| Re: Singular jacobian in broyden [message #42788] |
Thu, 24 February 2005 02:34 |
Jerome Colin
Messages: 8
Registered: February 2005
|
Junior Member |
|
|
Ralf Schaa wrote:
> Jerome Colin wrote:
>
>> Hello,
>>
>> I try to use the Broyden function to resolve a set of three equations.
>> I've declared the three equations in a function, and provided initial
>> guess. The code is compiled well, but I get the message :
>> 'Singular jacobian in broydn'
>>
>> As I'm not familiar with this method, I'd appreciate any help and/or
>> comments about the source of this error.
>
>
> Jerome,
>
> Please have a look at the numerical recipes, Chapter 9.7 "Root Finding
> and Nonlinear Sets of Equations". On page 382 you will find the
> paragraph "Multidimensional Secant Methods: Broyden�s Method" and on
> page 386 there is a comment on singular jaciobians.
> Download the pdf-files from numerical recipes e.g. at:
> http://www.library.cornell.edu/nr/cbookfpdf.html
>
> I had difficulties using Broyden method's as implemented in IDL
> following the numerical recipes, since it is very strong depending on
> the start vector.
> I used methods in Mathematica though and had 'the feeling' it worked
> better, but I could not proove it.
>
> -Ralf
Thank you Ralf !
Jerome
|
|
|
|
| Re: Singular jacobian in broyden [message #42789 is a reply to message #42788] |
Thu, 24 February 2005 01:53  |
Ralf Schaa
Messages: 37
Registered: June 2001
|
Member |
|
|
Jerome Colin wrote:
> Hello,
>
> I try to use the Broyden function to resolve a set of three equations.
> I've declared the three equations in a function, and provided initial
> guess. The code is compiled well, but I get the message :
> 'Singular jacobian in broydn'
>
> As I'm not familiar with this method, I'd appreciate any help and/or
> comments about the source of this error.
Jerome,
Please have a look at the numerical recipes, Chapter 9.7 "Root Finding
and Nonlinear Sets of Equations". On page 382 you will find the
paragraph "Multidimensional Secant Methods: Broyden�s Method" and on
page 386 there is a comment on singular jaciobians.
Download the pdf-files from numerical recipes e.g. at:
http://www.library.cornell.edu/nr/cbookfpdf.html
I had difficulties using Broyden method's as implemented in IDL
following the numerical recipes, since it is very strong depending on
the start vector.
I used methods in Mathematica though and had 'the feeling' it worked
better, but I could not proove it.
-Ralf
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
