Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I'm trying to use Newton's method to solve the following system of equations, where f and g are functions of x and y. (h,a,f,c,d,b and k are just constants).

$f(y,x)=\left[\begin{array}{c} y^{1}\\ y^{2}\\ y^{3}\\ y^{4} \end{array}\right]-\left[\begin{array}{c} y_{0}^{1}\\ y_{0}^{2}\\ y_{0}^{3}\\ y_{0}^{4} \end{array}\right]+h\left[\begin{array}{cccc} a_{1} & f_{1} & 0 & 0\\ c_{1} & a_{1} & f_{1} & 0\\ 0 & c_{1} & a_{1} & f_{1}\\ 0 & 0 & c_{1} & a_{1} \end{array}\right]\left[\begin{array}{c} y^{1}\\ y^{2}\\ y^{3}\\ y^{4} \end{array}\right]+\left[\begin{array}{cccc} b_{1} & 0 & 0 & 0\\ d_{1} & b_{1} & 0 & 0\\ 0 & d_{1} & b_{1} & 0\\ 0 & 0 & d_{1} & b_{1} \end{array}\right]\left[\begin{array}{c} x^{1}\\ x^{2}\\ x^{3}\\ x^{4} \end{array}\right]-\left[\begin{array}{c} k_{1}\\ 0\\ 0\\ k_{2} \end{array}\right]=0$

$g(y,x)=\left[\begin{array}{c} x^{1}\\ x^{2}\\ x^{3}\\ x^{4} \end{array}\right]-\left[\begin{array}{c} x_{0}^{1}\\ x_{0}^{2}\\ x_{0}^{3}\\ x_{0}^{4} \end{array}\right]+h\left[\begin{array}{cccc} a_{2} & f_{2} & 0 & 0\\ c_{2} & a_{2} & f_{2} & 0\\ 0 & c_{2} & a_{2} & f_{2}\\ 0 & 0 & c_{2} & a_{2} \end{array}\right]\left[\begin{array}{c} y^{1}\\ y^{2}\\ y^{3}\\ y^{4} \end{array}\right]+\left[\begin{array}{cccc} b_{2} & 0 & 0 & 0\\ d_{2} & b_{2} & 0 & 0\\ 0 & d_{2} & b_{2} & 0\\ 0 & 0 & d_{2} & b_{2} \end{array}\right]\left[\begin{array}{c} x^{1}\\ x^{2}\\ x^{3}\\ x^{4} \end{array}\right]-\left[\begin{array}{c} k_{1}\\ 0\\ 0\\ k_{2} \end{array}\right]=0$

Am I right in saying that the Newton Iteration equations be:

$y^{k+1}=y^{k}-\frac{f(y,x)}{\frac{\partial f(y,x)}{\partial y}}$ and $x^{k+1}=x^{k}-\frac{g(y,x)}{\frac{\partial g(x,y)}{\partial x}}$.

Or would the denominators be partial derivatives: $\left(x\frac{\partial f}{\partial y}+y\frac{\partial f}{\partial x}\right)$.

I know it was not necessary to write the entire functions but I have some more questions about the equations depending on the answer to this one.

Many Thanks!

share|improve this question
    
I've just been thinking a lot about this and realized that I'm not even sure how I can apply Newton Raphson method to these equations. If applying it to the first equations will give x or y? ... Please help if have any ideas. –  Hooman Jan 19 '12 at 13:29
    
Note: It's very possible I'm horribly wrong. With that said: I've never seen Newton's method applied to functions that aren't single variable. mathworld.wolfram.com/NewtonsMethod.html Mathworld does not seem to recognize this approach, either. Sorry if I am wrong, but I think Newton's method does not apply to functions that are more than one variable. –  000 Jan 21 '12 at 5:29
    
There's a method called the Multivariate Newton Raphson Method (MNRM) which is used for functions of more than one variable. But my equations are a bit more complicated. So I'm not quite sure how this could be applied in this case. –  Hooman Jan 22 '12 at 12:26
    
I think I might know the answer now. Will post as soon as I am sure. –  Hooman Jan 22 '12 at 14:31

1 Answer 1

What you have is a matrix with two unknowns and four equations. This implies that there isn't necessarily a consistent solution. However, if you are trying to find the nearby local optima solution, you can use a computational method called gradient descent. It is a multivariable analog to Newton's Method. It works a finite (sometimes adaptive, depending on algorithm) step size.

The recurrent equation is: $$\vec x_{i+1}=\vec x_i - h*\vec \nabla f(\vec x)$$ $h$ is the step size, and is usually small. There are advanced techniques for computing the "optimum" step size in a given situation, but the techniques are often not worth the added calculation complexity. As the system gets close to a solution, the gradient often becomes unstable due to numerical conditioning. As a result, it tends to bounce around the actual solution in random directions. However, at this point, the solution found is usually close enough for practical purposes.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.