FindRoot NSolve RootSearch
- To: mathgroup at smc.vnet.net
- Subject: [mg43081] FindRoot NSolve RootSearch
- From: jankiew at op.pl
- Date: Mon, 11 Aug 2003 02:15:58 -0400 (EDT)
- Sender: owner-wri-mathgroup at wolfram.com
Hi everyone,
I have a following problem
The equation below contains 3 parameters x1[n], x2[n], x3[n], which are random real positive numbers less then one, and n goes from 1 till 10000. I'm looking for real and positive solutions p in a huge range.
So my final answer consists of many thousands of solutions.
The range is so big because there is another positive and real parameter "a" which can change from zero till let's say 5.
And now the confusion starts:
When I'm using NSolve (it works for only few a's), or FindRoot, or RootSearch, I have different answers.
FindRoot always complains about the precission issues, but increasing of it doesn't help, also choosing better starting values (in a smaller region) didn't work.
RootSearch works perfectly (there are no complains), but it's not original function form Wolfram, so I don't know if I can trust it.
So can I trust it? Or is there any other way to deal with precision issues of FindRoot?
RootSearch[
SetPrecision[
(9.4*10^8 + 1.4*10^8)^2 -(9.4*10^8)^2 + p^(2 + a)/(10^28)^a
(x1[n] -
(x2[n] (9.4*10^8)^(1 + a) + x3[n] (1.4*10^8)^(1 + a))
/(9.4*10^8 + 1.4*10^8)^(1 + a)) == 4*7.2*10^(-4) p,
20],
{p, 10^5, 10^28}]
Thanks a lot