mpqc-users — A mailing list for MPQC users.

[Mpqc-users] WARNING: SCF::compute_vector(): unlink of temporary checkpoint file

[Sambasiva R. C. <chi...@in...>]

Dear all,

I am on Ubuntu 12.04 LTS 64 bit
 mpqc/2.3.1-openmpi-1.7.5


I have the following problem to run HF calculations.

 iter     1 energy =  -39.3112895219 delta = 8.68295e-02
                197088 integrals
  iter     2 energy =  -40.1591399225 delta = 3.80776e-02
StateOutBin::put_array_void: failed
link of temporary checkpoint file
           "/data/chinnams/TestMpqc/BondFragment2.wfn.1.tmp" failed with
error: No such file or directory
                197088 integrals
  iter     3 energy =  -40.1938656865 delta = 1.21268e-02
  WARNING: SCF::compute_vector(): unlink of temporary checkpoint file
           "/data/chinnams/TestMpqc/BondFragment2.wfn.2.tmp" failed with
error: No such file or directory


Could you please help me to resolve this problem?


best regards
Rao

Re: [Mpqc-users] MPQC compiling problem

[Curtis J. <cl...@sa...>]

Hi Jemmy,

Here are a few suggestions that may or may not help with the problems
you are having:

1) Use --with-libdirs and --with-libs to chose the native libraries.

2) Examine config.log to see why the check for mpi.h failed.  Double
check your include path.

3) Do you get the undefined VTT for both PGI and Pathscale?  This looks
like a compiler bug I've seen long ago, but am surprised to see this
with a recent compiler.  Please post the list if you find a workaround.

4) Are you sure you are linking blas and lapack libraries compiled with
the same fortran compiler used for MPQC?

In general, it is easiest to link against MPI by using the compilation
scripts provided with the MPI distribution.  That way you don't have to
bother getting the right include directories and library
names/directories. I've posted a few examples of the configure flags
that I use to the MPQC wiki:

http://mpqc.wiki.sourceforge.net/

Curt

Jemmy Hu wrote:
> Hi All,
> 
> We tried to build MPQC with MPI on our linux clusters without success.
> One type of clusters is Opteron runs red hat based XC OS and hpmpi from HP
> with GNU, PathScale and PGI compilers available.
> Another is a alpha cluster with compaq compilers
> 
> The problems I observed are,
> 1) MPQC does not work with the native libraries such as acml on Opteron
> and cxml on alpha clusters. The configure log file tells:
> checking for daxpy_ in -lblas... no
> checking for daxpy_ in -lf77blas... no
> WARNING: Could not link to the BLAS library.  It can be obtained at
> http://www.netlib.org/blas.  Use --with-libdirs and/or --with-libs
> to specify the name of the library.
> configure: error: BLAS is required to complete the build
> 
> Instead, we compiled blas and lapack using g77 on the clusters.
> Then the problem turns to be
> 2) --with-include does not work for GNU compilers
> With this configure options:
> #!/bin/bash
> # run configure with options
> ./configure --with-cxx='g++' --with-f77='g77' --with-cc='gcc' 
> --prefix='/work/jemmyhu/mpqc' --with-include='-I/opt/hpmpi/include' 
> --with-libs='-lmpiCC -lmpi -lblas -llapack' 
> --with-libdirs='-L/opt/hpmpi/lib/linux_amd64 
> -L/work/jemmyhu/lapack-3.1.1' --enable-always-use-mpi
> 
> The configure error is:
> checking for backtrace_symbols_fd... yes
> checking for mpi.h... no
> configure: error: --enable-always-use-mpi is set but MPI is not available
> 
> Remove --enable-always-use-mpi, the configure is still
> checking for mpi.h... no
> but GNU works for MPQC for a serial version on Opteron cluster.
> 
> 3) Neither Pathscale (version 2.2.1) nor PGI (version 6.1) work for MPQC.
> They do work with the --with-include and mpi libraries:
> 
> The configure options are:
> #!/bin/bash
> # run configure with options
> ./configure --with-cxx='CC' --without-f77 --with-cc='cc' 
> --prefix='/work/jemmyhu/mpqc' --with-include='-I/opt/hpmpi/include' 
> --with-libs='-lmpiCC -lmpi -lblas -llapack' 
> --with-libdirs='-L/opt/hpmpi/lib/linux_amd64 
> -L/work/jemmyhu/lapack-3.1.1'
> 
> configure succeeded and mpi links are fine as
> checking for mpi.h... yes
> checking for MPI_Init... yes
> checking for MPI_File_open... yes
> checking for MPI_Init_thread... yes
> checking for MPI_Abort in -lmpi++... yes
> 
> however, the compile gives lots of errors such as
> ../../../lib/libSCwfn.a(hcorewfn.o)(.text+0x2b0): In function 
> `sc::HCoreWfn::HCoreWfn(sc::StateIn&)':
> : undefined reference to `VTT for sc::HCoreWfn'
> 
> 4) Strange thing for blas and lapack
> As stated above, compile blas and lapack using g77, then use GNU compilers
> to build MPQC is fine except for the mpi link problem.
> We also tried to compile blas and lapack on the cluster using pgf77 or
> pathf77
> without problems. However, to use those compilers to configure MPQC,
> it says that BLAS is not there:
> 
> checking for daxpy_ in -lblas... no
> checking for daxpy_ in -lf77blas... no
> WARNING: Could not link to the BLAS library.  It can be obtained at
> http://www.netlib.org/blas.  Use --with-libdirs and/or --with-libs
> to specify the name of the library.
> configure: error: BLAS is required to complete the build
> 
> As a result, we can only build a serial version with GNU.
> 
> Note that: all compilers and mpi libraries work well for standard
> c++ mpi codes on our systems.
> 
> Thanks for any helps in advance.
> --
> Jemmy Hu                   SHARCNET
> HPC Analyst                University of Waterloo
> (519)888-4567 x37854       PHY2004
> je...@sh...        http://www.sharcnet.ca
> 
> 
> 
> 
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc.
> Still grepping through log files to find problems?  Stop.
> Now Search log events and configuration files using AJAX and a browser.
> Download your FREE copy of Splunk now >> http://get.splunk.com/
> _______________________________________________
> Mpqc-users mailing list
> Mpq...@li...
> https://lists.sourceforge.net/lists/listinfo/mpqc-users
> 

-- 
Curtis Janssen, cl...@sa..., +1 925-294-1509
Sandia National Labs, MS 9158, PO Box 969, Livermore, CA 94551, USA

[Mpqc-users] MPQC compiling problem

[Jemmy Hu <j8...@sc...>]

Hi All,

We tried to build MPQC with MPI on our linux clusters without success.
One type of clusters is Opteron runs red hat based XC OS and hpmpi from HP
with GNU, PathScale and PGI compilers available.
Another is a alpha cluster with compaq compilers

The problems I observed are,
1) MPQC does not work with the native libraries such as acml on Opteron
and cxml on alpha clusters. The configure log file tells:
checking for daxpy_ in -lblas... no
checking for daxpy_ in -lf77blas... no
WARNING: Could not link to the BLAS library.  It can be obtained at
http://www.netlib.org/blas.  Use --with-libdirs and/or --with-libs
to specify the name of the library.
configure: error: BLAS is required to complete the build

Instead, we compiled blas and lapack using g77 on the clusters.
Then the problem turns to be
2) --with-include does not work for GNU compilers
With this configure options:
#!/bin/bash
# run configure with options
./configure --with-cxx=3D'g++' --with-f77=3D'g77' --with-cc=3D'gcc' =20
--prefix=3D'/work/jemmyhu/mpqc' --with-include=3D'-I/opt/hpmpi/include' =20
--with-libs=3D'-lmpiCC -lmpi -lblas -llapack' =20
--with-libdirs=3D'-L/opt/hpmpi/lib/linux_amd64 =20
-L/work/jemmyhu/lapack-3.1.1' --enable-always-use-mpi

The configure error is:
checking for backtrace_symbols_fd... yes
checking for mpi.h... no
configure: error: --enable-always-use-mpi is set but MPI is not available

Remove --enable-always-use-mpi, the configure is still
checking for mpi.h... no
but GNU works for MPQC for a serial version on Opteron cluster.

3) Neither Pathscale (version 2.2.1) nor PGI (version 6.1) work for MPQC.
They do work with the --with-include and mpi libraries:

The configure options are:
#!/bin/bash
# run configure with options
./configure --with-cxx=3D'CC' --without-f77 --with-cc=3D'cc' =20
--prefix=3D'/work/jemmyhu/mpqc' --with-include=3D'-I/opt/hpmpi/include' =20
--with-libs=3D'-lmpiCC -lmpi -lblas -llapack' =20
--with-libdirs=3D'-L/opt/hpmpi/lib/linux_amd64 =20
-L/work/jemmyhu/lapack-3.1.1'

configure succeeded and mpi links are fine as
checking for mpi.h... yes
checking for MPI_Init... yes
checking for MPI_File_open... yes
checking for MPI_Init_thread... yes
checking for MPI_Abort in -lmpi++... yes

however, the compile gives lots of errors such as
../../../lib/libSCwfn.a(hcorewfn.o)(.text+0x2b0): In function =20
`sc::HCoreWfn::HCoreWfn(sc::StateIn&)':
: undefined reference to `VTT for sc::HCoreWfn'

4) Strange thing for blas and lapack
As stated above, compile blas and lapack using g77, then use GNU compilers
to build MPQC is fine except for the mpi link problem.
We also tried to compile blas and lapack on the cluster using pgf77 or pathf=
77
without problems. However, to use those compilers to configure MPQC,
it says that BLAS is not there:

checking for daxpy_ in -lblas... no
checking for daxpy_ in -lf77blas... no
WARNING: Could not link to the BLAS library.  It can be obtained at
http://www.netlib.org/blas.  Use --with-libdirs and/or --with-libs
to specify the name of the library.
configure: error: BLAS is required to complete the build

As a result, we can only build a serial version with GNU.

Note that: all compilers and mpi libraries work well for standard
c++ mpi codes on our systems.

Thanks for any helps in advance.
--=20
Jemmy Hu                   SHARCNET
HPC Analyst                University of Waterloo
(519)888-4567 x37854       PHY2004
je...@sh...        http://www.sharcnet.ca

Re: [Mpqc-users] Problem running example mp2h2o.in on > 20 processors with MPQC 2.3.1

[Jens T. <j.m...@dl...>]

Hi Curtis,

Many thanks for confirming what the problem was. Unfortunately I don't 
have access to a thread-safe MPI on the system in question, but I now 
that I know what's happening I should be able to work round any problems.

Best wishes,

Jens


Curtis Janssen wrote:
> Hi Jens,
>
> Thanks for the detailed analysis.  This is a bug in MPQC.  Since the
> case you are running is really too small to use 32 processors, this
> problem doesn't arise too often.  I just checked the trunk and this
> problem exists there, too.  I submitted the bug and it is request 1777368.
>
> If you are using a thread-safe MPI you could try MTMPIMemoryGrp instead
> of ARMCIMemoryGrp.  On InfiniBand with OFED 1.2 and MPICH2 1.0.5p4 (with
> the TCP/IP transport using IPoIB) it works quite well and gives good
> performance.
>
> Curt
>
> Jens Thomas wrote:
>   
>> Hello,
>>
>> I'm trying to get to grips with MPQC on an cluster running Suse 9.3 on
>> Opteron processors and using SCORE (http://www.pccluster.org/).
>>
>> I built the code with AMRCI from version 4.0.6 of the Global Arrays,
>> using the "--with-default-parallel=armcimpi" option to the configure script.
>>
>> The validation for check0 ran fine on 4 processors (1 node), but hung
>> running the mp2 examples when I tried running on 32 processors.
>>
>> After a bit of playing about running mp2h2o.in, I think I've
>> tracked the problem down to the following:
>>
>> The code hangs just after the first SCF when it prints:
>>
>> Memory used for integral storage:       31826758 Bytes
>>
>> The problem manifests itself in the call:
>>
>> mem->set_localsize(size_t(nijmax)*nbasis*nbasis*sizeof(double));
>>
>> in the file:
>>
>> mpqc-2.3.1/src/lib/chemistry/qc/mbpt/csgrad.cc
>>
>> When run on 21 processors, processor 21 does not have any work to do so
>> localsize is called with 0.
>>
>> This call appears to be resolved in the file:
>>
>> mpqc-2.3.1/src/lib/util/group/memarmci.cc
>>
>> where, within the set_localsize function, there is a statement:
>>
>> if (localsize == 0) return;
>>
>> I think this is because set_localsize is called with an argument of 0 by
>> ARMCIMemoryGrp::finalize, and in this case the call just clears up the
>> memory and mutexes.
>>
>> However, in the case that hangs, only processor 21 returns at that
>> point, and the rest all proceed to the call:
>>
>> ARMCI_Create_mutexes(1);
>>
>> which is a collective operation and therefore hangs as processor 21
>> never makes this call.
>>
>> As I've only just started using MPQC and have only just got through the
>> first few pages through my first C++ book ;-) , I wanted to check
>> whether this is indeed a problem with the code or a problem with my
>> use/understanding of it.
>>
>> I'll happily supply any more information if that would be of use.
>>
>> Best wishes,
>>
>> Jens
>>
>>
>> --
>> ===================================================================
>> Jens Thomas,                email:  j.m...@dl...
>> STFC Daresbury Lab,         tel:    +44-1925-603849
>> Warrington,                 fax:    +44-1925-603634
>> WA4 4AD,  UK.               http:   http://www.cse.scitech.ac.uk
>> ===================================================================
>>
>>
>>
>> -------------------------------------------------------------------------
>> This SF.net email is sponsored by: Splunk Inc.
>> Still grepping through log files to find problems?  Stop.
>> Now Search log events and configuration files using AJAX and a browser.
>> Download your FREE copy of Splunk now >>  http://get.splunk.com/
>> _______________________________________________
>> Mpqc-users mailing list
>> Mpq...@li...
>> https://lists.sourceforge.net/lists/listinfo/mpqc-users
>>
>>     
>
>   


-- 
===================================================================
Jens Thomas,                email:  j.m...@dl...
STFC Daresbury Lab,         tel:    +44-1925-603849
Warrington,                 fax:    +44-1925-603634
WA4 4AD,  UK.               http:   http://www.cse.scitech.ac.uk
===================================================================

Re: [Mpqc-users] Problem running example mp2h2o.in on > 20 processors with MPQC 2.3.1

[Curtis J. <cl...@sa...>]

Hi Jens,

Thanks for the detailed analysis.  This is a bug in MPQC.  Since the
case you are running is really too small to use 32 processors, this
problem doesn't arise too often.  I just checked the trunk and this
problem exists there, too.  I submitted the bug and it is request 1777368.

If you are using a thread-safe MPI you could try MTMPIMemoryGrp instead
of ARMCIMemoryGrp.  On InfiniBand with OFED 1.2 and MPICH2 1.0.5p4 (with
the TCP/IP transport using IPoIB) it works quite well and gives good
performance.

Curt

Jens Thomas wrote:
> Hello,
> 
> I'm trying to get to grips with MPQC on an cluster running Suse 9.3 on
> Opteron processors and using SCORE (http://www.pccluster.org/).
> 
> I built the code with AMRCI from version 4.0.6 of the Global Arrays,
> using the "--with-default-parallel=armcimpi" option to the configure script.
> 
> The validation for check0 ran fine on 4 processors (1 node), but hung
> running the mp2 examples when I tried running on 32 processors.
> 
> After a bit of playing about running mp2h2o.in, I think I've
> tracked the problem down to the following:
> 
> The code hangs just after the first SCF when it prints:
> 
> Memory used for integral storage:       31826758 Bytes
> 
> The problem manifests itself in the call:
> 
> mem->set_localsize(size_t(nijmax)*nbasis*nbasis*sizeof(double));
> 
> in the file:
> 
> mpqc-2.3.1/src/lib/chemistry/qc/mbpt/csgrad.cc
> 
> When run on 21 processors, processor 21 does not have any work to do so
> localsize is called with 0.
> 
> This call appears to be resolved in the file:
> 
> mpqc-2.3.1/src/lib/util/group/memarmci.cc
> 
> where, within the set_localsize function, there is a statement:
> 
> if (localsize == 0) return;
> 
> I think this is because set_localsize is called with an argument of 0 by
> ARMCIMemoryGrp::finalize, and in this case the call just clears up the
> memory and mutexes.
> 
> However, in the case that hangs, only processor 21 returns at that
> point, and the rest all proceed to the call:
> 
> ARMCI_Create_mutexes(1);
> 
> which is a collective operation and therefore hangs as processor 21
> never makes this call.
> 
> As I've only just started using MPQC and have only just got through the
> first few pages through my first C++ book ;-) , I wanted to check
> whether this is indeed a problem with the code or a problem with my
> use/understanding of it.
> 
> I'll happily supply any more information if that would be of use.
> 
> Best wishes,
> 
> Jens
> 
> 
> --
> ===================================================================
> Jens Thomas,                email:  j.m...@dl...
> STFC Daresbury Lab,         tel:    +44-1925-603849
> Warrington,                 fax:    +44-1925-603634
> WA4 4AD,  UK.               http:   http://www.cse.scitech.ac.uk
> ===================================================================
> 
> 
> 
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc.
> Still grepping through log files to find problems?  Stop.
> Now Search log events and configuration files using AJAX and a browser.
> Download your FREE copy of Splunk now >>  http://get.splunk.com/
> _______________________________________________
> Mpqc-users mailing list
> Mpq...@li...
> https://lists.sourceforge.net/lists/listinfo/mpqc-users
> 

-- 
Curtis Janssen, cl...@sa..., +1 925-294-1509
Sandia National Labs, MS 9158, PO Box 969, Livermore, CA 94551, USA

[Mpqc-users] Problem running example mp2h2o.in on > 20 processors with MPQC 2.3.1

[Jens T. <j.m...@dl...>]

Hello,

I'm trying to get to grips with MPQC on an cluster running Suse 9.3 on
Opteron processors and using SCORE (http://www.pccluster.org/).

I built the code with AMRCI from version 4.0.6 of the Global Arrays,
using the "--with-default-parallel=armcimpi" option to the configure script.

The validation for check0 ran fine on 4 processors (1 node), but hung
running the mp2 examples when I tried running on 32 processors.

After a bit of playing about running mp2h2o.in, I think I've
tracked the problem down to the following:

The code hangs just after the first SCF when it prints:

Memory used for integral storage:       31826758 Bytes

The problem manifests itself in the call:

mem->set_localsize(size_t(nijmax)*nbasis*nbasis*sizeof(double));

in the file:

mpqc-2.3.1/src/lib/chemistry/qc/mbpt/csgrad.cc

When run on 21 processors, processor 21 does not have any work to do so
localsize is called with 0.

This call appears to be resolved in the file:

mpqc-2.3.1/src/lib/util/group/memarmci.cc

where, within the set_localsize function, there is a statement:

if (localsize == 0) return;

I think this is because set_localsize is called with an argument of 0 by
ARMCIMemoryGrp::finalize, and in this case the call just clears up the
memory and mutexes.

However, in the case that hangs, only processor 21 returns at that
point, and the rest all proceed to the call:

ARMCI_Create_mutexes(1);

which is a collective operation and therefore hangs as processor 21
never makes this call.

As I've only just started using MPQC and have only just got through the
first few pages through my first C++ book ;-) , I wanted to check
whether this is indeed a problem with the code or a problem with my
use/understanding of it.

I'll happily supply any more information if that would be of use.

Best wishes,

Jens


-- 
===================================================================
Jens Thomas,                email:  j.m...@dl...
STFC Daresbury Lab,         tel:    +44-1925-603849
Warrington,                 fax:    +44-1925-603634
WA4 4AD,  UK.               http:   http://www.cse.scitech.ac.uk
===================================================================

[Mpqc-users] User-defined grid in MPQC

[Joseph K. <jp...@sa...>]

User defined grids (unauthoritative):

Drilling down through the CLKS documentation it looks like this should
give you an arbitrary EulerMaclaurin/Lebedev integrator when included in
the CLKS section of a keyval input:

    integrator<RadialAngularIntegrator>:(
      radial<EulerMaclaurinRadialIntegrator>: (
        nr = 75
      )
      angular<LebedevLaikovIntegrator>:(
        n = 302
      )
    )

I'm not sure if there's a way to do something more general than this
(pruning?).  Any authoritative info out there?  

Joe


> > On Wed, 2007-05-09 at 14:19 -0600, rmuller wrote:
> >> Joe,
> >>
> >> Is there an easy way to define a user-defined DFT grid in MPQC? I
> >> thought that one way I could check my PyQuante grid is by running
> >> MPQC using the PyQuante grids.
> >>
> >> In flipping through the dft source, I see that it appears there are
> >> hooks for user-defined grids. Is it easy to do this?
> >>
> >> R.
> >>
> > -- 
> > Joseph P. Kenny
> > Sandia National Laboratories, Livermore CA
> > Scalable Computing Research and Development
> >
> 
-- 
Joseph P. Kenny
Sandia National Laboratories, Livermore CA
Scalable Computing Research and Development

Re: [Mpqc-users] your VIAlewGRA

[Giada H. <na...@ma...>]

Good day,

Vi_gra $1, 80
Ci_lis $3, 00
Lev_tra $3, 35

http://www.progenyid%.com ( Important Remove "%" )

--
dipping his pen into some ink, and starting to write.
Next Monday,  he said as he scribbled, I am likely to develop a cough,
owing to the unlucky conjunction of Mars and Jupiter.  He looked up at

Re: [Mpqc-users] PHkocARMACY

[Teodosio C. <li...@fa...>]

Hi,

Purchase your medication from our site because it costs less.

http://www.vodrx*.com - Remove "*" to make the link working!

--

He forced down some dinner after Divination, then returned to the empty
classroom with Hermione, using the Invisibility Cloak to avoid the
teachers. They kept practicing until past midnight. They would have

Re: [Mpqc-users] ED354

[Aniruddha C. <rea...@ha...>]

Good day,

Viazzgra  $1, 80
Ciazzlis  $3, 00
Levizztra $3, 35

http://www.printeryml.*com ( Important ! Remove "*" )

--
almost ordinary; their owners had clearly tried to make them as
Muggle-like as possible, but had slipped up by adding chimneys, or
bellpulls, or weather vanes. However, here and there was a tent so

[Mpqc-users] Restart with both .wfn and .ckpt files

[Brian H. <nsi...@ya...>]

Hi,
While performing a geometry optimization, I see that I
can restart from a saved Optimize object,  OR,  while
performing an energy calculation I can restart from a
saved MolecularEnergy object. What I'd like to do,
while performing an optimization, is restore the saved
Optimize object AND the saved MolecularEnergy object.

For example, I've finished 5 optimize iterations and
am on the 50th energy calc iteration when I kill the
process. Now I'd like to restart with my 5th iteration
optimize data PLUS my 50th iteration energy calc
state.

If what I want makes sense, would hacking the mpqc.cc
code to (basically) do the following work?

if (haveCkptFile && haveWfnFile)
  opt << SavableState::key_restore_state(ckptStateIn,
"opt");
  mole << SavableState::key_restore_state(wfnStateIn,
"mole");
  opt->set_function(mole.pointer());

Thanks,
--Brian.



 
____________________________________________________________________________________
No need to miss a message. Get email on-the-go 
with Yahoo! Mail for Mobile. Get started.
http://mobile.yahoo.com/mail 

Re: [Mpqc-users] dual-core amd question

[<lso...@cs...>]

On Thu, Jan 18, 2007 at 01:33:22PM -0800, Francesco Pietra wrote:
> Thank you. Then, one additional question that I forgot
> before. Planning to replace the 1GB modules with 2GB
> modules, is that correct to replace couples of modules
> (one left, one right) at different times, and not all
> eigth modules at the same time? I plan to use the
> machine with a mixture of 1G-1GB and 2GB-2GB couples
> of memory modules until I get money to replace all.

Follow the manuals instructions and make sure you keep the memory in
each bank on a cpu matches.  So you can have 2 x 1GB on one cpu and 2 x
2GB on another and that would be fine.  Having 1GB + 2GB banks on one
cpu will hurt performance however since it can only run the first 1GB
dual channel, and the second 1GB (on the 2GB bank) would be single
channel access.  Each CPU has two memory banks, usually with 2 or 4
memory sockets per bank.  Just keep the size of each bank on a given cpu
identical.  The motherboard user manual should tell you exactly how to
put memory in to keep the pairs matched up.

--
Len Sorensen

Re: [Mpqc-users] dual-core amd question

[<lso...@cs...>]

On Thu, Jan 18, 2007 at 11:12:51AM -0800, Francesco Pietra wrote:
> Question about MPQC job, with command for
> multiprocessor with POSIX threads:
> 
>  -thread "PthreadThreadGrp>:(num_threads = 4)"
> 
> Hardware: Tyan S2895 K8WE with all eigth memory
> modules; two dual-core opteron cpus. OS: debian amd64
> etch.

A dual core opteron has two complete opteron processors in one chip.
With two chips you have a total of 4 cores so you have 4 CPUs.

> The output of a MPQC job says (than the job completes
> successfully):
> 
> Using ShmMessageGrp for message passing (number of
> nodes = 4).
>   Using PthreadThreadGrp for threading (number of
> threads = 1).
>   Using ShmMemoryGrp for distributed shared memory.
>   Total number of processors = 4
> 
> Why 4 processors? Moreover, with such a system is that
> command optimized or should MPI used instead? In other
> words, is my arrangement "shared memory" or should be
> considered a cluster? 
> 
> That because someone has recently warned me that
> "dual-opteron is not shared memory [I knew that of
> course].  Each cpu has its own memory, and they can
> access each other's. If your program uses MPI then
> that's the best way of having it."

On most opteron systems, all the CPUs are connected to each other, and
the ram connected to various CPUs, and the extra time to access memory
attached to another cpu is only bus cycle per extra cpu, so in your case
at most one extra cycle.  No message passing algorithm will ever get
anywhere close to the performance of just accessing the memory on a
system like yours so shared memory between threads/processes is as fast
as it can get in your case.

> I had no doubt, until now, that the above arrangement
> of memory modules makes my system shared memory.
> Cannot try MPI because the progran is not compiled for
> clusters.

You do not have a cluster.  Your CPUs are as tightly linked as they can
realisticly be.  As far as I recall the interconnect between the two
CPUs runs at 4GB/s each way, which should be plenty for passing data
between threads.  I don't think the memory directly connected to each
cpu is that much faster than that.

--
Len Sorensen

Re: [Mpqc-users] dual-core amd question

[Francesco P. <chi...@ya...>]

Thank you. Then, one additional question that I forgot
before. Planning to replace the 1GB modules with 2GB
modules, is that correct to replace couples of modules
(one left, one right) at different times, and not all
eigth modules at the same time? I plan to use the
machine with a mixture of 1G-1GB and 2GB-2GB couples
of memory modules until I get money to replace all.

francesco pietra 


--- Jo Shields <jm...@os...> wrote:

> On Thu, 2007-01-18 at 11:12 -0800, Francesco Pietra
> wrote:
> > Why 4 processors? Moreover, with such a system is
> that
> > command optimized or should MPI used instead? In
> other
> > words, is my arrangement "shared memory" or should
> be
> > considered a cluster? 
> 
> 4 processors because 2 x 2 = 4
> 
> > That because someone has recently warned me that
> > "dual-opteron is not shared memory [I knew that of
> > course].  Each cpu has its own memory, and they
> can
> > access each other's. If your program uses MPI then
> > that's the best way of having it."
> 
> Dual Opteron is definitely shared memory in the
> modern sense of the term
> - it's actually NUMA, which means each processor has
> its own supply of
> memory but can request memory from other processors
> (whereas with
> 'traditional' SMP, there's one big pool of memory
> and one pool of
> processors). Massive 'SMP' systems such as SGI's
> Itanium-powered Altix
> are NUMA.
> 
> Performance, however, is a bigger question. Is there
> a hit when
> accessing a different processor's memory? Yes. Would
> MPI be faster?
> Possibly, but probably not on a 2x2 machine. Opteron
> suffers from
> diminishing returns (especially with a 2.6 kernel),
> so adding more cores
> takes you further & further from optimal performance
> - meaning MPI
> becomes a more interesting prospect at that point.
> 
> --Jo Shields
> 
> 



 
____________________________________________________________________________________
Food fight? Enjoy some healthy debate 
in the Yahoo! Answers Food & Drink Q&A.
http://answers.yahoo.com/dir/?link=list&sid=396545367

Re: [Mpqc-users] dual-core amd question

[Jo S. <jm...@os...>]

On Thu, 2007-01-18 at 11:12 -0800, Francesco Pietra wrote:
> Why 4 processors? Moreover, with such a system is that
> command optimized or should MPI used instead? In other
> words, is my arrangement "shared memory" or should be
> considered a cluster? 

4 processors because 2 x 2 = 4

> That because someone has recently warned me that
> "dual-opteron is not shared memory [I knew that of
> course].  Each cpu has its own memory, and they can
> access each other's. If your program uses MPI then
> that's the best way of having it."

Dual Opteron is definitely shared memory in the modern sense of the term
- it's actually NUMA, which means each processor has its own supply of
memory but can request memory from other processors (whereas with
'traditional' SMP, there's one big pool of memory and one pool of
processors). Massive 'SMP' systems such as SGI's Itanium-powered Altix
are NUMA.

Performance, however, is a bigger question. Is there a hit when
accessing a different processor's memory? Yes. Would MPI be faster?
Possibly, but probably not on a 2x2 machine. Opteron suffers from
diminishing returns (especially with a 2.6 kernel), so adding more cores
takes you further & further from optimal performance - meaning MPI
becomes a more interesting prospect at that point.

--Jo Shields

[Mpqc-users] dual-core amd question

[Francesco P. <chi...@ya...>]

Question about MPQC job, with command for
multiprocessor with POSIX threads:

 -thread "PthreadThreadGrp>:(num_threads = 4)"

Hardware: Tyan S2895 K8WE with all eigth memory
modules; two dual-core opteron cpus. OS: debian amd64
etch.

The output of a MPQC job says (than the job completes
successfully):

Using ShmMessageGrp for message passing (number of
nodes = 4).
  Using PthreadThreadGrp for threading (number of
threads = 1).
  Using ShmMemoryGrp for distributed shared memory.
  Total number of processors = 4

Why 4 processors? Moreover, with such a system is that
command optimized or should MPI used instead? In other
words, is my arrangement "shared memory" or should be
considered a cluster? 

That because someone has recently warned me that
"dual-opteron is not shared memory [I knew that of
course].  Each cpu has its own memory, and they can
access each other's. If your program uses MPI then
that's the best way of having it."

I had no doubt, until now, that the above arrangement
of memory modules makes my system shared memory.
Cannot try MPI because the progran is not compiled for
clusters.

Thanks for your kind attention.

francesco pietra


 
____________________________________________________________________________________
The fish are biting. 
Get more visitors on your site using Yahoo! Search Marketing.
http://searchmarketing.yahoo.com/arp/sponsoredsearch_v2.php

Re: [Mpqc-users] your uncovere

[Jameel L. <be...@da...>]

Hi
=20
AM_MBlEN $ 2. 90
VA_ALlUM $ 1. 25
Vl_lAGRA $ 3. 30
Cl_lALlS $ 3. 75
XA_ANAX $ 1. 50
=20
http://www.vaserionkunyionpion.com
=20
=20
=20
leave now, you may lose the opportunity to see further than you have
ever=20
I dont want to see anything except a headache cure, said Harry.
He stood up. The class backed away. They all looked unnerved.

Re: [Mpqc-users] your jeste

[Alessio T. <pie...@ba...>]

AMB_BlEN $ 2, 90
VAL_LlUM $ 1, 25
VlA_AGRA $ 3, 30
ClA_ALlS $ 3, 75
XAN_NAX $ 1, 50
=20
http://www.afeunkiontungenrandesinkin.com
=20
=20
=20
There was a pleasant feeling of anticipation in the air that day. Nobody
was very attentive in lessons, being much more interested in the arrival
that evening of the people from Beauxbatons and Durmstrang; even Potions

Re: [Mpqc-users] blaze Vtaagra

[Carmelo B. <ol...@ja...>]

V / A A G R A $ 3. 30=20
more info at http://www.unhujadesewuinderun.com

=20
=20
How are you supposed to defend yourself against something youve never
seen? A wizard whos about to put an illegal curse on you isnt going to
tell you what hes about to do. Hes not going to do it nice and polite

Re: [Mpqc-users] thras v/aggra

[Hampus L. <blu...@va...>]

V/AGGRA $ 3. 30=20
and 200 more items http://www.xaserunhfneguankuntynfans.com

=20
=20
But if they were the Death Eaters, why did they Disapparate when they
saw the Dark Mark? said Ron. Theyd have been pleased to see it,
wouldnt they?=20

Re: [Mpqc-users] new ingravescen

[Jenni H. <del...@ma...>]

VfAGRA from $3. 30 and other goods.

http://www.paleruijidefuhnswerun.com
=20
Wed just like to congratulate Mr. Bagman on his splendid performance
for England in the Quidditch match against Turkey last Saturday, the
witch said breathlessly.

Re: [Mpqc-users] MP2-R12 single point energy

[Francesco P. <chi...@ya...>]

Thank you for the message below. Yes, I blurred
things, very sorry. I reeducated myself through
Labanowski's "Simplified introduction to ab initio
basis set. Terms and notation" on internet, and other
readings.

Then, unable to implement the suggestion below for
Slater Type Orbitals, I went to use the "MP2-R12
energy" input proposed for water on "mpqc input" on
the sourceforge web site (cc-pvDZ and aug-cc-pVDZ
basis set).

Using command "mpqc -thread
> "<PthreadThreadGrp>:(num_threads = 4)"
> *.in |tee *.out"

on my machine with a couple of dual-opterons, 8GB ram,
debian-amd64, and mpc231 recompiled for libint,

for my big molecules, resetting
symmetry = 1
memory = 7GB 
at both places, the output gave erratic energy values
for the
 energy during the preliminary iter steps with cc-pVDZ
(values from -1017 to +389 during 14 iter processes).
Therefore, I killed the process at this stage.

Then I decided to check my system with the original
"MP2-R12 energy" input above (for the water
moolecule), simply resetting

memory = 7GB

at both places. Using the same -thread command above,
the computation went to end instantly. Input and
output files are attached. Here the energy decreased
steadily during the preliminary iter processes.

I would be much obliged for you having a look at the
attached output file. Is anything wrong there? I would
like to be sure that things run correctly, before
going to check the MP2-R12 energy calculation with a
couple of conformers much simpler than those I
unsuccessfully used.

Thanks a lot for your kind attention

francesco pietra



--- "Kenny, Joseph P" <jp...@sa...> wrote:

> Problem is that the basis for your guess
> wavefunction (STO-3G) has sp shells, nothing to do
> with aug-cc-pVDZ basis.  You need to use a basis set
> without these sp shells.  User-generated basis set
> splitting STO-3G sp shells would work.
> 
> Joe
> 
> -----Original Message-----
> From: mpq...@li... on
> behalf of Francesco Pietra
> Sent: Thu 12/7/2006 11:32 PM
> To: Kenny, Joseph P
> Cc: Matt Leininger; mpq...@li...
> Subject: Re: [Mpqc-users] MP2-R12 single point
>  
> Basis set name was correct in the input file, while
> it
> was wrong in the file name, now corrected (attached
> here, together with the output file; other
> recipients
> receive files from my system, so that at least at
> your
> address they should be delivered). 
> 
> Replacing aug-cc-pVDZ with cc-pVDZ gave similar
> output
> file. Therefore, I must have badly constructed the
> input.
> 
> Commad was: $ mpqc -thread
> "<PthreadThreadGrp>:(num_threads = 4)"
> coleq_singlepoint_aug-cc-pVDZ_MP2-R12.in |tee
> coleq_singlepoint_aug-cc-pVDZ_MP2-R12.out
> 
> Thanks a lot for your care.
> 
> francesco pietra
> --- Joseph Kenny <jp...@sa...> wrote:
> 
> > Cints will work with these basis sets.  Note, the
> > keyword is
> > "aug-cc-pVDZ", not "aug-ccpVDZ", was this your
> > problem?  Shouldn't get a
> > complaint about fully general contraction from
> Cints
> > with this type of
> > basis set -- need input/output for further
> analysis
> > and I didn't get any
> > attachments.
> > 
> > Does source forge strip attachments???  Didn't
> think
> > so, this message
> > isn't in archive yet, so I can't check for
> > attachment there.
> > 
> > Joe
> > 
> > On Thu, 2006-12-07 at 13:33 -0800, Francesco
> Pietra
> > wrote:
> > > --- Matt Leininger <mll...@hp...>
> > > wrote:
> > > 
> > > > On Thu, 2006-12-07 at 00:45 -0800, Francesco
> > Pietra
> > > > wrote:
> > > > > Very sorry to come again for help about
> > MP2-R12
> > > > > 
> > > > > I tried MP2-R12 single-point energy
> > computation by
> > > > > merely replacing "aug-ccpVDZ" with "cc-pVXZ"
> > in
> > > > 
> > > >    The cc-pVXZ is a series of basis sets where
> > > > X={D,T,Q,5,6}.  Replace X
> > > > with one of these options should correct the
> > error
> > > > below.
> > > > 
> > > >   Thanks,
> > > > 
> > > > 	- Matt
> > > 
> > > Hi Matt:
> > > Using cc-pVDZ, the errors below are corrected
> > indeed
> > > but the obstacle of Cints with also this basis
> set
> > is
> > > not. In fact I wrote this morning to the users:
> > > 
> > > I naively thought that cc-pVXZ is different. I
> > tried
> > > MP2-R12 with the same input file by replacing
> > > aug-ccpVDZ with cc-pVDZ, getting again that
> Cints
> > can
> > > not deal with basis set with fully general
> > > contraction. Suggested to use IntegralV3. I must
> > > conclude that I did not understand what you
> > suggested
> > > to do.
> > > 
> > > The input file is attached again deprived of
> most
> > > coordinate data to make the file smaller. This
> > file
> > > was constructed from the MP2_single-point input
> > file
> > > that you sent me (and worked), modified
> according
> > to
> > > examples on the mpqc manual. It may well be that
> I
> > > badly arranged the input file.
> > > 
> > > Cheers
> > > francesco pietra
> > > 
> > > 
> > > > 
> > > > > previous input file (which is attached here
> > > > again).
> > > > > Unfortunately, the resulting *.in file is
> far
> > from
> > > > OK.
> > > > > Trying to start the process with the -thread
> > > > command,
> > > > > like previously, no computation was started:
> > > > > 
> > > > > Machine: x86_64-pc-linux-gnu
> > > > > Using ProcMessageGrp for message passing
> > (number
> > > > of
> > > > > nodes = 1)
> > > > > Using Pthread ThreadGrp for threading
> (number
> > of
> > > > > threads = 4)
> > > > > Using ProcMemoryGrp for distributed shared
> > memory
> > > > > Total number of processors = 4
> > > > > Using IntegralV3 by default for molecular
> > > > integrals
> > > > > evaluation
> > > > > Reading file /usr/share/mpqc/atominfo.kv
> > > > > GaussianBasisSet::count_shells_ couldn't
> find
> > > > > ":basis:carbon:cc-pVXZ":
> > > > > PrefixKeyVal: error: "The keyword was not
> > found"
> > > > >    prefix:
> > > > > "basis"
> > > > > keyval:
> > > > > ParseKeyVal: error: "The keyword was not
> > found"
> > > > > 
> > > > > Thanks a lot for help
> > > > > 
> > > > > francesco pietra
> > > > > 
> > > > > 
> > > > > 
> > > > > 
> > > > > --- Joseph Kenny <jp...@sa...> wrote:
> > > > > 
> > > > > > Hi Francesco, 
> > > > > > 
> > > > > > This all looks as expected.  Cints can't
> > handle
> > > > > > general contractions
> > > > > > that look like this:
> > > > > > 
> > > > > >   (type: [am = p am = s]
> > > > > >    {exp coef:1 coef:0} = {
> > > > > >             0.636289700     -0.09996723   
>  
> > > > > > 0.15591627
> > > > > >             0.147860100      0.39951283   
>  
> > > > > > 0.60768372
> > > > > >             0.048088700      0.70011547   
>  
> > > > > > 0.39195739
> > > > > >    })
> > > > > > 
> > > > > > You can do the same calculation, just less
> > > > > > efficiently, by splitting
> > > > > > this type of shell into two shells, one s
> > shell
> > > > and
> > > > > > one p shell.  In
> > > > > > 3.2.1 you'd need to do this with a user
> > defined
> > > > > > basis set.  In the trunk
> > > > > > there is a SplitGaussianBasisSet object
> that
> 
=== message truncated ===


 
____________________________________________________________________________________
Cheap talk?
Check out Yahoo! Messenger's low PC-to-Phone call rates.
http://voice.yahoo.com

Re: [Mpqc-users] MP2-R12 single point

[Kenny, J. P <jp...@sa...>]

Problem is that the basis for your guess wavefunction (STO-3G) has sp =
shells, nothing to do with aug-cc-pVDZ basis.  You need to use a basis =
set without these sp shells.  User-generated basis set splitting STO-3G =
sp shells would work.

Joe

-----Original Message-----
From: mpq...@li... on behalf of Francesco =
Pietra
Sent: Thu 12/7/2006 11:32 PM
To: Kenny, Joseph P
Cc: Matt Leininger; mpq...@li...
Subject: Re: [Mpqc-users] MP2-R12 single point
=20
Basis set name was correct in the input file, while it
was wrong in the file name, now corrected (attached
here, together with the output file; other recipients
receive files from my system, so that at least at your
address they should be delivered).=20

Replacing aug-cc-pVDZ with cc-pVDZ gave similar output
file. Therefore, I must have badly constructed the
input.

Commad was: $ mpqc -thread
"<PthreadThreadGrp>:(num_threads =3D 4)"
coleq_singlepoint_aug-cc-pVDZ_MP2-R12.in |tee
coleq_singlepoint_aug-cc-pVDZ_MP2-R12.out

Thanks a lot for your care.

francesco pietra
--- Joseph Kenny <jp...@sa...> wrote:

> Cints will work with these basis sets.  Note, the
> keyword is
> "aug-cc-pVDZ", not "aug-ccpVDZ", was this your
> problem?  Shouldn't get a
> complaint about fully general contraction from Cints
> with this type of
> basis set -- need input/output for further analysis
> and I didn't get any
> attachments.
>=20
> Does source forge strip attachments???  Didn't think
> so, this message
> isn't in archive yet, so I can't check for
> attachment there.
>=20
> Joe
>=20
> On Thu, 2006-12-07 at 13:33 -0800, Francesco Pietra
> wrote:
> > --- Matt Leininger <mll...@hp...>
> > wrote:
> >=20
> > > On Thu, 2006-12-07 at 00:45 -0800, Francesco
> Pietra
> > > wrote:
> > > > Very sorry to come again for help about
> MP2-R12
> > > >=20
> > > > I tried MP2-R12 single-point energy
> computation by
> > > > merely replacing "aug-ccpVDZ" with "cc-pVXZ"
> in
> > >=20
> > >    The cc-pVXZ is a series of basis sets where
> > > X=3D{D,T,Q,5,6}.  Replace X
> > > with one of these options should correct the
> error
> > > below.
> > >=20
> > >   Thanks,
> > >=20
> > > 	- Matt
> >=20
> > Hi Matt:
> > Using cc-pVDZ, the errors below are corrected
> indeed
> > but the obstacle of Cints with also this basis set
> is
> > not. In fact I wrote this morning to the users:
> >=20
> > I naively thought that cc-pVXZ is different. I
> tried
> > MP2-R12 with the same input file by replacing
> > aug-ccpVDZ with cc-pVDZ, getting again that Cints
> can
> > not deal with basis set with fully general
> > contraction. Suggested to use IntegralV3. I must
> > conclude that I did not understand what you
> suggested
> > to do.
> >=20
> > The input file is attached again deprived of most
> > coordinate data to make the file smaller. This
> file
> > was constructed from the MP2_single-point input
> file
> > that you sent me (and worked), modified according
> to
> > examples on the mpqc manual. It may well be that I
> > badly arranged the input file.
> >=20
> > Cheers
> > francesco pietra
> >=20
> >=20
> > >=20
> > > > previous input file (which is attached here
> > > again).
> > > > Unfortunately, the resulting *.in file is far
> from
> > > OK.
> > > > Trying to start the process with the -thread
> > > command,
> > > > like previously, no computation was started:
> > > >=20
> > > > Machine: x86_64-pc-linux-gnu
> > > > Using ProcMessageGrp for message passing
> (number
> > > of
> > > > nodes =3D 1)
> > > > Using Pthread ThreadGrp for threading (number
> of
> > > > threads =3D 4)
> > > > Using ProcMemoryGrp for distributed shared
> memory
> > > > Total number of processors =3D 4
> > > > Using IntegralV3 by default for molecular
> > > integrals
> > > > evaluation
> > > > Reading file /usr/share/mpqc/atominfo.kv
> > > > GaussianBasisSet::count_shells_ couldn't find
> > > > ":basis:carbon:cc-pVXZ":
> > > > PrefixKeyVal: error: "The keyword was not
> found"
> > > >    prefix:
> > > > "basis"
> > > > keyval:
> > > > ParseKeyVal: error: "The keyword was not
> found"
> > > >=20
> > > > Thanks a lot for help
> > > >=20
> > > > francesco pietra
> > > >=20
> > > >=20
> > > >=20
> > > >=20
> > > > --- Joseph Kenny <jp...@sa...> wrote:
> > > >=20
> > > > > Hi Francesco,=20
> > > > >=20
> > > > > This all looks as expected.  Cints can't
> handle
> > > > > general contractions
> > > > > that look like this:
> > > > >=20
> > > > >   (type: [am =3D p am =3D s]
> > > > >    {exp coef:1 coef:0} =3D {
> > > > >             0.636289700     -0.09996723    =20
> > > > > 0.15591627
> > > > >             0.147860100      0.39951283    =20
> > > > > 0.60768372
> > > > >             0.048088700      0.70011547    =20
> > > > > 0.39195739
> > > > >    })
> > > > >=20
> > > > > You can do the same calculation, just less
> > > > > efficiently, by splitting
> > > > > this type of shell into two shells, one s
> shell
> > > and
> > > > > one p shell.  In
> > > > > 3.2.1 you'd need to do this with a user
> defined
> > > > > basis set.  In the trunk
> > > > > there is a SplitGaussianBasisSet object that
> > > > > automates this process, but
> > > > > it hasn't made it into a release yet.
> > > > >=20
> > > > > In practice, this is rarely a limitation, as
> we
> > > > > typically use cc-pVXZ
> > > > > type basis sets for correlated calculations.
> > > > >=20
> > > > > Joe=20
> > > > >=20
> > > > > On Wed, 2006-12-06 at 10:23 -0800, Francesco
> > > Pietra
> > > > > wrote:
> > > > > > Compiled for debian amd64 for libint along
> > > > > (hopefully)
> > > > > > the directions indicated below, mpqc 2.3.1
> > > with
> > > > > > command:
> > > > > >=20
> > > > > > $ mpqc -thread
> > > "<PthreadThreadGrp>:(num_threads =3D
> > > > > 4)"
> > > > > > coleq_singlepoint_aug-ccpVDZ_MP2-R12.in |
> tee
> > > > > > coleq_singlepoint_aug-ccpVDZ_MP2-R12.out
> > > > > >=20
> > > > > > started telling that it'll use 892 basis,
> > > starting
> > > > > > energy ca 2800, ther performed a STO,
> thoughat
> > > > > that
> > > > > > point:
> > > > > >=20
> > > > > > Projecting the guess density.
> > > > > > The nr of electrons in the guess density =3D
> 212
> > > > > > mpqc error, exception raised:
> > > > > > IntegralCints cannot handle basis set with
> > > fully
> > > > > > general contractions yet, try IntegralIV3
> > > instead.
> > > > > >=20
> > > > > > I must confess that I am unable to
> implement
> > > these
> > > > > > suggestions and hope on help.
> > > > > >=20
> > > > > > Attached are the in and out files
> concerned.
> > > For
> > > > > the
> > > > > > "in" I made an arrangement from files that
> I
> > > used
> > > > > > successfully for single-point MP2
> calculation,
>=20
=3D=3D=3D message truncated =3D=3D=3D


=20
_________________________________________________________________________=
___________
Do you Yahoo!?
Everyone is raving about the all-new Yahoo! Mail beta.
http://new.mail.yahoo.com

Re: [Mpqc-users] MP2-R12 single point

[Francesco P. <chi...@ya...>]

Basis set name was correct in the input file, while it
was wrong in the file name, now corrected (attached
here, together with the output file; other recipients
receive files from my system, so that at least at your
address they should be delivered). 

Replacing aug-cc-pVDZ with cc-pVDZ gave similar output
file. Therefore, I must have badly constructed the
input.

Commad was: $ mpqc -thread
"<PthreadThreadGrp>:(num_threads = 4)"
coleq_singlepoint_aug-cc-pVDZ_MP2-R12.in |tee
coleq_singlepoint_aug-cc-pVDZ_MP2-R12.out

Thanks a lot for your care.

francesco pietra
--- Joseph Kenny <jp...@sa...> wrote:

> Cints will work with these basis sets.  Note, the
> keyword is
> "aug-cc-pVDZ", not "aug-ccpVDZ", was this your
> problem?  Shouldn't get a
> complaint about fully general contraction from Cints
> with this type of
> basis set -- need input/output for further analysis
> and I didn't get any
> attachments.
> 
> Does source forge strip attachments???  Didn't think
> so, this message
> isn't in archive yet, so I can't check for
> attachment there.
> 
> Joe
> 
> On Thu, 2006-12-07 at 13:33 -0800, Francesco Pietra
> wrote:
> > --- Matt Leininger <mll...@hp...>
> > wrote:
> > 
> > > On Thu, 2006-12-07 at 00:45 -0800, Francesco
> Pietra
> > > wrote:
> > > > Very sorry to come again for help about
> MP2-R12
> > > > 
> > > > I tried MP2-R12 single-point energy
> computation by
> > > > merely replacing "aug-ccpVDZ" with "cc-pVXZ"
> in
> > > 
> > >    The cc-pVXZ is a series of basis sets where
> > > X={D,T,Q,5,6}.  Replace X
> > > with one of these options should correct the
> error
> > > below.
> > > 
> > >   Thanks,
> > > 
> > > 	- Matt
> > 
> > Hi Matt:
> > Using cc-pVDZ, the errors below are corrected
> indeed
> > but the obstacle of Cints with also this basis set
> is
> > not. In fact I wrote this morning to the users:
> > 
> > I naively thought that cc-pVXZ is different. I
> tried
> > MP2-R12 with the same input file by replacing
> > aug-ccpVDZ with cc-pVDZ, getting again that Cints
> can
> > not deal with basis set with fully general
> > contraction. Suggested to use IntegralV3. I must
> > conclude that I did not understand what you
> suggested
> > to do.
> > 
> > The input file is attached again deprived of most
> > coordinate data to make the file smaller. This
> file
> > was constructed from the MP2_single-point input
> file
> > that you sent me (and worked), modified according
> to
> > examples on the mpqc manual. It may well be that I
> > badly arranged the input file.
> > 
> > Cheers
> > francesco pietra
> > 
> > 
> > > 
> > > > previous input file (which is attached here
> > > again).
> > > > Unfortunately, the resulting *.in file is far
> from
> > > OK.
> > > > Trying to start the process with the -thread
> > > command,
> > > > like previously, no computation was started:
> > > > 
> > > > Machine: x86_64-pc-linux-gnu
> > > > Using ProcMessageGrp for message passing
> (number
> > > of
> > > > nodes = 1)
> > > > Using Pthread ThreadGrp for threading (number
> of
> > > > threads = 4)
> > > > Using ProcMemoryGrp for distributed shared
> memory
> > > > Total number of processors = 4
> > > > Using IntegralV3 by default for molecular
> > > integrals
> > > > evaluation
> > > > Reading file /usr/share/mpqc/atominfo.kv
> > > > GaussianBasisSet::count_shells_ couldn't find
> > > > ":basis:carbon:cc-pVXZ":
> > > > PrefixKeyVal: error: "The keyword was not
> found"
> > > >    prefix:
> > > > "basis"
> > > > keyval:
> > > > ParseKeyVal: error: "The keyword was not
> found"
> > > > 
> > > > Thanks a lot for help
> > > > 
> > > > francesco pietra
> > > > 
> > > > 
> > > > 
> > > > 
> > > > --- Joseph Kenny <jp...@sa...> wrote:
> > > > 
> > > > > Hi Francesco, 
> > > > > 
> > > > > This all looks as expected.  Cints can't
> handle
> > > > > general contractions
> > > > > that look like this:
> > > > > 
> > > > >   (type: [am = p am = s]
> > > > >    {exp coef:1 coef:0} = {
> > > > >             0.636289700     -0.09996723     
> > > > > 0.15591627
> > > > >             0.147860100      0.39951283     
> > > > > 0.60768372
> > > > >             0.048088700      0.70011547     
> > > > > 0.39195739
> > > > >    })
> > > > > 
> > > > > You can do the same calculation, just less
> > > > > efficiently, by splitting
> > > > > this type of shell into two shells, one s
> shell
> > > and
> > > > > one p shell.  In
> > > > > 3.2.1 you'd need to do this with a user
> defined
> > > > > basis set.  In the trunk
> > > > > there is a SplitGaussianBasisSet object that
> > > > > automates this process, but
> > > > > it hasn't made it into a release yet.
> > > > > 
> > > > > In practice, this is rarely a limitation, as
> we
> > > > > typically use cc-pVXZ
> > > > > type basis sets for correlated calculations.
> > > > > 
> > > > > Joe 
> > > > > 
> > > > > On Wed, 2006-12-06 at 10:23 -0800, Francesco
> > > Pietra
> > > > > wrote:
> > > > > > Compiled for debian amd64 for libint along
> > > > > (hopefully)
> > > > > > the directions indicated below, mpqc 2.3.1
> > > with
> > > > > > command:
> > > > > > 
> > > > > > $ mpqc -thread
> > > "<PthreadThreadGrp>:(num_threads =
> > > > > 4)"
> > > > > > coleq_singlepoint_aug-ccpVDZ_MP2-R12.in |
> tee
> > > > > > coleq_singlepoint_aug-ccpVDZ_MP2-R12.out
> > > > > > 
> > > > > > started telling that it'll use 892 basis,
> > > starting
> > > > > > energy ca 2800, ther performed a STO,
> thoughat
> > > > > that
> > > > > > point:
> > > > > > 
> > > > > > Projecting the guess density.
> > > > > > The nr of electrons in the guess density =
> 212
> > > > > > mpqc error, exception raised:
> > > > > > IntegralCints cannot handle basis set with
> > > fully
> > > > > > general contractions yet, try IntegralIV3
> > > instead.
> > > > > > 
> > > > > > I must confess that I am unable to
> implement
> > > these
> > > > > > suggestions and hope on help.
> > > > > > 
> > > > > > Attached are the in and out files
> concerned.
> > > For
> > > > > the
> > > > > > "in" I made an arrangement from files that
> I
> > > used
> > > > > > successfully for single-point MP2
> calculation,
> 
=== message truncated ===


 
____________________________________________________________________________________
Do you Yahoo!?
Everyone is raving about the all-new Yahoo! Mail beta.
http://new.mail.yahoo.com

Re: [Mpqc-users] MP2-R12 single point

[Joseph K. <jp...@sa...>]

Cints will work with these basis sets.  Note, the keyword is
"aug-cc-pVDZ", not "aug-ccpVDZ", was this your problem?  Shouldn't get a
complaint about fully general contraction from Cints with this type of
basis set -- need input/output for further analysis and I didn't get any
attachments.

Does source forge strip attachments???  Didn't think so, this message
isn't in archive yet, so I can't check for attachment there.

Joe

On Thu, 2006-12-07 at 13:33 -0800, Francesco Pietra wrote:
> --- Matt Leininger <mll...@hp...>
> wrote:
> 
> > On Thu, 2006-12-07 at 00:45 -0800, Francesco Pietra
> > wrote:
> > > Very sorry to come again for help about MP2-R12
> > > 
> > > I tried MP2-R12 single-point energy computation by
> > > merely replacing "aug-ccpVDZ" with "cc-pVXZ" in
> > 
> >    The cc-pVXZ is a series of basis sets where
> > X={D,T,Q,5,6}.  Replace X
> > with one of these options should correct the error
> > below.
> > 
> >   Thanks,
> > 
> > 	- Matt
> 
> Hi Matt:
> Using cc-pVDZ, the errors below are corrected indeed
> but the obstacle of Cints with also this basis set is
> not. In fact I wrote this morning to the users:
> 
> I naively thought that cc-pVXZ is different. I tried
> MP2-R12 with the same input file by replacing
> aug-ccpVDZ with cc-pVDZ, getting again that Cints can
> not deal with basis set with fully general
> contraction. Suggested to use IntegralV3. I must
> conclude that I did not understand what you suggested
> to do.
> 
> The input file is attached again deprived of most
> coordinate data to make the file smaller. This file
> was constructed from the MP2_single-point input file
> that you sent me (and worked), modified according to
> examples on the mpqc manual. It may well be that I
> badly arranged the input file.
> 
> Cheers
> francesco pietra
> 
> 
> > 
> > > previous input file (which is attached here
> > again).
> > > Unfortunately, the resulting *.in file is far from
> > OK.
> > > Trying to start the process with the -thread
> > command,
> > > like previously, no computation was started:
> > > 
> > > Machine: x86_64-pc-linux-gnu
> > > Using ProcMessageGrp for message passing (number
> > of
> > > nodes = 1)
> > > Using Pthread ThreadGrp for threading (number of
> > > threads = 4)
> > > Using ProcMemoryGrp for distributed shared memory
> > > Total number of processors = 4
> > > Using IntegralV3 by default for molecular
> > integrals
> > > evaluation
> > > Reading file /usr/share/mpqc/atominfo.kv
> > > GaussianBasisSet::count_shells_ couldn't find
> > > ":basis:carbon:cc-pVXZ":
> > > PrefixKeyVal: error: "The keyword was not found"
> > >    prefix:
> > > "basis"
> > > keyval:
> > > ParseKeyVal: error: "The keyword was not found"
> > > 
> > > Thanks a lot for help
> > > 
> > > francesco pietra
> > > 
> > > 
> > > 
> > > 
> > > --- Joseph Kenny <jp...@sa...> wrote:
> > > 
> > > > Hi Francesco, 
> > > > 
> > > > This all looks as expected.  Cints can't handle
> > > > general contractions
> > > > that look like this:
> > > > 
> > > >   (type: [am = p am = s]
> > > >    {exp coef:1 coef:0} = {
> > > >             0.636289700     -0.09996723     
> > > > 0.15591627
> > > >             0.147860100      0.39951283     
> > > > 0.60768372
> > > >             0.048088700      0.70011547     
> > > > 0.39195739
> > > >    })
> > > > 
> > > > You can do the same calculation, just less
> > > > efficiently, by splitting
> > > > this type of shell into two shells, one s shell
> > and
> > > > one p shell.  In
> > > > 3.2.1 you'd need to do this with a user defined
> > > > basis set.  In the trunk
> > > > there is a SplitGaussianBasisSet object that
> > > > automates this process, but
> > > > it hasn't made it into a release yet.
> > > > 
> > > > In practice, this is rarely a limitation, as we
> > > > typically use cc-pVXZ
> > > > type basis sets for correlated calculations.
> > > > 
> > > > Joe 
> > > > 
> > > > On Wed, 2006-12-06 at 10:23 -0800, Francesco
> > Pietra
> > > > wrote:
> > > > > Compiled for debian amd64 for libint along
> > > > (hopefully)
> > > > > the directions indicated below, mpqc 2.3.1
> > with
> > > > > command:
> > > > > 
> > > > > $ mpqc -thread
> > "<PthreadThreadGrp>:(num_threads =
> > > > 4)"
> > > > > coleq_singlepoint_aug-ccpVDZ_MP2-R12.in | tee
> > > > > coleq_singlepoint_aug-ccpVDZ_MP2-R12.out
> > > > > 
> > > > > started telling that it'll use 892 basis,
> > starting
> > > > > energy ca 2800, ther performed a STO, thoughat
> > > > that
> > > > > point:
> > > > > 
> > > > > Projecting the guess density.
> > > > > The nr of electrons in the guess density = 212
> > > > > mpqc error, exception raised:
> > > > > IntegralCints cannot handle basis set with
> > fully
> > > > > general contractions yet, try IntegralIV3
> > instead.
> > > > > 
> > > > > I must confess that I am unable to implement
> > these
> > > > > suggestions and hope on help.
> > > > > 
> > > > > Attached are the in and out files concerned.
> > For
> > > > the
> > > > > "in" I made an arrangement from files that I
> > used
> > > > > successfully for single-point MP2 calculation,
> > > > taking
> > > > > suggestions as to R12 from the manual.
> > > > > 
> > > > > Thanks a lot for answering
> > > > > francesco pietra
> > > > > 
> > > > > (I changed email address and I have not yet
> > found
> > > > the
> > > > > way to change my coordinates on GNU web sites)
> > > > > 
> > > > > --- Joseph Kenny <jp...@sa...> wrote:
> > > > > 
> > > > > > Hi,
> > > > > > 
> > > > > > MP2-R12 requires from an external package --
> > > > libint.
> > > > > >  See here for
> > > > > > details:
> > > > > > 
> > > > > > http://www.mpqc.org/mpqc-html/compile.html
> > > > > > 
> > > > > > Once libint is compiled properly (make sure
> > r12
> > > > > > integrals are turned on
> > > > > > in configure), you need to recompile mpqc
> > with
> > > > the
> > > > > > libint include/lib
> > > > > > paths given in --with-include and
> > > > --with-libdirs.
> > > > > > 
> > > > > > Joe
> > > > > > 
> > > > > > On Sun, 2006-12-03 at 01:54 -0800, Francesco
> > > > Pietra
> > > > > > wrote:
> > > > > > > I tried MP2-R12 single point on the basis
> > of
> > > > the
> > > > > > MP2
> > > > > > > layout, mofified as for MP2-R12 Energy
> > from
> > > > the
> > > > > > mpqc
> > > > > > > manual.
> > > > > > > 
> > > > > > > Either I did mistakes or the layout has to
> > be
> > > > > > change.
> > > > > > > I got:
> > > > > > > 
> > > > > > > ClassDesc::load_class looking for
> > "./classes"
> > > > > > > ClassDesc::load_class("MBPT_R12"): load
> > failed
> > > > > > > Either "MBPT2_R12" is an invalid class
> > name or
> > > > the
> > > > > > > code for "MBPT2_R12" was not linked into
> > the
> > > > > > > executable.
> > > > > > > 
> > > > > > > I understand from the latter WW that the
> > > > standard
> > > > > > > mpqc2.3.1 (from Debian) is not ready for
> > such
> > > > a
> > > > > > > calculation and has to be recompiled.
> > > > > > > 
> > > > > > > Tanks for answering
> > > > > > > 
> > > > > > > francesco pietra
> > > > > > > 
> > > > > > > 
> > > > > > >  
> > 
> === message truncated ===
> 
> 
> 
>  
> ____________________________________________________________________________________
> Want to start your own business?
> Learn how on Yahoo! Small Business.
> http://smallbusiness.yahoo.com/r-index
> 
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share your
> opinions on IT & business topics through brief surveys - and earn cash
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> Mpqc-users mailing list
> Mpq...@li...
> https://lists.sourceforge.net/lists/listinfo/mpqc-users
-- 
Joseph P. Kenny
Sandia National Laboratories, Livermore CA
Scalable Computing Research and Development