[CSC 435] Matrices for Accuracy Testing

[Andrew J. Pounds]

Okay Bryan -- I couldn't find one -- but I built a system that should 
do.   You will find the matrices and vectors on the website.  To check 
your code, take the chi squared statistic (the sum of the squares of the 
residuals) between your solution vector x and the x vector I provided.   
If your linear solver is working properly you should get a very small 
residual.

http://theochem.mercer.edu/chm435

Look in the DATA section.


On 02/21/14 12:10, Bryan B Danley wrote:
> Do you have something similar for a large linear system that we can test our methods on? I have a few 4x4s coded up from examples in the books, but something larger would be nice, as well.
> ________________________________________
> From: csc435-bounces at theochem.mercer.edu <csc435-bounces at theochem.mercer.edu> on behalf of Andrew J. Pounds <pounds_aj at mercer.edu>
> Sent: Wednesday, February 19, 2014 1:39 PM
> To: csc435 at theochem.mercer.edu
> Subject: [CSC 435] Matrices for Accuracy Testing
>
> Guys -- while trying to look for a good example of a matrix that needs
> conditioning I found myself trying to test the accuracy of some of my
> own matrix codes.  This prompted me to build a 100x100 random matrix and
> its inverse.   It occurred to me that such a matrix could be EXTREMELY
> useful if you are trying to test the accuracy of your matrix multiply
> routines using serial, OpenMP, pThreads, and MPI methods.  Basically if
> you multiply the matrix and its inverse together you should get the
> identity matrix.  Summing up the diagonal should equal the dimension of
> the matrix.
>
> If you would like to use these for testing, the files are located on the
> class website in the DATA section.
>
> http://theochem.mercer.edu/csc435
>
>
>
>
>
> --
> Andrew J. Pounds, Ph.D.  (pounds_aj at mercer.edu)
> Professor of Chemistry and Computer Science
> Mercer University,  Macon, GA 31207   (478) 301-5627
> http://faculty.mercer.edu/pounds_aj
>
> _______________________________________________
> csc435 mailing list
> csc435 at theochem.mercer.edu
> http://theochem.mercer.edu/mailman/listinfo/csc435


-- 
Andrew J. Pounds, Ph.D.  (pounds_aj at mercer.edu)
Professor of Chemistry and Computer Science
Mercer University,  Macon, GA 31207   (478) 301-5627
http://faculty.mercer.edu/pounds_aj