Then A can be decomposed into a diagonal component D, a lower triangular part L and an upper triangular part U:
The solution is then obtained iteratively via
where is the kth approximation or iteration of and is the next or k + 1 iteration of . The element-based formula is thus:
The computation of requires each element in x(k) except itself. Unlike the Gauss–Seidel method, we can't overwrite with , as that value will be needed by the rest of the computation. The minimum amount of storage is two vectors of size n.
Algorithm
Input:initial guess to the solution, (diagonal dominant) matrix , right-hand side vector , convergence criterion
Output:solution when convergence is reachedComments:pseudocode based on the element-based formula abovewhile convergence not reached dofor i := 1 step until n dofor j := 1 step until n doif j ≠ i thenendendendend
Convergence
The standard convergence condition (for any iterative method) is when the spectral radius of the iteration matrix is less than 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:
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
Examples
Example 1
A linear system of the form with initial estimate is given by
We use the equation , described above, to estimate . First, we rewrite the equation in a more convenient form , where and . From the known values
we determine as
Further, is found as
With and calculated, we estimate as :
The next iteration yields
This process is repeated until convergence (i.e., until is small). The solution after 25 iterations is
Example 2
Suppose we are given the following linear system:
If we choose (0, 0, 0, 0) as the initial approximation, then the first approximate solution is given by
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.
0.6
2.27272
-1.1
1.875
1.04727
1.7159
-0.80522
0.88522
0.93263
2.05330
-1.0493
1.13088
1.01519
1.95369
-0.9681
0.97384
0.98899
2.0114
-1.0102
1.02135
The exact solution of the system is (1, 2, −1, 1).
Python example
importnumpyasnpITERATION_LIMIT=1000# initialize the matrixA=np.array([[10.,-1.,2.,0.],[-1.,11.,-1.,3.],[2.,-1.,10.,-1.],[0.0,3.,-1.,8.]])# initialize the RHS vectorb=np.array([6.,25.,-11.,15.])# prints the systemprint("System:")foriinrange(A.shape[0]):row=["{}*x{}".format(A[i,j],j+1)forjinrange(A.shape[1])]print(f'{" + ".join(row)} = {b[i]}')print()x=np.zeros_like(b)forit_countinrange(ITERATION_LIMIT):ifit_count!=0:print("Iteration {0}: {1}".format(it_count,x))x_new=np.zeros_like(x)foriinrange(A.shape[0]):s1=np.dot(A[i,:i],x[:i])s2=np.dot(A[i,i+1:],x[i+1:])x_new[i]=(b[i]-s1-s2)/A[i,i]ifx_new[i]==x_new[i-1]:breakifnp.allclose(x,x_new,atol=1e-10,rtol=0.):breakx=x_newprint("Solution: ")print(x)error=np.dot(A,x)-bprint("Error:")print(error)
Weighted Jacobi method
The weighted Jacobi iteration uses a parameter to compute the iteration as
with being the usual choice.[1]
From the relation , this may also be expressed as
.
Convergence in the symmetric positive definite case
In case that the system matrix is of symmetric positive-definite type one can show convergence.
Let be the iteration matrix.
Then, convergence is guaranteed for
where is the maximal eigenvalue.
The spectral radius can be minimized for a particular choice of as follows