Post Reply 
Newton Method
05-12-2016, 09:17 PM
Post: #5
RE: Newton Method
(05-12-2016 08:47 PM)Namir Wrote:  A better way to calculate h is using something like:

h=0.001*(1+ABS(x))

Hmmm.... I do not think this is "better". Consider a function where the derivative changes rapidly near zero, e.g. sin(10^6 · x). Since the smallest possible h is 0,001 it may become much larger than x itself. That's why I prefer defining h as a certain fraction of x.

(05-12-2016 08:47 PM)Namir Wrote:  which does not require testing the value of x.

Well, that's done in two steps. And thus even shorter than "ABS 1 +". ;-)

Dieter
Find all posts by this user
Quote this message in a reply
Post Reply 


Messages In This Thread
Newton Method - bshoring - 05-11-2016, 07:54 PM
RE: Newton Method - Dieter - 05-11-2016, 08:40 PM
RE: Newton Method - Namir - 05-12-2016, 08:47 PM
RE: Newton Method - Dieter - 05-12-2016 09:17 PM
RE: Newton Method - Namir - 05-13-2016, 01:59 PM
RE: Newton Method - bshoring - 05-14-2016, 07:49 PM
RE: Newton Method - Dieter - 05-14-2016, 10:23 PM
RE: Newton Method - Namir - 05-15-2016, 02:00 AM
RE: Newton Method - Dieter - 05-15-2016, 01:11 PM
RE: Newton Method - Namir - 05-15-2016, 01:57 PM
RE: Newton Method - bshoring - 05-12-2016, 12:10 AM
RE: Newton Method - bshoring - 05-14-2016, 06:36 AM
RE: Newton Method - SlideRule - 08-24-2016, 12:08 PM
RE: Newton Method - Pekis - 08-24-2016, 12:20 PM
RE: Newton Method - Csaba Tizedes - 08-24-2016, 08:37 PM
RE: Newton Method - Duane Hess - 08-25-2016, 04:47 AM
RE: Newton Method - Namir - 08-25-2016, 05:00 AM
RE: Newton Method - rprosperi - 08-25-2016, 04:19 PM
RE: Newton Method - bshoring - 08-27-2016, 03:09 AM
RE: Newton Method - Duane Hess - 08-27-2016, 06:31 AM
RE: Newton Method - SlideRule - 08-28-2016, 02:57 AM
RE: Newton Method - rprosperi - 08-29-2016, 12:46 AM



User(s) browsing this thread: 6 Guest(s)