lapackpp-devel — Discussion of Lapack++ development

Re: [Lapackpp-devel] Lapackpp scaled matrix addition

[Matti V. <mav...@ja...>]

On Tue, 15 May 2007, Christian Stimming wrote:

> Am Montag, 14. Mai 2007 16:59 schrieb Matti Varjokallio:
> > > >Am I correct in saying that currently the possibilities to compute a
> > > >scaled matrix sum in lapack++ are:
> > > >1) Do a basic double-loop
> > > >2) Blas_Mat_Mat_Mult with the second argument as identity matrix
> > > >3) Blas_Scale for the second argument and use +-operator which is
> > > >marked deprecated
> > > >
> > > > It seems that there is a Blas_Add_Mult -function only for vectors, but
> > > > not for matrices? Is there some specific reason for this or could a
> > > > wrapper for dgema be added?
> > >
> > > Oh, there exists dgema that does this? In that case it would be nice to
> > > have a wrapper for this, indeed. If you're up to submitting a patch, I'd
> > > happily include it into lapackpp; otherwise I might be able to do this
> > > maybe in June.
> >
> > Sorry it turned out that I was browsing through some Compaq-specific
> > extensions and thought it was plain LAPACK-interface. Also ESSL gives
> > dgeadd for this purpose, but there seems to be no such function in LAPACK.
> 
> Ah, ok. Yes, that's what I knew as well. Even though I needed this at times, 
> there wasn't any function available except for Blas_Mat_Mat_Mult with an 
> identity matrix.
> 
> > Would it make sense to use daxpy (scaled vector addition) to make
> > something like Blas_Add_Mult(LaGenMatDouble&, double, const
> > &LaGenMatDouble)?
> 
> I'm not so sure. We're talking about an element-wise sum of matrices? In that 
> case daxpy does not make sense because it always calculates the (scaled) 
> inner product of two vectors, if I recall correctly. 
> 
> I'm open for patches that implement this function "by hand" in lapackpp. 
> However I've regularly been surprised by the speed of the existing LAPACK 
> function. In cases like this I'd like to hear (with some convincing numbers) 
> that a hand-written function is indeed faster than Blas_Mat_Mat_Mult 
> with "weird" input arguments (identity etc.). Profiling numbers could be 
> obtained by e.g. valgrind's tool callgrind or others.
> 
> Regards,
> 
> Christian


Hi again,

Daxpy wasn't scaled inner product, but scaled vector addition. 
Blas_Add_Mult uses it internally. It seems to work 
for matrices as well, so I suggest:

void Blas_Add_Mult_Mat(LaGenMatDouble &A,
                       double alpha,
                       const LaGenMatDouble &B)
{
  assert(A.rows() == B.rows());
  assert(A.cols() == B.cols());
  assert(A.inc(0) == B.inc(0));
  assert(A.inc(1) == B.inc(1));
  
  integer n = A.rows()*A.cols();
  integer inca = A.inc(0), incb = B.inc(0);
  
  F77NAME(daxpy)(&n, &alpha, &B(0,0), &incb, &A(0,0), &inca);
}


...although I tried it only couple of occasions. I don't know if those 
inc-asserts are needed. Just naming it Blas_Add_Mult won't work because 
LaVectorDouble was inherited from LaGenMatDouble.


-Matti

Re: [Lapackpp-devel] Lapackpp scaled matrix addition

[Christian S. <sti...@tu...>]

Am Montag, 14. Mai 2007 16:59 schrieb Matti Varjokallio:
> > >Am I correct in saying that currently the possibilities to compute a
> > >scaled matrix sum in lapack++ are:
> > >1) Do a basic double-loop
> > >2) Blas_Mat_Mat_Mult with the second argument as identity matrix
> > >3) Blas_Scale for the second argument and use +-operator which is
> > >marked deprecated
> > >
> > > It seems that there is a Blas_Add_Mult -function only for vectors, but
> > > not for matrices? Is there some specific reason for this or could a
> > > wrapper for dgema be added?
> >
> > Oh, there exists dgema that does this? In that case it would be nice to
> > have a wrapper for this, indeed. If you're up to submitting a patch, I'd
> > happily include it into lapackpp; otherwise I might be able to do this
> > maybe in June.
>
> Sorry it turned out that I was browsing through some Compaq-specific
> extensions and thought it was plain LAPACK-interface. Also ESSL gives
> dgeadd for this purpose, but there seems to be no such function in LAPACK.

Ah, ok. Yes, that's what I knew as well. Even though I needed this at times, 
there wasn't any function available except for Blas_Mat_Mat_Mult with an 
identity matrix.

> Would it make sense to use daxpy (scaled vector addition) to make
> something like Blas_Add_Mult(LaGenMatDouble&, double, const
> &LaGenMatDouble)?

I'm not so sure. We're talking about an element-wise sum of matrices? In that 
case daxpy does not make sense because it always calculates the (scaled) 
inner product of two vectors, if I recall correctly. 

I'm open for patches that implement this function "by hand" in lapackpp. 
However I've regularly been surprised by the speed of the existing LAPACK 
function. In cases like this I'd like to hear (with some convincing numbers) 
that a hand-written function is indeed faster than Blas_Mat_Mat_Mult 
with "weird" input arguments (identity etc.). Profiling numbers could be 
obtained by e.g. valgrind's tool callgrind or others.

Regards,

Christian

Re: [Lapackpp-devel] Lapackpp scaled matrix addition

[Christian S. <sti...@tu...>]

Quoting Matti Varjokallio <mav...@ja...>:
> Am I correct in saying that currently the possibilities to compute a
> scaled matrix sum in lapack++ are:
> 1) Do a basic double-loop
> 2) Blas_Mat_Mat_Mult with the second argument as identity matrix
> 3) Blas_Scale for the second argument and use +-operator which is
> marked deprecated
>
> It seems that there is a Blas_Add_Mult -function only for vectors, but not
> for matrices? Is there some specific reason for this or could a wrapper
> for dgema be added?

Oh, there exists dgema that does this? In that case it would be nice  
to have a wrapper for this, indeed. If you're up to submitting a  
patch, I'd happily include it into lapackpp; otherwise I might be able  
to do this maybe in June.

Regards,

Christian

Re: [Lapackpp-devel] Lapack++ Problem

[Tarun T. <tar...@gm...>]

Hi Folks

I have downloaded the Lapack++ v2.5.0 from
http://lapackpp.sourceforge.net/and using it in MS Visual Studio 2005.
I am able to compile the source code
but not able to run the test cases like tgc,tgd,tGenSolve and tvd. No matter
what argument I pass into, it takes some arbitrary values. I am not able to
figure out if there is some bug in the code itself or something else.

Guyz please if someone can put in their two cents. Any help in this regard
would be highly appreciated.

Thanks

Re: [Lapackpp-devel] Lapack++ configuration

[Christian S. <sti...@tu...>]

Am Donnerstag, 22. M=E4rz 2007 14:04 schrieb Jens A. Berger:
> Hi tried to configure, make lapackpp-2.5.1 under SUSE 10.1 with suse rpm
> packages for blas and lapack installed under  /usr/lib/liblapack.so and
> /usr/lib/libblas.so:
>
> jberger@lt4:~/devel/FD/lapackpp-2.5.1 8> ./configure
>
> bla bla .......
>
> checking for cblas library... configure: WARNING: not found
> checking whether ATLAS is usable... configure: WARNING: no
> checking for sgemm_ in -lblas... no
> checking for sgemm_ in -lcxml... no
> checking for sgemm_ in -ldxml... no
> checking for sgemm_ in -lscs... no
> checking for sgemm_ in -lcomplib.sgimath... no
> checking for sgemm_ in -lblas... (cached) no
> checking for sgemm_ in -lblas... (cached) no
> checking for sgemm_ in -lblas32... no
> configure: error: Blas/Lapack was not found.
> *** This means Lapack++ and matrix support cannot be compiled.
> *** This makes this library unusable. Please get blas and lapack
> *** installed. If you do have these installed, use the options
> *** --with-blas=3D<libname> or --with-lapack=3D<libname> and/or set
> *** the env variable LDFLAGS to include the appropriate linker
> *** flags.

I can only guess that maybe you don't have the gcc-fortran package installe=
d=20
(might be called gcc-g77 in suse10.1; in suse 10.2 it's called gcc-fortran)=
=2E=20
To be sure, you'd have to look into the config.log file at approx. 66% into=
=20
the file where these actual error messages are. Maybe it simply says "g77:=
=20
not found" or "-lgfortran: not found" similar, which simply means there is =
no=20
fortran compiler or its library available.

Christian

>
> tried:
> ./configure --with-blas=3Dlibblas --with-lapack-libs=3D/usr/lib/
> --with-lapack=3Dliblapack
> and:
> ./configure --with-blas=3D/usr/lib/libblas.so
> --with-lapack=3D/usr/lib/liblapack.so
>
> but does not solve the problem. What did I do wrong?
>
> Thanks in advance, Jens

Re: [Lapackpp-devel] Compiling error: laversion.h

[Christian S. <sti...@tu...>]

1. Please reply always on the mailing list, never to individual developers.

Am Mittwoch, 28. Februar 2007 14:18 schrieb Hudelist, Florian:
> Hi Christian,
> Thanks for the quick reply! I followed the installation instructions
> exactly. During the make install process I got this message:
> ...
> Making install in include
> make[1]: Entering directory `/home/fh22/LAPACK++/lapackpp-2.5.1/include'
> make  install-am
> make[2]: Entering directory `/home/fh22/LAPACK++/lapackpp-2.5.1/include'
> make[3]: Entering directory `/home/fh22/LAPACK++/lapackpp-2.5.1/include'
> make[3]: Nothing to be done for `install-exec-am'.
> test -z "/usr/local/include/lapackpp" || mkdir -p --
> "/usr/local/include/lapackpp"
> mkdir: cannot create directory `/usr/local/include/lapackpp': Permission
> denied

Please first read INSTALL and learn a bit of unix administration - this 
mailing list here is specifially about issues with lapackpp, not with unix 
administration in general.

You can use e.g.
 ./configure --prefix=$HOME/lapackpp
or
 ./configure --prefix=$HOME/usr

> I do not have administrator rights so I can't write in
> /usr/local/include. I added the
> -I/home/fh22/LAPACK++/lapackpp-2.5.1/include to my compilation command
> hoping that it still works.

No. This is wrong.

> Do I have to contact the administrator to do this for me? (I already
> asked him and he is absolutely not keen on installling new software)

For further questions on how to install software into your own user's 
directory, please ask your administrators, not this list.

Best Regards,

Christian Stimming

Re: [Lapackpp-devel] Compiling error: laversion.h

[Christian S. <sti...@tu...>]

Am Mittwoch, 28. Februar 2007 13:57 schrieb Hudelist, Florian:
> I just started to use LAPACK++ and I get the following errormessage from
> my compiler:
>
>
> slioch:temp> gcc test.cpp
>
> /home/fh22/LAPACK++/lapackpp-2.5.1/include/lacomplex.h:45:23:
> laversion.h: No such file or directory
>
>
> I had a look in this folder and found a file called laversion.h.in. I
> just renamed this funcion to laversion.h or is that something more
> serious?

?!?? How did you expect to install lapackpp? The instructions (file INSTALL, 
also known as RTFM) clearly say you have to do

  ./configure --prefix=/your/preferred/install/location
  make 
  make install

which will also make sure to create and install laversion.h correctly. It is 
*wrong* to copy laversion.h.in to laversion.h.

Christian

[Lapackpp-devel] Compiling error: laversion.h

[Hudelist, F. <fh...@hw...>]

Hi,

I just started to use LAPACK++ and I get the following errormessage from
my compiler:

=20

slioch:temp> gcc test.cpp

In file included from
/home/fh22/LAPACK++/lapackpp-2.5.1/include/lapackc.h:14,

                 from
/home/fh22/LAPACK++/lapackpp-2.5.1/include/lapack.h:10,

                 from
/home/fh22/LAPACK++/lapackpp-2.5.1/include/lapackpp.h:16,

                 from
/home/fh22/LAPACK++/lapackpp-2.5.1/include/lapack++.h:1,

                 from test.cpp:2:

/home/fh22/LAPACK++/lapackpp-2.5.1/include/lacomplex.h:45:23:
laversion.h: No such file or directory

In file included from
/home/fh22/LAPACK++/lapackpp-2.5.1/include/blas1pp.h:36,

                 from
/home/fh22/LAPACK++/lapackpp-2.5.1/include/blaspp.h:39,

                 from
/home/fh22/LAPACK++/lapackpp-2.5.1/include/lapackpp.h:52,

                 from
/home/fh22/LAPACK++/lapackpp-2.5.1/include/lapack++.h:1,

                 from test.cpp:2:

/home/fh22/LAPACK++/lapackpp-2.5.1/include/blas1.h:9:52: laversion.h: No
such file or directory

test.cpp: In function `int main()':

test.cpp:19: `zgeev' undeclared (first use this function)

test.cpp:19: (Each undeclared identifier is reported only once for each

   function it appears in.)

=20

I had a look in this folder and found a file called laversion.h.in. I
just renamed this funcion to laversion.h or is that something more
serious?

=20

After I changed that I still get the error:=20

=20

test.cpp: In function `int main()':

test.cpp:19: `zgeev' undeclared (first use this function)

test.cpp:19: (Each undeclared identifier is reported only once for each

   function it appears in.)

=20

My source code is:=20

=20

#include <iostream>

#include <lapack++.h>

# include <complex>

using namespace std;

=20

=20

int main(){

            complex<double> M[3][3], eigVecL[3][3],eigVecR[3][3],
eigVal[3];

            double lWork[6],rWork[6];

            int info, dim=3D3;

           =20

           =20

            M[1][1]=3Dcomplex<double>(1,1);

            M[1][2]=3Dcomplex<double>(1,2);

            M[2][1]=3Dcomplex<double>(2,1);

            M[0][1]=3Dcomplex<double>(0,1);

            M[0][2]=3Dcomplex<double>(0,2);

=20

            zgeev('V', 'V', dim, M, dim, eigVal, eigVecL, 10, eigVecR,
10, rWork, 64*dim, rWork, info);

}

=20

What could be the problem for that?

=20

Thanks,

Florian

Re: [Lapackpp-devel] Newbie & "float" support

[Christian S. <sti...@tu...>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Luigi Bardelli schrieb:
> Dear all,
>    I have just started using lapackpp, and I would like to ask you: do lapackpp
> provide matrix and vector classes with type float? 

Matrix yes, vector no. (Although the LaGenVectorDouble is just a
convenience representation of the LaGenMatDouble with one dimension set
to one, so this restriction shouldn't be too difficult.)

> If yes, are they "supported"
> and tested as well as the double ones? (I have to work with large amounts of
> data where the float precision is more than adequate)

Unfortunately, no. All the "interesting" functions like matrix
multiplications and solvers are only implemented for the double
precision. You would have to add a Float version of each of these
functions. In principle, this isn't too difficult, because in almost all
cases you would copy&paste the Double code to Float, and replace the
dfoo function of LAPACK by its ffoo version. Although the function
declarations of LAPACK for the f* functions are also missing and would
need to be added to include/blas1.h etc.

> At http://lapackpp.sourceforge.net/html/hierarchy.html the class LaGenMatFloat
> is reported but no other classes...

Well, effectively only Double is supported. For Float there is the
matrix class but no other function available.

Christian
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.1 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iQCVAwUBRdw1K2XAi+BfhivFAQLgaAQAj7VkZgaRmoxRRUckpcbAMmb/FWdHmB7u
ENlyIxZ3Oze3VqNgw8W7TH/XaJOdod//iuAe5tQxoNpomkLDqdiqk72Qt8hKwsSS
RlkNMYlGA8JuQUCOpi2f6NcqadScaLJZpYHuz8GPuSVT4n+nbsF8MRC380DtpYPn
jlEa1pyDsGw=
=/xCE
-----END PGP SIGNATURE-----

[Lapackpp-devel] Newbie & "float" support

[Luigi B. <lui...@ya...>]

Dear all,
   I have just started using lapackpp, and I would like to ask you: do lapackpp
provide matrix and vector classes with type float? If yes, are they "supported"
and tested as well as the double ones? (I have to work with large amounts of
data where the float precision is more than adequate)

At http://lapackpp.sourceforge.net/html/hierarchy.html the class LaGenMatFloat
is reported but no other classes...

Thank you
        Luigi


	

	
		
___________________________________ 
L'email della prossima generazione? Puoi averla con la nuova Yahoo! Mail: 
http://it.docs.yahoo.com/nowyoucan.html

Re: [Lapackpp-devel] shared libs install problem

[Christian S. <sti...@tu...>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Dear Dan,

Dan Frankel schrieb:
> 	I installed lapackpp-2.5.1 on a Fedora 6 Linux box and am having
> problems with shared libraries.  The configure and make seemed to work
> okay, but when I do a 'make check' I get the  error shown below, which I
> list as ERROR #1.  Also,  when I try to link against lapackpp I get an
> error shown below as ERROR #2. Are these the same problem?  Any idea
> what may be happening?  I'd greatly appreciate any suggestions.  

Unfortunately I have never seem such an error, especially the first one:

> /usr/local/share/lapackpp-2.5.1/blaspp/testing/.libs/lt-tblasd++: error
> while loading shared
> libraries: /usr/local/share/lapackpp-2.5.1/src/.libs/liblapackpp.so.15:
> cannot restore segment prot after reloc: Permission denied

Basically I'd guess this could be caused by an erroneous linker or
compiler or libtool installation. Do you have more than one version of
each of these installed (e.g. on in /usr/local, the other in /usr),
which will then interfere with each other?

> ERROR #2 -- Occurs at compile time for my C++ class which
> instantiates a LaGenMatDouble:
> 
> /usr/local/share/lapackpp-2.5.0/include/lavd.h:53: undefined reference
> to `LaGenMatDouble::~LaGenMatDouble()'

Huh? This error talks about a different lapackpp version 2.5.0?
Anyway, in terms of the header file declaration this particular function
has always been correctly declared. The error rather means that the
liblapackpp.so library which is being linked against is somehow broken.
If the tests above didn't work anyway, the actual compiling won't work
as well.

Christian
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.1 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iQCVAwUBRdwQCGXAi+BfhivFAQJOQQQAmRXZr/Tzy8IGieoNyfRQnR2CqBDfuC4j
CTDYqGLUR5P+0LHZkD7DrRQ8s/eNmL0P9cC7mDpGsFSA7+haiM7zruAC9ubsbZUI
fV7eU6TSy5kM3geEORx+oQfOIau1LV9XfvAbzHq+1pjJ67dxXA2MkeBjlGS+f/P0
+tnKTQ7sgzU=
=7uTZ
-----END PGP SIGNATURE-----

[Lapackpp-devel] shared libs install problem

[Dan F. <wa...@co...>]

Hello,

	I installed lapackpp-2.5.1 on a Fedora 6 Linux box and am having
problems with shared libraries.  The configure and make seemed to work
okay, but when I do a 'make check' I get the  error shown below, which I
list as ERROR #1.  Also,  when I try to link against lapackpp I get an
error shown below as ERROR #2. Are these the same problem?  Any idea
what may be happening?  I'd greatly appreciate any suggestions.  

Thanks very much,

Dan Frankel

ERROR #1:

[root@localhost lapackpp-2.5.1]# make check
Making check in include
make[1]: Entering directory `/usr/local/share/lapackpp-2.5.1/include'
make  check-am
make[2]: Entering directory `/usr/local/share/lapackpp-2.5.1/include'
make[2]: Nothing to be done for `check-am'.
make[2]: Leaving directory `/usr/local/share/lapackpp-2.5.1/include'
make[1]: Leaving directory `/usr/local/share/lapackpp-2.5.1/include'
Making check in blaspp
make[1]: Entering directory `/usr/local/share/lapackpp-2.5.1/blaspp'
Making check in src
make[2]: Entering directory `/usr/local/share/lapackpp-2.5.1/blaspp/src'
make[2]: Nothing to be done for `check'.
make[2]: Leaving directory `/usr/local/share/lapackpp-2.5.1/blaspp/src'
Making check in testing
make[2]: Entering directory
`/usr/local/share/lapackpp-2.5.1/blaspp/testing'
make  tblasd++ check_blas1pp \
          blas++_test
make[3]: Entering directory
`/usr/local/share/lapackpp-2.5.1/blaspp/testing'
make[3]: `tblasd++' is up to date.
make[3]: `check_blas1pp' is up to date.
make[3]: Nothing to be done for `blas++_test'.
make[3]: Leaving directory
`/usr/local/share/lapackpp-2.5.1/blaspp/testing'
make  check-TESTS
make[3]: Entering directory
`/usr/local/share/lapackpp-2.5.1/blaspp/testing'
tblasd++ 10 10 > tblasd++10.out
/usr/local/share/lapackpp-2.5.1/blaspp/testing/.libs/lt-tblasd++: error
while loading shared
libraries: /usr/local/share/lapackpp-2.5.1/src/.libs/liblapackpp.so.15:
cannot restore segment prot after reloc: Permission denied
FAIL: blas++_test
/usr/local/share/lapackpp-2.5.1/blaspp/testing/.libs/lt-check_blas1pp:
error while loading shared
libraries: /usr/local/share/lapackpp-2.5.1/src/.libs/liblapackpp.so.15:
cannot restore segment prot after reloc: Permission denied
FAIL: check_blas1pp
===================
2 of 2 tests failed
===================
make[3]: *** [check-TESTS] Error 1
make[3]: Leaving directory
`/usr/local/share/lapackpp-2.5.1/blaspp/testing'
make[2]: *** [check-am] Error 2
make[2]: Leaving directory
`/usr/local/share/lapackpp-2.5.1/blaspp/testing'
make[1]: *** [check-recursive] Error 1
make[1]: Leaving directory `/usr/local/share/lapackpp-2.5.1/blaspp'
make: *** [check-recursive] Error 1
[root@localhost lapackpp-2.5.1]# 


ERROR #2 -- Occurs at compile time for my C++ class which
instantiates a LaGenMatDouble:

/usr/local/share/lapackpp-2.5.0/include/lavd.h:53: undefined reference
to `LaGenMatDouble::~LaGenMatDouble()'

Re: [Lapackpp-devel] Hilfe bei Lapack++ auf Windows XP

[Christian S. <sti...@tu...>]

Am Donnerstag, 15. Februar 2007 17:44 schrieb Christoph Schneider:
> for the integration of the lapack++ package into the c++ environment
> (Microsoft Visual Studio 2003 .Net) I followed the instructions
> "For compiling on Windows with the Microsoft Visual Studio C++ (MSVC)
> compiler" on http://lapackpp.sourceforge.net/ and the compilation/build
> of the lapackpp.vcproj succeeded, ending up in an compiled LAPACKPP.DLL.
>
> So I am not using the precompiled dll.

Okay. Then I assume you also compiled the three example program that are 
included in lapackpp.vcproj? These obviously linked without this linker 
error, so you should try to compare the compiler/linker settings of the 
example programs with the ones of your own program.

> As I stated before, I would like to use the linear algebra solvers of
> lapack++ for a c++ programm. So I started to declare the matrixes by
>
> LaGenMatDouble A(N,N);
>
> in my program, but received the following errors while compiling the c++
> file with the Microsoft Visual Studio C++ (MSVC) compiler:
>
> baum.obj : error LNK2019: unresolved external symbol
> "__declspec(dllimport) public: virtual __thiscall
> LaGenMatDouble::~LaGenMatDouble(void)" (__imp_??1LaGenMatDouble@@UAE@XZ)
> referenced in function _main

First I'd suggest you should check whether the linker actually finds your 
newly compiled version of lapackpp.dll and lapackpp.lib. Well, other than 
that... I don't have MSVC and/or winXP at hand, so I can really only suggest 
you should compare your own program to lapackpp's example program.

We would be glad to hear about the final outcome of your problem.

Regards

Christian

Re: [Lapackpp-devel] Hilfe bei Lapack++ auf Windows XP

[Christian S. <sti...@tu...>]

Hello,

1. Please always write to the lapackpp-devel mailinglist, *never* to=20
individial developers. (Answering in english, but German is ok for me)

Am Donnerstag, 15. Februar 2007 13:52 schrieb Christoph Schneider:
> Ich will ein C++ Programm schreiben, mit Microsoft Visual Studio 2003
> unter Windows XP und m=F6chte gerne das Lapack++ paket benutzen. Die
> Installation habe ich wie auf der sourceforge Seite
> http://lapackpp.sourceforge.net/html/index.html vorgenommen.
>
> Es kommt jedoch beim Comilieren bereits bei der Variablendeklaration zu
> folgender Fehlermeldung:
>
> baum.obj : error LNK2019: unresolved external symbol
> "__declspec(dllimport) public: virtual __thiscall
> LaGenMatDouble::~LaGenMatDouble(void)" (__imp_??1LaGenMatDouble@@UAE@XZ)
> referenced in function _main

Seems to me you are trying to use the pre-compiled DLL but you are compilin=
g=20
your own project with MSVC. To quote the page you've referenced: "Watch out=
:=20
This DLL works only with the gcc compiler, not with the Microsoft Visual=20
Studio C++ (MSVC) compiler!"=20

If that's not your problem, please describe more clearly *how* you installe=
d=20
lapackpp.

Christian


> baum.obj : error LNK2019: unresolved external symbol
> "__declspec(dllimport) public: __thiscall
> LaGenMatDouble::LaGenMatDouble(int,int)"
> (__imp_??0LaGenMatDouble@@QAE@HH@Z) referenced in function _main
> Debug/tree.exe : fatal error LNK1120: 2 unresolved externals
>
>
> das Programm baum.cpp lautet:
>
> **************************
>
> #include <lapackpp.h>
>
> int N;
>
> main()
> {
> N=3D4;
> LaGenMatDouble A(N,N);
>
> }
>
> ******************************
>
>
> Wo k=F6nnte deiner Meinung der Fehler liegen? Kann er beim Linken nicht
> auf die notwendige DLL zugreifen?
>
> Vielen Dank im voraus
> Christoph Schneider

Re: [Lapackpp-devel] So what is the advantage of lapackpp over lapack?

[Christian S. <sti...@tu...>]

Am Sonntag, 4. Februar 2007 15:06 schrieb sun:
> I know that LAPACK has both C and Fortran versions.
> What is the advatage of lapackpp then?

LAPACK =3D=3D C, Fortran
Lapackpp =3D=3D C++

That's the answer.

Christian

> Is it object orient and easy to use?
>
> On 2/4/07, Christian Stimming <sti...@tu...> wrote:
> > Dear Mr. Ozturk,
> >
> > no, lapackpp does not offer any iterative or sparse solvers.
> >
> > If you need such kind of solvers, I would suggest you should take a look
> > at
> > the original LAPACK solvers (available in Fortran). Lapackpp should only
> > provide a C++ wrapper to the more familiar parts of LAPACK, but your
> > specialized request should rather be solved by using the LAPACK functio=
ns
> > directly.
> >
> > Christian
> >
> > Am Sonntag, 4. Februar 2007 12:29 schrieb oz...@ee...:
> > > Dear Mr Christian Stimming,
> > >
> > > I am pursueing MSc. at Bilkent university in Turkey. In my thesis, I =
am
> > > solving large sparse linear equations(beyond 10000*10000). Currently I
> >
> > am
> >
> > > making use of lapackpp in .net2003. However as far as I know, all
> > > matrix structures and methods for solving equations are defined for
> > > dense
> >
> > matrices
> >
> > > which limits the size of the problem that I am able solve. Is there a=
ny
> > > patch or an addition to lapackpp for iterative and/or direct  sparse
> >
> > matrix
> >
> > > solvers? I would appreciate any help coming from you since such an
> >
> > ability
> >
> > > in solving sparse matrices enables me to write my paper and complete =
my
> > > thesis as soon as possible. My thesis is analyzing QoS in OBS networks
> > > using a novel numerical method.
> > >
> > >
> > > Kind regards,
> > >
> > > Onur =C3=96zt=C3=BCrk
> > > EEE @ Bilkent University
> > > Ankara, Turkey

Re: [Lapackpp-devel] lapackpp@Bilkent University

[Christian S. <sti...@tu...>]

Dear Mr. Ozturk,

no, lapackpp does not offer any iterative or sparse solvers.

If you need such kind of solvers, I would suggest you should take a look at=
=20
the original LAPACK solvers (available in Fortran). Lapackpp should only=20
provide a C++ wrapper to the more familiar parts of LAPACK, but your=20
specialized request should rather be solved by using the LAPACK functions=20
directly.

Christian

Am Sonntag, 4. Februar 2007 12:29 schrieb oz...@ee...:
> Dear Mr Christian Stimming,
>
> I am pursueing MSc. at Bilkent university in Turkey. In my thesis, I am
> solving large sparse linear equations(beyond 10000*10000). Currently I am
> making use of lapackpp in .net2003. However as far as I know, all matrix
> structures and methods for solving equations are defined for dense matric=
es
> which limits the size of the problem that I am able solve. Is there any
> patch or an addition to lapackpp for iterative and/or direct  sparse matr=
ix
> solvers? I would appreciate any help coming from you since such an ability
> in solving sparse matrices enables me to write my paper and complete my
> thesis as soon as possible. My thesis is analyzing QoS in OBS networks
> using a novel numerical method.
>
>
> Kind regards,
>
> Onur =D6zt=FCrk
> EEE @ Bilkent University
> Ankara, Turkey

Re: [Lapackpp-devel] Problem With EigenVector Calculation for a 3x3 Matrix

[sun <pyt...@gm...>]

Hi, first we know that both results of Lapackpp and Matlab are correct
although sometime some of eigenvectors become mirrored. So in my point of
view, that is not a problem of Lapackpp.

As you concern, if I understand correctly, you are not happy with the
negative sign of eigenvectors.
I think you can self-control the results: Just get the eigenvector of
minimal eigenvalue and then do a simple judgement, if the eigenvector has a
positive sign, it is OK. Otherwise you just reverse the sign by yourself.

sun

On 2/4/07, Satyajit Sarangi <ssa...@gm...> wrote:
>
> Hi Sun,
>           Thanx for your reply. However, if you check the eigen vectors
> you will find that there is an arbitrary change of sign in 1st and 2nd eigen
> vectors which I have shown. also there is a problem that one matrix is kind
> of a mirror image of the other however, with some more negative signs. I am
> using the eigen vectors to estimate normals for a Point in 3D space. So I
> have to use the eigen vector corresponding to the least eigen value. This
> requires me to take the x,y,z value of the eigen vector. So even if the
> vector is a mirror image then it becomes a problem for me. So please comment
> on this. I had noticed the precision but that is not my problem here.
> Regards
> Satyajit

Re: [Lapackpp-devel] Problem With EigenVector Calculation for a 3x3 Matrix

[sun <pyt...@gm...>]

Yes, what you have gotten just show that the results from Lapackpp and
Matlab are same.
The only difference is that the results display, i.e., the default display
precision of number between Lapackpp and Matlab is different.

Please type 'format long' in Matlab to set to long precision and then take a
look at the results.


sun
On 2/4/07, Satyajit Sarangi <ssa...@gm...> wrote:
>
> Hi All,
>       I am trying to calculate the EigenVectors for a 3x3 matrix. However,
> I get different results from Matlab and LAPACKPP. I am supposed to use the
> Eigen Vector directly for my calculations and hence wanted to know where I
> might be going wrong. I am giving a code snippet which I am using for the
> calculation.
>
> LaGenMatDouble CoVarianceMatrix(3,3);  // Create a 3x3 Matrix
>
>     LaVectorDouble EIG_VAL(3);
>     LaVectorDouble EIG_VAL_IM(3);
>     LaVectorDouble EIGVECS;
>
> // Some Code which creates the CoVariance Matrix.
>
>         LaEigSolve(CoVarianceMatrix, EIG_VAL, EIG_VAL_IM, EIGVECS);    //
> To solve for the Eigen Vectors and Eigen Values
>
> ////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
>
> // This comparison belows shows the difference between my results and
> Matlab's results
> ////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
>
>
> LAPACKPP Output:
>
> CoVariance Matrix
> 242.265 50.7184 13.7922
> 50.7184 14.1482 20.3449
> 13.7922 20.3449 279.355
>
> Eigen Values: (From LAPACKPP)
> 2.32756
> 244.362
> 289.079
>
> Eigen Values: (From Matlab)
>     2.3275     0         0
>          0     244.3618      0
>          0         0        289.0789
>
>
> Eigen Vectors: ( From LAPACKPP)
> 0.203024   -0.884631  0.419774
> -0.97723  -0.156061  0.143755
> 0.06166  0.439401  0.896172
>
> Eigen Vectors: ( From Matlab)
>     0.2030    0.8846    0.4198
>    -0.9772    0.1561    0.1438
>     0.0617   -0.4394    0.8962
>
>
>
> ===========================================
>
> CoVariance Matrix
> 356.18 3.72945 31.3579
> 3.72945 20.3312 72.4796
> 31.3579 72.4796 342.337
>
> Eigen Values: (From LAPACKPP)
> 388.504
> 325.6
> 4.74432
>
> Eigen Values: (From Matlab)
>     4.7443         0         0
>          0  325.5996         0
>          0         0  388.5043
>
> Eigen Vectors: (From LAPACK)
> 0.697496  0.716539  -0.00842256
> 0.145208  -0.15284  -0.977525
> 0.701722  -0.680597  0.210652
>
> Eigen Vectors: (From Matlab)
>    -0.0084   -0.7165   -0.6975
>    -0.9775    0.1528   -0.1452
>     0.2107    0.6806   -0.7017
>
> I would really appreciate it if someone can help me out with this as I
> need it urgently.
> Regards
> Satyajit
>
>
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job
> easier.
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> lapackpp-devel mailing list
> lap...@li...
> https://lists.sourceforge.net/lists/listinfo/lapackpp-devel
>
>

[Lapackpp-devel] Problem With EigenVector Calculation for a 3x3 Matrix

[Satyajit S. <ssa...@gm...>]

Hi All,
      I am trying to calculate the EigenVectors for a 3x3 matrix. However, I
get different results from Matlab and LAPACKPP. I am supposed to use the
Eigen Vector directly for my calculations and hence wanted to know where I
might be going wrong. I am giving a code snippet which I am using for the
calculation.

LaGenMatDouble CoVarianceMatrix(3,3);  // Create a 3x3 Matrix

    LaVectorDouble EIG_VAL(3);
    LaVectorDouble EIG_VAL_IM(3);
    LaVectorDouble EIGVECS;

// Some Code which creates the CoVariance Matrix.

        LaEigSolve(CoVarianceMatrix, EIG_VAL, EIG_VAL_IM, EIGVECS);    // To
solve for the Eigen Vectors and Eigen Values

////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
// This comparison belows shows the difference between my results and
Matlab's results
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////

LAPACKPP Output:

CoVariance Matrix
242.265 50.7184 13.7922
50.7184 14.1482 20.3449
13.7922 20.3449 279.355

Eigen Values: (From LAPACKPP)
2.32756
244.362
289.079

Eigen Values: (From Matlab)
    2.3275     0         0
         0     244.3618      0
         0         0        289.0789


Eigen Vectors: ( From LAPACKPP)
0.203024  -0.884631  0.419774
-0.97723  -0.156061  0.143755
0.06166  0.439401  0.896172

Eigen Vectors: ( From Matlab)
    0.2030    0.8846    0.4198
   -0.9772    0.1561    0.1438
    0.0617   -0.4394    0.8962



===========================================

CoVariance Matrix
356.18 3.72945 31.3579
3.72945 20.3312 72.4796
31.3579 72.4796 342.337

Eigen Values: (From LAPACKPP)
388.504
325.6
4.74432

Eigen Values: (From Matlab)
    4.7443         0         0
         0  325.5996         0
         0         0  388.5043

Eigen Vectors: (From LAPACK)
0.697496  0.716539  -0.00842256
0.145208  -0.15284  -0.977525
0.701722  -0.680597  0.210652

Eigen Vectors: (From Matlab)
   -0.0084   -0.7165   -0.6975
   -0.9775    0.1528   -0.1452
    0.2107    0.6806   -0.7017

I would really appreciate it if someone can help me out with this as I need
it urgently.
Regards
Satyajit

Re: [Lapackpp-devel] [Lapackpp-cvs] CVS update hangs

[Christian S. <sti...@tu...>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Dominik Wagenfuehr schrieb:
> Sorry for the second answer....
> 
> I have deleted my old testing-folder and it hangs at
> 
> U lapackpp/testing/tutil.cc
> 
> Maybe there is something wrong with this file.

No, I don't think so. I currently don't have any problems with cvs
update or cvs checkout from the developer cvs
:ext:cs...@la...:/cvsroot/lapackpp but
sourceforge regularly can have server problems. Just try again sometime
later.

Christian
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.1 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iQCVAwUBRb8Y6GXAi+BfhivFAQICWAP+NJemiEnzhfdg1ZvaIJI+FCnxReFe1hcQ
vrzg6K9Fw0R6y0xuHqg0GeikuPEoYo9UgOovWFuzXAcc5fYQMbJr9A2OMPrXltd9
LoCLWP4T6qJZI8ex1FuTo9jDaHBA/oqSLMmTCmPRJ12988sRrjN20SJUM/34e1NK
BAZ+VL5woaw=
=qM+q
-----END PGP SIGNATURE-----

Re: [Lapackpp-devel] [Lapackpp-cvs] CVS update hangs

[Dominik W. <dom...@ar...>]

Sorry for the second answer....

I have deleted my old testing-folder and it hangs at

U lapackpp/testing/tutil.cc

Maybe there is something wrong with this file.

Greetings, Dominik


Christian Stimming schrieb:
> Dominik Wagenfuehr schrieb:
> >> Hello Christian,
> >>
> >>> In your particular case you should use the "developer cvs" access
> (which
> >>> you have because you're a developer member of lapackpp)
> >> Maybe a stupid question, but what is the site password or where can I
> >> get it? The CVS site http://sourceforge.net/cvs/?group_id=99696
> does not
> >> say much about this.
>
> That's your sourceforge-wide password for your username. The same
> password that you use to login to the sf.net website.
>
> Christian

Re: [Lapackpp-devel] [Lapackpp-cvs] CVS update hangs

[Dominik W. <dom...@ar...>]

Okay, thanks....

Unfortunately it still hangs at

cvs checkout: Updating lapackpp/testing

even if I use the developer CVS. Did it work on your machine?

Greetings, Dominik


Christian Stimming schrieb:
> Dominik Wagenfuehr schrieb:
> >> Hello Christian,
> >>
> >>> In your particular case you should use the "developer cvs" access
> (which
> >>> you have because you're a developer member of lapackpp)
> >> Maybe a stupid question, but what is the site password or where can I
> >> get it? The CVS site http://sourceforge.net/cvs/?group_id=99696
> does not
> >> say much about this.
>
> That's your sourceforge-wide password for your username. The same
> password that you use to login to the sf.net website.
>
> Christian

Re: [Lapackpp-devel] [Lapackpp-cvs] CVS update hangs

[Christian S. <sti...@tu...>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Dominik Wagenfuehr schrieb:
> Hello Christian,
> 
>> In your particular case you should use the "developer cvs" access (which
>> you have because you're a developer member of lapackpp)
> 
> Maybe a stupid question, but what is the site password or where can I
> get it? The CVS site http://sourceforge.net/cvs/?group_id=99696 does not
> say much about this.

That's your sourceforge-wide password for your username. The same
password that you use to login to the sf.net website.

Christian
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.1 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iQCVAwUBRb8SKWXAi+BfhivFAQLKjwQAgirbMZh9RuKGz24ZL5BIzCU4oxUCjQSH
dL/wd3iByyYkv7q0JSWoq/f3VsEwbZ9imfT/oIrZ54mzH4YcmZE0D8FBB4UwO4CB
wompQ81jMDypODCskIEnBemjqRGoPZnqHkoPQYXuqcIHW050XXzUgR/V56X6cG2r
qjMzZo1T890=
=uUHq
-----END PGP SIGNATURE-----

Re: [Lapackpp-devel] [Lapackpp-cvs] CVS update hangs

[Dominik W. <dom...@ar...>]

Hello Christian,

> In your particular case you should use the "developer cvs" access (which
> you have because you're a developer member of lapackpp)

Maybe a stupid question, but what is the site password or where can I
get it? The CVS site http://sourceforge.net/cvs/?group_id=99696 does not
say much about this.

Greetings, Dominik

Re: [Lapackpp-devel] [Lapackpp-cvs] CVS update hangs

[Christian S. <sti...@tu...>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Dominik,

Dominik Wagenfuehr schrieb:
> I try to get the latest CVS files per
> 
> cvs -d:pserver:ano...@la...:/cvsroot/lapackpp
> login
> cvs -z3
> -d:pserver:ano...@la...:/cvsroot/lapackpp co
> -P lapackpp
> 
> But at
> 
> cvs checkout: Updating lapackpp/testing
> 
> it hangs and won't go on. Do you know this problem?

usually this just means sourceforge's cvs servers have a problem. This
can very well last 1-2 days until it is resolved.

In your particular case you should use the "developer cvs" access (which
you have because you're a developer member of lapackpp), because those
servers are usually maintained separately from the anonymous cvs access.

Regards,

Christian
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.1 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iQCVAwUBRb8JgWXAi+BfhivFAQIsbAQAnG1QqwzS6xMo2DXeY5t+CoVZBEAk3cPV
InQsh1BlnEJouzhQYNl3CR5U2Sr6i3lWUx5X4N+ny13QFhvf/XtyqBpjSye6LzjV
3SrYOB/d3T+JjSEB55nQY12xh2kd80NepLlQa6QqwPnJdBCSgNGH9eZFnFQdrjZf
/Wlgmeyxw7w=
=1c2c
-----END PGP SIGNATURE-----