|
[Sponsors] |
April 25, 2010, 02:28 |
Solving Linear System of Equations in UDF
|
#1 |
Member
MOHSEN Zendehbad
Join Date: Aug 2009
Posts: 35
Rep Power: 17 |
Somewhere in my UDF i should solve linear system of equations(Ax = b), where i have A matrix and b vector and i need to compute x vector,
Now do any one know if there is an internal linear solver in fluent udf that i can use to compute x vector? REGARDS |
|
April 25, 2010, 10:05 |
|
#2 |
Senior Member
Join Date: Nov 2009
Posts: 411
Rep Power: 20 |
I think I've seen in one of the *.h files of Fluent a declaration for a Gauss solver (I don't have a Fluent installed on this machine to give you an exact name).
However I'm sure you can find on Internet a simple Gauss solver in pure C, you can simply use this function in your UDF. Do |
|
April 26, 2010, 15:56 |
|
#3 |
Senior Member
Join Date: Nov 2009
Posts: 411
Rep Power: 20 |
Precisely the above mentioned code is in linalg.h in your Fluent directory, this is a piece of code from the file:
void Gauss_Elimination(const int n_rows, const int n_cols, real A[], real X[], real B[], int swap_rows) This is what you need ! (I hope) Do |
|
April 26, 2010, 19:42 |
|
#4 | |
Senior Member
Arjun
Join Date: Mar 2009
Location: Nurenberg, Germany
Posts: 1,286
Rep Power: 34 |
Quote:
why you want to use fluent. Do you need this as a part of simulation. if you do not need it as part of simultion than there are many libraries that would do it for you. |
||
May 8, 2010, 08:48 |
|
#5 |
Member
MOHSEN Zendehbad
Join Date: Aug 2009
Posts: 35
Rep Power: 17 |
Isn't there any iterative method (such as GMRES or conjugate gradient) since gauss elimination is not responding to huge matrices?
REGARDS MZ |
|
May 8, 2010, 11:39 |
|
#6 |
Senior Member
Join Date: Nov 2009
Posts: 411
Rep Power: 20 |
Hello,
I've found only Gauss, but you can search in all Fluent's *.h files from the udf's directory, maybe I've missed something (probably there are less then 100 *.h files so you can do a quick check). If you want a more complex solver I'm sure you can link your udf with an open source math library (I said open source because you will need to compile this just like you compile your own UDF). Unfortunately you need a pure C library, I think GSL ( http://www.gnu.org/software/gsl/ ) may be a good choice, not sure how easy it will be for you to recompile it. Or you can write a C wrapper for Lapack which is a state of the art linear algebra library, probably this will be more feasible for any operating system (Lapack is written in Fortran). Do you use Linux or Windows ? Do |
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
convergence problem when use pisoFoam, LES for wind tunnel case | Forrest_Lei | OpenFOAM | 3 | July 19, 2011 07:00 |
Unknown error | sivakumar | OpenFOAM Pre-Processing | 9 | September 9, 2008 13:53 |
Convergence moving mesh | lr103476 | OpenFOAM Running, Solving & CFD | 30 | November 19, 2007 15:09 |
IcoFoam parallel woes | msrinath80 | OpenFOAM Running, Solving & CFD | 9 | July 22, 2007 03:58 |
advice needed in solving linear system | yf yap | Main CFD Forum | 5 | February 5, 2001 01:21 |