" id="mw5Q">],</span> <span class="linenos" data-line="8" id="mw5g"></span> <span class="p" id="mw5w">[</span><span class="mf" id="mw6A">2.</span><span class="p" id="mw6Q">,</span> <span class="o" id="mw6g">-</span><span class="mf" id="mw6w">1.</span><span class="p" id="mw7A">,</span> <span class="mf" id="mw7Q">10.</span><span class="p" id="mw7g">,</span> <span class="o" id="mw7w">-</span><span class="mf" id="mw8A">1.</span><span class="p" id="mw8Q">],</span> <span class="linenos" data-line="9" id="mw8g"></span> <span class="p" id="mw8w">[</span><span class="mf" id="mw9A">0.0</span><span class="p" id="mw9Q">,</span> <span class="mf" id="mw9g">3.</span><span class="p" id="mw9w">,</span> <span class="o" id="mw-A">-</span><span class="mf" id="mw-Q">1.</span><span class="p" id="mw-g">,</span> <span class="mf" id="mw-w">8.</span><span class="p" id="mw_A">]])</span> <span class="linenos" data-line="10" id="mw_Q"></span><span class="c1" id="mw_g"># initialize the RHS vector</span> <span class="linenos" data-line="11" id="mw_w"></span><span class="n" id="mwAQA">b</span> <span class="o" id="mwAQE">=</span> <span class="n" id="mwAQI">np</span><span class="o" id="mwAQM">.</span><span class="kp" id="mwAQQ">array</span><span class="p" id="mwAQU">([</span><span class="mf" id="mwAQY">6.</span><span class="p" id="mwAQc">,</span> <span class="mf" id="mwAQg">25.</span><span class="p" id="mwAQk">,</span> <span class="o" id="mwAQo">-</span><span class="mf" id="mwAQs">11.</span><span class="p" id="mwAQw">,</span> <span class="mf" id="mwAQ0">15.</span><span class="p" id="mwAQ4">])</span> <span class="linenos" data-line="12" id="mwAQ8"></span> <span class="linenos" data-line="13" id="mwARA"></span><span class="c1" id="mwARE"># prints the system</span> <span class="linenos" data-line="14" id="mwARI"></span><span class="nb" id="mwARM">print</span><span class="p" id="mwARQ">(</span><span class="s2" id="mwARU">"System:"</span><span class="p" id="mwARY">)</span> <span class="linenos" data-line="15" id="mwARc"></span><span class="k" id="mwARg">for</span> <span class="n" id="mwARk">i</span> <span class="ow" id="mwARo">in</span> <span class="nb" id="mwARs">range</span><span class="p" id="mwARw">(</span><span class="n" id="mwAR0">A</span><span class="o" id="mwAR4">.</span><span class="kp" id="mwAR8">shape</span><span class="p" id="mwASA">[</span><span class="mi" id="mwASE">0</span><span class="p" id="mwASI">]):</span> <span class="linenos" data-line="16" id="mwASM"></span> <span class="n" id="mwASQ">row</span> <span class="o" id="mwASU">=</span> <span class="p" id="mwASY">[</span><span class="sa" id="mwASc">f</span><span class="s2" id="mwASg">"</span><span class="si" id="mwASk">{</span><span class="n" id="mwASo">A</span><span class="p" id="mwASs">[</span><span class="n" id="mwASw">i</span><span class="p" id="mwAS0">,</span><span class="w" id="mwAS4"> </span><span class="n" id="mwAS8">j</span><span class="p" id="mwATA">]</span><span class="si" id="mwATE">}</span><span class="s2" id="mwATI">*x</span><span class="si" id="mwATM">{</span><span class="n" id="mwATQ">j</span><span class="w" id="mwATU"> </span><span class="o" id="mwATY">+</span><span class="w" id="mwATc"> </span><span class="mi" id="mwATg">1</span><span class="si" id="mwATk">}</span><span class="s2" id="mwATo">"</span> <span class="k" id="mwATs">for</span> <span class="n" id="mwATw">j</span> <span class="ow" id="mwAT0">in</span> <span class="nb" id="mwAT4">range</span><span class="p" id="mwAT8">(</span><span class="n" id="mwAUA">A</span><span class="o" id="mwAUE">.</span><span class="kp" id="mwAUI">shape</span><span class="p" id="mwAUM">[</span><span class="mi" id="mwAUQ">1</span><span class="p" id="mwAUU">])]</span> <span class="linenos" data-line="17" id="mwAUY"></span> <span class="nb" id="mwAUc">print</span><span class="p" id="mwAUg">(</span><span class="sa" id="mwAUk">f</span><span class="s1" id="mwAUo">'</span><span class="si" id="mwAUs">{</span><span class="s2" id="mwAUw">" + "</span><span class="o" id="mwAU0">.</span><span class="n" id="mwAU4">join</span><span class="p" id="mwAU8">(</span><span class="n" id="mwAVA">row</span><span class="p" id="mwAVE">)</span><span class="si" id="mwAVI">}</span><span class="s1" id="mwAVM"> = </span><span class="si" id="mwAVQ">{</span><span class="n" id="mwAVU">b</span><span class="p" id="mwAVY">[</span><span class="n" id="mwAVc">i</span><span class="p" id="mwAVg">]</span><span class="si" id="mwAVk">}</span><span class="s1" id="mwAVo">'</span><span class="p" id="mwAVs">)</span> <span class="linenos" data-line="18" id="mwAVw"></span><span class="nb" id="mwAV0">print</span><span class="p" id="mwAV4">()</span> <span class="linenos" data-line="19" id="mwAV8"></span> <span class="linenos" data-line="20" id="mwAWA"></span><span class="n" id="mwAWE">x</span> <span class="o" id="mwAWI">=</span> <span class="n" id="mwAWM">np</span><span class="o" id="mwAWQ">.</span><span class="kp" id="mwAWU">zeros_like</span><span class="p" id="mwAWY">(</span><span class="n" id="mwAWc">b</span><span class="p" id="mwAWg">)</span> <span class="linenos" data-line="21" id="mwAWk"></span><span class="k" id="mwAWo">for</span> <span class="n" id="mwAWs">it_count</span> <span class="ow" id="mwAWw">in</span> <span class="nb" id="mwAW0">range</span><span class="p" id="mwAW4">(</span><span class="n" id="mwAW8">ITERATION_LIMIT</span><span class="p" id="mwAXA">):</span> <span class="linenos" data-line="22" id="mwAXE"></span> <span class="k" id="mwAXI">if</span> <span class="n" id="mwAXM">it_count</span> <span class="o" id="mwAXQ">!=</span> <span class="mi" id="mwAXU">0</span><span class="p" id="mwAXY">:</span> <span class="linenos" data-line="23" id="mwAXc"></span> <span class="nb" id="mwAXg">print</span><span class="p" id="mwAXk">(</span><span class="sa" id="mwAXo">f</span><span class="s2" id="mwAXs">"Iteration </span><span class="si" id="mwAXw">{</span><span class="n" id="mwAX0">it_count</span><span class="si" id="mwAX4">}</span><span class="s2" id="mwAX8">: </span><span class="si" id="mwAYA">{</span><span class="n" id="mwAYE">x</span><span class="si" id="mwAYI">}</span><span class="s2" id="mwAYM">"</span><span class="p" id="mwAYQ">)</span> <span class="linenos" data-line="24" id="mwAYU"></span> <span class="n" id="mwAYY">x_new</span> <span class="o" id="mwAYc">=</span> <span class="n" id="mwAYg">np</span><span class="o" id="mwAYk">.</span><span class="kp" id="mwAYo">zeros_like</span><span class="p" id="mwAYs">(</span><span class="n" id="mwAYw">x</span><span class="p" id="mwAY0">)</span> <span class="linenos" data-line="25" id="mwAY4"></span> <span class="linenos" data-line="26" id="mwAY8"></span> <span class="k" id="mwAZA">for</span> <span class="n" id="mwAZE">i</span> <span class="ow" id="mwAZI">in</span> <span class="nb" id="mwAZM">range</span><span class="p" id="mwAZQ">(</span><span class="n" id="mwAZU">A</span><span class="o" id="mwAZY">.</span><span class="kp" id="mwAZc">shape</span><span class="p" id="mwAZg">[</span><span class="mi" id="mwAZk">0</span><span class="p" id="mwAZo">]):</span> <span class="linenos" data-line="27" id="mwAZs"></span> <span class="n" id="mwAZw">s1</span> <span class="o" id="mwAZ0">=</span> <span class="n" id="mwAZ4">np</span><span class="o" id="mwAZ8">.</span><span class="kp" id="mwAaA">dot</span><span class="p" id="mwAaE">(</span><span class="n" id="mwAaI">A</span><span class="p" id="mwAaM">[</span><span class="n" id="mwAaQ">i</span><span class="p" id="mwAaU">,</span> <span class="p" id="mwAaY">:</span><span class="n" id="mwAac">i</span><span class="p" id="mwAag">],</span> <span class="n" id="mwAak">x</span><span class="p" id="mwAao">[:</span><span class="n" id="mwAas">i</span><span class="p" id="mwAaw">])</span> <span class="linenos" data-line="28" id="mwAa0"></span> <span class="n" id="mwAa4">s2</span> <span class="o" id="mwAa8">=</span> <span class="n" id="mwAbA">np</span><span class="o" id="mwAbE">.</span><span class="kp" id="mwAbI">dot</span><span class="p" id="mwAbM">(</span><span class="n" id="mwAbQ">A</span><span class="p" id="mwAbU">[</span><span class="n" id="mwAbY">i</span><span class="p" id="mwAbc">,</span> <span class="n" id="mwAbg">i</span> <span class="o" id="mwAbk">+</span> <span class="mi" id="mwAbo">1</span><span class="p" id="mwAbs">:],</span> <span class="n" id="mwAbw">x</span><span class="p" id="mwAb0">[</span><span class="n" id="mwAb4">i</span> <span class="o" id="mwAb8">+</span> <span class="mi" id="mwAcA">1</span><span class="p" id="mwAcE">:])</span> <span class="linenos" data-line="29" id="mwAcI"></span> <span class="n" id="mwAcM">x_new</span><span class="p" id="mwAcQ">[</span><span class="n" id="mwAcU">i</span><span class="p" id="mwAcY">]</span> <span class="o" id="mwAcc">=</span> <span class="p" id="mwAcg">(</span><span class="n" id="mwAck">b</span><span class="p" id="mwAco">[</span><span class="n" id="mwAcs">i</span><span class="p" id="mwAcw">]</span> <span class="o" id="mwAc0">-</span> <span class="n" id="mwAc4">s1</span> <span class="o" id="mwAc8">-</span> <span class="n" id="mwAdA">s2</span><span class="p" id="mwAdE">)</span> <span class="o" id="mwAdI">/</span> <span class="n" id="mwAdM">A</span><span class="p" id="mwAdQ">[</span><span class="n" id="mwAdU">i</span><span class="p" id="mwAdY">,</span> <span class="n" id="mwAdc">i</span><span class="p" id="mwAdg">]</span> <span class="linenos" data-line="30" id="mwAdk"></span> <span class="k" id="mwAdo">if</span> <span class="n" id="mwAds">x_new</span><span class="p" id="mwAdw">[</span><span class="n" id="mwAd0">i</span><span class="p" id="mwAd4">]</span> <span class="o" id="mwAd8">==</span> <span class="n" id="mwAeA">x_new</span><span class="p" id="mwAeE">[</span><span class="n" id="mwAeI">i</span><span class="o" id="mwAeM">-</span><span class="mi" id="mwAeQ">1</span><span class="p" id="mwAeU">]:</span> <span class="linenos" data-line="31" id="mwAeY"></span> <span class="k" id="mwAec">break</span> <span class="linenos" data-line="32" id="mwAeg"></span> <span class="linenos" data-line="33" id="mwAek"></span> <span class="k" id="mwAeo">if</span> <span class="n" id="mwAes">np</span><span class="o" id="mwAew">.</span><span class="kp" id="mwAe0">allclose</span><span class="p" id="mwAe4">(</span><span class="n" id="mwAe8">x</span><span class="p" id="mwAfA">,</span> <span class="n" id="mwAfE">x_new</span><span class="p" id="mwAfI">,</span> <span class="n" id="mwAfM">atol</span><span class="o" id="mwAfQ">=</span><span class="mf" id="mwAfU">1e-10</span><span class="p" id="mwAfY">,</span> <span class="n" id="mwAfc">rtol</span><span class="o" id="mwAfg">=</span><span class="mf" id="mwAfk">0.</span><span class="p" id="mwAfo">):</span> <span class="linenos" data-line="34" id="mwAfs"></span> <span class="k" id="mwAfw">break</span> <span class="linenos" data-line="35" id="mwAf0"></span> <span class="linenos" data-line="36" id="mwAf4"></span> <span class="n" id="mwAf8">x</span> <span class="o" id="mwAgA">=</span> <span class="n" id="mwAgE">x_new</span> <span class="linenos" data-line="37" id="mwAgI"></span> <span class="linenos" data-line="38" id="mwAgM"></span><span class="nb" id="mwAgQ">print</span><span class="p" id="mwAgU">(</span><span class="s2" id="mwAgY">"Solution: "</span><span class="p" id="mwAgc">)</span> <span class="linenos" data-line="39" id="mwAgg"></span><span class="nb" id="mwAgk">print</span><span class="p" id="mwAgo">(</span><span class="n" id="mwAgs">x</span><span class="p" id="mwAgw">)</span> <span class="linenos" data-line="40" id="mwAg0"></span><span class="n" id="mwAg4">error</span> <span class="o" id="mwAg8">=</span> <span class="n" id="mwAhA">np</span><span class="o" id="mwAhE">.</span><span class="kp" id="mwAhI">dot</span><span class="p" id="mwAhM">(</span><span class="n" id="mwAhQ">A</span><span class="p" id="mwAhU">,</span> <span class="n" id="mwAhY">x</span><span class="p" id="mwAhc">)</span> <span class="o" id="mwAhg">-</span> <span class="n" id="mwAhk">b</span> <span class="linenos" data-line="41" id="mwAho"></span><span class="nb" id="mwAhs">print</span><span class="p" id="mwAhw">(</span><span class="s2" id="mwAh0">"Error:"</span><span class="p" id="mwAh4">)</span> <span class="linenos" data-line="42" id="mwAh8"></span><span class="nb" id="mwAiA">print</span><span class="p" id="mwAiE">(</span><span class="n" id="mwAiI">error</span><span class="p" id="mwAiM">)</span>

Weighted Jacobi method

The weighted Jacobi iteration uses a parameter $ω$ to compute the iteration as

x^{(k + 1)} = ω D^{- 1} (b - (L + U) x^{(k)}) + (1 - ω) x^{(k)}

with $ω = 2 / 3$ being the usual choice. From the relation $L + U = A - D$ , this may also be expressed as

\begin{matrix} x^{(k + 1)} & = ω D^{- 1} b + (I - ω D^{- 1} A) x^{(k)} \\ = x^{(k)} + ω D^{- 1} r^{(k)}, \end{matrix}

where $r^{(k)} = b - A x^{(k)}$ is the algebraic residual at iteration $k$ .

Convergence in the symmetric positive definite case

If the system matrix $A$ is symmetric positive-definite, one can show convergence.

Let $C = C_{ω} = I - ω D^{- 1} A$ be the iteration matrix. Then, convergence is guaranteed for

ρ (C_{ω}) < 1 ⟺ 0 < ω < \frac{2}{λ_{max} (D^{- 1} A)},

where $λ_{max}$ is the maximal eigenvalue.

The spectral radius can be minimized for a particular choice of $ω = ω_{opt}$ as follows $min_{ω} ρ (C_{ω}) = ρ (C_{ω_{opt}}) = 1 - \frac{2}{κ (D^{- 1} A) + 1} for ω_{opt} := \frac{2}{λ_{min} (D^{- 1} A) + λ_{max} (D^{- 1} A)},$ where $κ$ is the matrix condition number.