The Museum of HP Calculators

HP Forum Archive 18

 HP 50g finding the nth derivativeMessage #1 Posted by Jonathan Vogel on 8 Mar 2008, 4:05 p.m. Is there a way I can find the nth derivative of a function using the hp 50g?

 Re: HP 50g finding the nth derivativeMessage #2 Posted by Mark W Paris on 10 Mar 2008, 10:56 a.m.,in response to message #1 by Jonathan Vogel Hi Jonathan -- I don't know if there's a key-sequence that'll do it for you, but I tend to doubt it. I checked the manual and didn't find any f^{(n)} function. You can write a short program using matrix functions to do it. Suppose you want the n-th derivative. The brute force way of doing this is to evaluate the function at n+1 points and take the finite differences dividing out (for n>0) some appropriate interval. Eg. f^{(1)}(x) = \frac{f(h)-f(-h)}{2h} (LaTeX form - should be pretty obvious) The error in this equation is O(h^2) since we've taken the symmetric diffrence. Now if we truncate the Taylor expansion for a function f(x) about a point f(x_0) at order n f(x)-f(x_0)=\sum_{m=1}^n f^{(m)}(x_0) x^m/m! we can write a linear system of equations for the derivatives f^{(m)} for m>0 in terms of the differences of the function evaluated at x_0 and multiples of some small number h as f^{(m)}(x_0) = \sum_{j=1}^n A^{-1}_{mj} [f_j-f(x_0)] where f_j = f(x=j*h) and A^{-1} is the inverse of the matrix A_{jm} = (j*h)^m/m! The method is accurate to O(h^(n-j)) for the j-th derivative. There are some tricks you can play to improve the accuracy of the algorithm, like taking symmetric intervals about x_0, ie. x=+/-h, x=+/-2h, x=+/-3h, ... but the idea is the same. Of course, as always, you should check any math people tell you yourself.

 Re: HP 50g finding the nth derivativeMessage #3 Posted by Marcus von Cube, Germany on 10 Mar 2008, 12:34 p.m.,in response to message #2 by Mark W Paris Why not do it symbolically in RPL? << -> fn var n << fn 1 n START var d NEXT EVAL >> >> 'dN' STO  It can be used on the stack or algebraically like this: 'dN(X^4,X,3)' EVAL  The latter leads to 24*X as a result. Marcus Edited: 10 Mar 2008, 4:13 p.m. after one or more responses were posted

 Re: HP 50g finding the nth derivativeMessage #4 Posted by Mark W Paris on 10 Mar 2008, 12:46 p.m.,in response to message #3 by Marcus von Cube, Germany Much nicer. Guess I should learn RPL.

Go back to the main exhibit hall