MathGroup Archive 2003

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: successive over relaxation -> SparceLinearSolve

  • To: mathgroup at smc.vnet.net
  • Subject: [mg38914] Re: successive over relaxation -> SparceLinearSolve
  • From: Mike <mikeh1980 at optusnet.com.au>
  • Date: Sat, 18 Jan 2003 00:39:22 -0500 (EST)
  • References: <200301090940.EAA07078@smc.vnet.net> <avm4ji$na0$1@smc.vnet.net> <avu05n$dd4$1@smc.vnet.net>
  • Sender: owner-wri-mathgroup at wolfram.com

According to WRI tech support a bug exists in Developer`SparseLinearSolve[].

I recently contacted WRI about another problem I was working on in which
Developer`SparseLinearSolve[] was giving different results to LinearSolve.
They stated that a bug existed in the current version however in the
development version it is apparently corrected.

I don't know how widely this error will effect other peoples problems but
for the previous problem I was working on  (which was an asymmetrical
matrix) it certainly made using it impossible. In my problem an instability
somehow arose. I was solving an equation where the output was the input at
the next time step. Within a few time steps instead of returning values
between 0. and 1. I was getting 10^6 and so on.

LinearSolve gave good results but slow because of the large matrix size.

Regards

Mike

On 13/1/03 8:16 PM, in article avu05n$dd4$1 at smc.vnet.net, "Jens-Peer Kuska"
<kuska at informatik.uni-leipzig.de> wrote:

> Hi,
> 
> and
> 
> Developer`SparseLinearSolve[]
> 
> does not a better job than any SOR ??
> 
> Regards
> Jens
> 
> Selwyn Hollis wrote:
>> 
>> Mike,
>> 
>> I doubt you'll find what you're looking for. I recently spent some time
>> trying to concoct an efficient Gauss-Seidel-SOR program in Mathematica
>> and left it before getting anything I was happy with. There are inherent
>> difficulties, I think. However, I believe it's an very interesting
>> problem to find the "best" way of implementing SOR in Mathematica.
>> 
>> ----
>> Selwyn Hollis
>> 
>> Mike wrote:
>>> Does anyone know of any sources of examples of successive over relaxation
>>> method using mathematica?
>>> 
>>> I came across a link on mathsource but the notebook actually links to ITPACK
>>> method. I was interested in a full implementation within mathematica.
>>> 
>>> 
>>> Thanks
>>> 
>>> Mike
>>> 
>>> 
>>> 
> 



  • Prev by Date: Re: Save as Special XHTML+MathML bug
  • Next by Date: RE: What is a smart way to do this?
  • Previous by thread: Re: successive over relaxation
  • Next by thread: Re: Re: successive over relaxation