Has anyone been able to compile matlisp on macosx for use with
cmucl.
Unfortunately I get the compile error
ld -G unknown option -G
-G seems to be the load shared library option that works in linux
but not osx.
Alternatively, does any one know of a lisp library for matrix operations.
I don't want to reinvent the wheel.
"Peter Scott" <·········@gmail.com> writes:
> If you're looking for an alternative, Sapaclisp may help. See
> <http://common-lisp.net/project/sapaclisp/>.
>
> -Peter
Note that I would NOT expect this to be usable for any sort of
"large-scale" computations --- as far as I can tell sapaclisp is using
generic arrays and arithmetic. Numerical linear algebra is one of
those areas where I feel that wrapping the canonical solution
(ATLAS/BLAS) is nearly always preferable to rolling one's own, even if
substantial work is involved.
Of course, I wish I'd known about sapaclisp a few weeks ago when I
wrote my own LU decomposition in order to work with high-precision
matrices in CLISP --- could've saved a day or two.
rif
In article <···············@five-percent-nation.mit.edu>, rif says...
>
>
>"Peter Scott" <·········@gmail.com> writes:
>
>> If you're looking for an alternative, Sapaclisp may help. See
>> <http://common-lisp.net/project/sapaclisp/>.
>>
>> -Peter
>
>Note that I would NOT expect this to be usable for any sort of
>"large-scale" computations --- as far as I can tell sapaclisp is using
>generic arrays and arithmetic. Numerical linear algebra is one of
>those areas where I feel that wrapping the canonical solution
>(ATLAS/BLAS) is nearly always preferable to rolling one's own, even if
>substantial work is involved.
>
>Of course, I wish I'd known about sapaclisp a few weeks ago when I
>wrote my own LU decomposition in order to work with high-precision
>matrices in CLISP --- could've saved a day or two.
>
>rif
>
Thanks Peter and rif. Unfortunately, I'm doing some "large scale"
computations, I'll see how things go. I'll also try and hack matlisp
on osx.
I'm now looking at http://www.femlisp.org/.
But I can't figure out the their licensing.
Fuad Abiad <·······@hotmail.com> writes:
> I'm now looking at http://www.femlisp.org/.
> But I can't figure out the their licensing.
Almost the best you can expect as a user: modified BSD. (Actually, this
information is not hard to find. Either click on the link "Savannah"
server where Femlisp is still hosted and the information will immediately
pop up, or read the file named "LICENSE" in the code.)
Nicolas.
Fuad Abiad wrote:
> Alternatively, does any one know of a lisp library for matrix operations.
> I don't want to reinvent the wheel.
If your matrices are of any non-trivial size, inventing an accurate
wheel is a specialist job, so you are indeed best off not attempting
it.
A couple of years back I needed eigenvalue/vector decompositions for
large, sparse matrices and after some googling ended up at BlueBit
Software http://www.bluebit.gr/. I haven't attempted to verify its
accuracy systematically, nor compare its performance against anything
else. It's working well enough for our purposes on both fronts, which
is all I'm bothered about.
The disadvantage with this library is that you need to be on Windoze to
use it (as its interfaces are .net and ActiveX). I can't believe
there's any reason what they couldn't compile very easily for macosx or
whatever, maybe nobody's ever made a good business case to them. Ask if
you need it, maybe they'll just say OK.
- nick
From: Raymond Toy
Subject: Re: matlisp/cmucl/macosx (or alterantive matrix library)
Date:
Message-ID: <sxd4qbkc5z7.fsf@rtp.ericsson.se>
>>>>> "Fuad" == Fuad Abiad <·······@hotmail.com> writes:
Fuad> Has anyone been able to compile matlisp on macosx for use with
Fuad> cmucl.
Fuad> Unfortunately I get the compile error
Fuad> ld -G unknown option -G
Are you using the CVS version? I have heard it works, but I cannot
myself test it because my Fortran compiler is incompatible with my C
compiler.
Ray
In article <···············@rtp.ericsson.se>, Raymond Toy says...
>
>>>>>> "Fuad" == Fuad Abiad <·······@hotmail.com> writes:
>
> Fuad> Has anyone been able to compile matlisp on macosx for use with
> Fuad> cmucl.
>
> Fuad> Unfortunately I get the compile error
> Fuad> ld -G unknown option -G
>
>Are you using the CVS version? I have heard it works, but I cannot
>myself test it because my Fortran compiler is incompatible with my C
>compiler.
>
>Ray
I was using 2_0 beta20021014
I just got the cvs version. Although configure recognizes darwin and
g77 and gives the right flags to ld . make does not compile the fortran
code and goes directly to loading lisp. So
it does work yet.
From: Raymond Toy
Subject: Re: matlisp/cmucl/macosx (or alterantive matrix library)
Date:
Message-ID: <sxdirzybp64.fsf@rtp.ericsson.se>
>>>>> "Fuad" == Fuad Abiad <·······@hotmail.com> writes:
Fuad> I was using 2_0 beta20021014
Fuad> I just got the cvs version. Although configure recognizes darwin and
Fuad> g77 and gives the right flags to ld . make does not compile the fortran
Fuad> code and goes directly to loading lisp. So
Fuad> it does work yet.
As I said, I can't really help much because my g77 doesn't work with
my gcc (different versions), so I can't even get past configure. But
I'm pretty sure at least one person has built matlisp on Mac OS X.
If you want to make this work, please subscribe to the matlisp mailing
list and we can try to figure out the problem.
Ray