[Arpack-ng] arpack-ng 3.1.4 - PI release

classic Classic list List threaded Threaded
6 messages Options
Sylvestre Ledru-4 Sylvestre Ledru-4
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

[Arpack-ng] arpack-ng 3.1.4 - PI release

Hello,

I am happy to announce the new release of arpack-ng. Many thanks to all
the patches received and bug reports.

  * libparpack2: missing dependency on MPI:
    http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=718790

  * Replace LAPACK second function with ARPACK's own arscnd in PARPACK

  * Fix issue #1259
    <http://forge.scilab.org/index.php/p/arpack-ng/issues/1259/> in
    DSEUPD and SSEUPD The Ritz vector purification step assumes
    workl(iq) still contains the original Q matrix. This is however
    overwritten by the call to xGEQR2 earlier. This patch fixes the
    issue by making a copy of the last row of the eigenvector matrix,
    after it is recomputed after QR by xORM2R. The work space
    WORKL(IW+NCV:IW+2*NCV) is not used later in the routine, and can be
    used for this.

  * Use configure supplied blas and lapack in the pkg-config. Thanks to
    Ward Poelmans (Closes: #1320
    <http://forge.scilab.org/index.php/p/arpack-ng/issues/1320/> )

  * Switch to automake 1.14 + libtool 2.4.2. Thanks to Ward Poelmans
    (Closes: #1321
    <http://forge.scilab.org/index.php/p/arpack-ng/issues/1321/> )

  * dseupd routine may lead to a segmentation fault Thanks to Edouard
    Canot (Closes: #1323
    <http://forge.scilab.org/index.php/p/arpack-ng/issues/1323/> )

  * dsaupd and 'BE' option returns wrong eigenvalues for a SPD matrix
    Thanks to Edouard Canot (Closes: #1329
    <http://forge.scilab.org/index.php/p/arpack-ng/issues/1329/> )

Download:
http://forge.scilab.org/index.php/p/arpack-ng/downloads/650/

This release is going to be available in a few hours in Debian unstable.

Sylvestre

_______________________________________________
Arpack-ng mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/arpack-ng
Sylvestre Ledru-4 Sylvestre Ledru-4
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: arpack-ng 3.1.4 - PI release

Le 13/11/2013 10:04, Sylvestre Ledru a écrit :

> Hello,
>
> I am happy to announce the new release of arpack-ng. Many thanks to all
> the patches received and bug reports.
>
>   * libparpack2: missing dependency on MPI:
>     http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=718790
>
>   * Replace LAPACK second function with ARPACK's own arscnd in PARPACK
>
>   * Fix issue #1259
>     <http://forge.scilab.org/index.php/p/arpack-ng/issues/1259/> in
>     DSEUPD and SSEUPD The Ritz vector purification step assumes
>     workl(iq) still contains the original Q matrix. This is however
>     overwritten by the call to xGEQR2 earlier. This patch fixes the
>     issue by making a copy of the last row of the eigenvector matrix,
>     after it is recomputed after QR by xORM2R. The work space
>     WORKL(IW+NCV:IW+2*NCV) is not used later in the routine, and can be
>     used for this.
>
>   * Use configure supplied blas and lapack in the pkg-config. Thanks to
>     Ward Poelmans (Closes: #1320
>     <http://forge.scilab.org/index.php/p/arpack-ng/issues/1320/> )
>
>   * Switch to automake 1.14 + libtool 2.4.2. Thanks to Ward Poelmans
>     (Closes: #1321
>     <http://forge.scilab.org/index.php/p/arpack-ng/issues/1321/> )
>
>   * dseupd routine may lead to a segmentation fault Thanks to Edouard
>     Canot (Closes: #1323
>     <http://forge.scilab.org/index.php/p/arpack-ng/issues/1323/> )
>
>   * dsaupd and 'BE' option returns wrong eigenvalues for a SPD matrix
>     Thanks to Edouard Canot (Closes: #1329
>     <http://forge.scilab.org/index.php/p/arpack-ng/issues/1329/> )
>
> Download:
> http://forge.scilab.org/index.php/p/arpack-ng/downloads/650/
Actual link:
http://forge.scilab.org/index.php/p/arpack-ng/downloads/651/

S

_______________________________________________
Arpack-ng mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/arpack-ng
Samuel Halliday Samuel Halliday
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: arpack-ng 3.1.4 - PI release

In reply to this post by Sylvestre Ledru-4
What happens if I use dynamically linked LAPACK/BLAS from, say, ATLAS? I hope you changed the name of the function, if you're expecting different behaviour.

Kind regards,
Sam Halliday

--
Sent from my iPhone

> On 13 Nov 2013, at 09:04, Sylvestre Ledru <[hidden email]> wrote:
>
> Hello,
>
> I am happy to announce the new release of arpack-ng. Many thanks to all
> the patches received and bug reports.
>
>  * libparpack2: missing dependency on MPI:
>    http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=718790
>
>  * Replace LAPACK second function with ARPACK's own arscnd in PARPACK
>
>  * Fix issue #1259
>    <http://forge.scilab.org/index.php/p/arpack-ng/issues/1259/> in
>    DSEUPD and SSEUPD The Ritz vector purification step assumes
>    workl(iq) still contains the original Q matrix. This is however
>    overwritten by the call to xGEQR2 earlier. This patch fixes the
>    issue by making a copy of the last row of the eigenvector matrix,
>    after it is recomputed after QR by xORM2R. The work space
>    WORKL(IW+NCV:IW+2*NCV) is not used later in the routine, and can be
>    used for this.
>
>  * Use configure supplied blas and lapack in the pkg-config. Thanks to
>    Ward Poelmans (Closes: #1320
>    <http://forge.scilab.org/index.php/p/arpack-ng/issues/1320/> )
>
>  * Switch to automake 1.14 + libtool 2.4.2. Thanks to Ward Poelmans
>    (Closes: #1321
>    <http://forge.scilab.org/index.php/p/arpack-ng/issues/1321/> )
>
>  * dseupd routine may lead to a segmentation fault Thanks to Edouard
>    Canot (Closes: #1323
>    <http://forge.scilab.org/index.php/p/arpack-ng/issues/1323/> )
>
>  * dsaupd and 'BE' option returns wrong eigenvalues for a SPD matrix
>    Thanks to Edouard Canot (Closes: #1329
>    <http://forge.scilab.org/index.php/p/arpack-ng/issues/1329/> )
>
> Download:
> http://forge.scilab.org/index.php/p/arpack-ng/downloads/650/
>
> This release is going to be available in a few hours in Debian unstable.
>
> Sylvestre
>
> _______________________________________________
> Arpack-ng mailing list
> [hidden email]
> http://lists.scilab.org/mailman/listinfo/arpack-ng
_______________________________________________
Arpack-ng mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/arpack-ng
Jordi Gutiérrez Hermoso Jordi Gutiérrez Hermoso
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: arpack-ng 3.1.4 - PI release

mOn Wed, 2013-11-13 at 09:33 +0000, Sam Halliday wrote:

> What happens if I use dynamically linked LAPACK/BLAS from, say,
> ATLAS? I hope you changed the name of the function, if you're
> expecting different behaviour.

What part are you referring to? I can't tell what you're replying to
when you're top-posting.

    https://en.wikipedia.org/wiki/Top_posting

But just guessing, if you mean about calling ARPACK's own arcsnd
instead of second from LAPACK? Yeah, there's no nameclash.

> Sent from my iPhone

Sent from my Emacs.

- Jordi G. H.



_______________________________________________
Arpack-ng mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/arpack-ng
rudimeier rudimeier
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [Arpack-ng] arpack-ng 3.1.4 - PI release

In reply to this post by Sylvestre Ledru-4
Hi,

is this 3.1.4 release already final? I wonder why there is no tag in the git repo.
Also in configure.ac is still 3.1.3 set.

BTW the new tar ball name arpack-ng_3.1.4.tar.gz with underscore instead of minus indicates that it was somehow manually created instead of using "make dist". Make dist would have also shown you the wrong version number.

cu,
Rudi
Sylvestre Ledru-3 Sylvestre Ledru-3
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: arpack-ng 3.1.4 - PI release

Right, sorry. That is fixed in 3.1.5
Thanks,
S
On 10/12/2013 19:50, rudimeier wrote:

> Hi,
>
> is this 3.1.4 release already final? I wonder why there is no tag in the git
> repo.
> Also in configure.ac is still 3.1.3 set.
>
> BTW the new tar ball name arpack-ng_3.1.4.tar.gz with underscore instead of
> minus indicates that it was somehow manually created instead of using "make
> dist". Make dist would have also shown you the wrong version number.
>
> cu,
> Rudi
>
>
>
> --
> View this message in context: http://mailinglists.scilab.org/Arpack-ng-arpack-ng-3-1-4-PI-release-tp4027818p4028006.html
> Sent from the Arpack-ng mailing list archive at Nabble.com.
> _______________________________________________
> Arpack-ng mailing list
> [hidden email]
> http://lists.scilab.org/mailman/listinfo/arpack-ng

_______________________________________________
Arpack-ng mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/arpack-ng
Loading...