Recurrence Relation for Iterative Solution of Ax=b

In summary, the equation {_nx} + M_1^{-1}(b-A_nx) + M_2^{-1}(b-A_nx) - M_2^{-1}AM_1^{-1}(b-A_nx) can be simplified to {_nx} + M_1^{-1}(b-A_nx) + M_2^{-1}(b-A_nx) - M_2^{-1}AM_1^{-1}(b-A_nx) by using the equations above.
  • #1
Kidnapster
2
0
[itex] _nx [/itex] represents the nth iteration.

Homework Statement



Show that [itex] {_{n+2}}x = _nx + M^{-1}(b-A{_nx}) [/itex]

Homework Equations



[itex] {_{n+1}}x = _nx + M_1^{-1}(b-A{_nx}) [/itex]
[itex] {_{n+2}}x = {_{n+1}}x + M_2^{-1}(b-A{_{n+1}x}) [/itex]
[itex] M = M_1(M_1+M_2-A)^{-1}M_2 [/itex]

The Attempt at a Solution



I tried was adding the two equations, which gives [itex] _{n+2}x = M_2^{-1}(b-A{_{n+1}x}) + _nx + M_1^{-1}(b-A{_nx}) [/itex]. But [itex] M_2^{-1}(b-A{_{n+1}}x) [/itex] devolves into a huge mess once I expand it. Subtraction is less helpful, as you are left without any [itex] _nx [/itex]. Directly substituting for [itex]_{n+1}x [/itex] in the second equation is even less helpful, as you are left with a huge mess of terms that can't really be factored. What am I missing?
 
Physics news on Phys.org
  • #2
Never mind, figured it out. Here's my solution:

[tex]
\begin{align*}
{_{n+1}x} &= {_nx} + M_1^{-1}(b-A_nx)\\
{_{n+2}x} &= {_{n+1}x} + M_1^{-1}(b-A_{n+1}x) \\
{_{n+2}x} &= {_nx} + M_1^{-1}(b-A_nx) + M_2^{-1}(b-A[{_nx} + M_1^{-1}(b-A_nx)])\\
{_{n+2}x} &= {_nx} + M_1^{-1}b - M_1^{-1}A_nx + M_2^{-1}b-M_2^{-1}A_nx - M_2^{-1}AM_1^{-1}b + M_2^{-1}AM_1^{-1}A_nx \\
{_{n+2}x} &= {_nx} + M_1^{-1}(b-A_nx) + M_2^{-1}(b-A_nx) - M_2^{-1}AM_1^{-1}(b-A_nx) \\
{_{n+2}x} &= {_nx} + M_1^{-1}(M_1+M_2-A)M_2^{-1}(b-A_nx)
\end{align*}

[/tex]
 

FAQ: Recurrence Relation for Iterative Solution of Ax=b

1. What is a recurrence relation in the context of solving Ax=b iteratively?

A recurrence relation is a mathematical formula that expresses the next term in a sequence as a function of the previous terms. In the context of solving Ax=b iteratively, it is a formula that defines the relationship between a current solution and the next solution in the iterative process.

2. How is a recurrence relation used to solve Ax=b iteratively?

A recurrence relation is used to calculate the next solution in the iterative process by substituting the previous solution into the formula. This process is repeated until a desired level of accuracy is achieved.

3. What are the advantages of using a recurrence relation for iterative solutions of Ax=b?

Using a recurrence relation allows for a more efficient and accurate solution of Ax=b compared to other methods, such as direct methods. It also allows for the use of parallel computing, which can further improve the speed of the solution.

4. Are there any limitations to using a recurrence relation for iterative solutions of Ax=b?

One limitation is that the convergence of the solution may depend on the choice of initial guess. Additionally, if the matrix A is ill-conditioned, the recurrence relation may not converge or may converge slowly.

5. How do you determine the optimal number of iterations to use with a recurrence relation for solving Ax=b?

The optimal number of iterations can be determined by evaluating the convergence rate of the recurrence relation, which can be influenced by the properties of the matrix A and the choice of initial guess. It is important to balance the number of iterations with the desired level of accuracy in order to achieve an efficient solution.

Back
Top