Solving a system on nonlinear equations using Halley's method
|
04-05-2014, 04:44 AM
(This post was last modified: 11-09-2015 09:23 AM by Namir.)
Post: #1
|
|||
|
|||
Solving a system on nonlinear equations using Halley's method
I have known for many years that Halley's method for solving single-variable nonlinear functions converged faster than Newton's method.
Early today I found the following 1985 article that discusses Halley's method for solving a system of nonlinear equations. Developing the matrix version of Halley from the regular version is not trivial. The article find is a gem! I wrote the following Matlab code (should be stored in file sne_halley.m) for the algorithm mentioned in the cited article. Code:
The code for test file fx1.m is: Code: function y = fx1(x) The code for test file fx2.m is: Code: function y = fx2(x) The code for test file fx3.m is: Code: function y = fx3(x) Sample run in Matlab is: [X, ~]=sne_halley(3, [4 2 3],1e-7,1e-8,1000,@fx1,@fx2,@fx3) You get: Code: % I think the Matlab code can be adapted to work on the HP-39gII, HP Prime, HP-71B with MATH ROM, HP-75C with MATH ROM, and the HP48S/SX/G/G+/Gx/49G/49G+/Gii/50 line of graphing calculators. The code for the nonlinear functions can be consolidated into a single program function that evaluate the different non-linear functions using an IF statement. Namir |
|||
04-05-2014, 07:46 AM
Post: #2
|
|||
|
|||
RE: Solving a system on nonlinear equations using Halley's method
(04-05-2014 04:44 AM)Namir Wrote: I think the Matlab code can be adapted to work on the HP Prime.This post probably fits better the forum HP Prime. Kind regards Thomas |
|||
06-06-2014, 01:48 AM
Post: #3
|
|||
|
|||
RE: Solving a system on nonlinear equations using Halley's method
Halley's method does converge faster (as do several others), but the domain of convergence is usually much smaller that that of Newton's method (which can be rather small itself.)
|
|||
06-06-2014, 03:02 AM
Post: #4
|
|||
|
|||
RE: Solving a system on nonlinear equations using Halley's method
Let me see if I am reading you correctly. Using Newton's method for nonlinear equation with multiple solution may not be (easily) able to converge to the solutions you are looking for? If this is what you mean, than I am in full agreement with you. HP Prime has problems converging to certain solutions of nonlinear equations. I recently wrote a similar Newton's method for 3 eqns on the HP-85 and I came across the same problem. This leads me to think if Newton's (and Halley's) methods are highly affected by the curvature of the nonlinear functions.
Yes? No? Namir |
|||
06-10-2014, 02:43 AM
Post: #5
|
|||
|
|||
RE: Solving a system on nonlinear equations using Halley's method
Both Newton's and Halley's methods (and Chebychev's and all the other higher order methods) strongly depend on the shape of the curves. The topic is usually covered in numerical analysis classes.
|
|||
06-12-2014, 01:39 PM
(This post was last modified: 06-12-2014 01:46 PM by Eddie W. Shore.)
Post: #6
|
|||
|
|||
RE: Solving a system on nonlinear equations using Halley's method
When I saw Halley, I immediately thought of Halley's Comet. It turns out that it is the same person who Halley's Comet is named after: Edmund Halley.
|
|||
« Next Oldest | Next Newest »
|
User(s) browsing this thread: