How realistic would it be to convert a Fortran 77 program to ACL6?
In particular, how do the matrix inversion capabilities compare?
And if they do not compare, is it possible to create them using Lisp alone?
In article <············@slb7.atl.mindspring.net>, Link Davis wrote:
>How realistic would it be to convert a Fortran 77 program to ACL6?
>
>In particular, how do the matrix inversion capabilities compare?
>
>And if they do not compare, is it possible to create them using Lisp alone?
You have all the numeric capabilities you need to implement matrix
inversion. Matrix inversion is not a simple problem; the results are
sensitive to the method that is applied due to floating point truncation
issues.
Because Common Lisp has built-in rational types, you could actually
perform an *exact* inverse using straightforward Gaussian elimination
on a matrix containing only rationals. Yet that same code could also
compute a reasonable approximation of a floating-point matrix inverse,
depending on the input.
Thank you very much.
Kaz Kylheku wrote in message ...
>In article <············@slb7.atl.mindspring.net>, Link Davis wrote:
>>How realistic would it be to convert a Fortran 77 program to ACL6?
>>
>>In particular, how do the matrix inversion capabilities compare?
>>
>>And if they do not compare, is it possible to create them using Lisp
alone?
>
>You have all the numeric capabilities you need to implement matrix
>inversion. Matrix inversion is not a simple problem; the results are
>sensitive to the method that is applied due to floating point truncation
>issues.
>
>Because Common Lisp has built-in rational types, you could actually
>perform an *exact* inverse using straightforward Gaussian elimination
>on a matrix containing only rationals. Yet that same code could also
>compute a reasonable approximation of a floating-point matrix inverse,
>depending on the input.
Thanks!
>Because Common Lisp has built-in rational types, you could actually
>perform an *exact* inverse using straightforward Gaussian elimination
>on a matrix containing only rationals. Yet that same code could also
>compute a reasonable approximation of a floating-point matrix inverse,
>depending on the input.
"Link Davis" <·········@mindspring.com> writes:
> How realistic would it be to convert a Fortran 77 program to ACL6?
>
> In particular, how do the matrix inversion capabilities compare?
>
> And if they do not compare, is it possible to create them using Lisp alone?
Have you come across matlisp ?
http://matlisp.sourceforge.net/
Jacek
>Have you come across matlisp ?
>
>http://matlisp.sourceforge.net/
>
>Jacek
Not until you mentioned it.
Thanks for the great link.
However, I hope to use Lisp exclusively.
Speed is not deciding factor since this is a learning project with modest
initial applications in mind.
"Link Davis" <·········@mindspring.com> writes:
> However, I hope to use Lisp exclusively.
Check out CMU's ai-repository:
<http://www-cgi.cs.cmu.edu/afs/cs/project/ai-repository/ai/lang/lisp/code/math/matrix>
Ole
>Check out CMU's ai-repository:
>
><http://www-cgi.cs.cmu.edu/afs/cs/project/ai-repository/ai/lang/lisp/code/m
ath/matrix>
>
> Ole
Thank you.
I wonder... are there any pure-Lisp FEA packages?
"Link Davis" <·········@mindspring.com> writes:
> I wonder... are there any pure-Lisp FEA packages?
There will be when I've finished :-) (Don't hold your breath.)