- — -

> Basicly, i have the values f(x+nh,y+nh), with n an

> integer in some range (for example [0,99]) and would

> like to approximate both of the partial derivatives

> in each of those points. After some googling i found

> some 1-dimensional method and was able to adapt that

> to the 2-dimensions using the R^2 analogy of the

> Taylor polynomial.

> Due to a technical limitation, i cannot use the whole

> set of points for approximation. Instead, all i have

> is a 3 by 3 grid of points centered at the point of

> interest. The following picture should make it

> absolutely clear:

> YYY

> YXY

> YYY

> The partial derivatives are to be calculated for the

> point marked with X. The values of the function are

> available for each of these 9 points.

> Using 1st degree taylor polynomials expanded around

> the middle point i can produce formulas that are a

> linear combination of the values divided by some

> constant. But I have no idea if my approach is

> appropriate for the situation, and if it is, which

> points should i use?

What’s the underlying partial differential equation

you are trying to solve ?

Best wishes

Torsten.