error to using PLUMED with lammps

[Charlie]

I tried to run the example with PLUMED package. But an internal ABORT
happened as follow:
.................................
Finding SHAKE clusters ...
0 = # of size 2 clusters
0 = # of size 3 clusters
0 = # of size 4 clusters
4000 = # of frozen angles
PLUMED: FOUND THE INPUT FILE plumed.dat
PLUMED: FOUND THE OUTPUT FILE metaout.dat
PROCESSOR 0 NATOMS 12000
Plumed Object being created...
Done!
PPPM initialization ...
G vector = 0.285155
grid = 24 24 24
stencil order = 6
RMS precision = 0.000104668
brick FFT buffer size/proc = 29791 13824 11532
Setting up run ...
Assertion failed in file helper_fns.c at line 337: 0
memcpy argument memory ranges overlap, dst_=0x6904cc80 src_=0x6904cc80
len_=4

internal ABORT - process 0
................................................................

I used the last version of lammps. the compile process is copied from
the plumed manual. Please help on this. Thanks.

[Davide Branduardi]

Hi Charlie,
did you checked if the version of lammps is supported by us?
Note that lammps changes quite often and it is safer to patch a version supported by plumed.
Good luck
D

> --
> You received this message because you are subscribed to the Google Groups "PLUMED users" group.
> To post to this group, send email to plumed...@googlegroups.com.
> To unsubscribe from this group, send email to plumed-users...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/plumed-users?hl=en.
>

[Charlie]

I used the 24Sep2010 version of LAMMPS. The error is still here.
Anyone has idea about this:

Assertion failed in file helper_fns.c at line 337: 0

memcpy argument memory ranges overlap, dst_=0x6ed7ed00 src_=0x6ed7ed00
len_=4

I guess something wrong happen to the memory management. But I can't
identify it. Here is followed by my screen information:

[charlie@hl373 torsion]$ lmplumed < input
LAMMPS (24 Sep 2010)
Scanning data file ...
3 = max bonds/atom
6 = max angles/atom
20 = max dihedrals/atom
Reading data file ...
orthogonal box = (-100 -100 -100) to (100 100 100)
1 by 1 by 1 processor grid
22 atoms
21 bonds
36 angles
66 dihedrals
Finding 1-2 1-3 1-4 neighbors ...
4 = max # of 1-2 neighbors
7 = max # of 1-3 neighbors
13 = max # of 1-4 neighbors
16 = max # of special neighbors
Finding SHAKE clusters ...
3 = # of size 2 clusters

0 = # of size 3 clusters

3 = # of size 4 clusters

0 = # of frozen angles

PLUMED: FOUND THE INPUT FILE myfile.cfg

PLUMED: FOUND THE OUTPUT FILE metaout.dat

PROCESSOR 0 NATOMS 22

Plumed Object being created...
Done!

Finding 1-2 1-3 1-4 neighbors ...
4 = max # of 1-2 neighbors
7 = max # of 1-3 neighbors
13 = max # of 1-4 neighbors
16 = max # of special neighbors
Setting up run ...
SHAKE stats (type/ave/delta) on step 0
3 1.0899 0.000401433
5 1.09024 0.000316568
7 1.00981 0.000141576

Assertion failed in file helper_fns.c at line 337: 0

memcpy argument memory ranges overlap, dst_=0x6ed7ed00 src_=0x6ed7ed00

len_=4

internal ABORT - process 0

On Feb 16, 6:08 pm, Davide Branduardi <davide.brandua...@gmail.com>
wrote:

[Fabio Pietrucci]

Hi Charlie,

a trivial question, if you run the same lammps input without
plumed, does it work?

Fabio

--
____________________
 CECAM Lausanne
 homepage: http://sites.google.com/site/fabiopietrucci

[Paolo Raiteri]

Hi Charlie,

I'm also using the latest version of LAMMPS with no problems.
I don't have the file helper_fns.c among the source files. A quick search on the web shows that it was part of the POEMS package in 2008. POEMS has then been unpdated and that file removed in the later releases. Can you double check which version of LAMMPS you have downloaded and patched? It's likely that PLUMED will not work with old versions of LAMMPS.

Cheers
Paolo

[Charlie]

I recently reinstalled our workstation and minimized the packages of
lammps. But plumed doesn't work here. Could you let me know the
details of your system and the vesions of compilers. I appreciate it~

here is the error information I got this time:

[charlie@localhost torsion]$ lmp < input
LAMMPS (9 Feb 2011)

Fatal error in PMPI_Allgather: Invalid buffer pointer, error stack:
PMPI_Allgather(896): MPI_Allgather(sbuf=0x8aee3f0, scount=1, MPI_INT,
rbuf=0x8aee3f0, rcount=1, MPI_INT, MPI_COMM_WORLD) failed
PMPI_Allgather(869): Buffers must not be aliased

On Feb 18, 8:14 am, Paolo Raiteri <paolo.rait...@gmail.com> wrote:
> Hi Charlie,
>
> I'm also using the latest version of LAMMPS with no problems.
> I don't have the file helper_fns.c among the source files. A quick search on the web shows that it was part of the POEMS package in 2008. POEMS has then been unpdated and that file removed in the later releases. Can you double check which version of LAMMPS you have downloaded and patched? It's likely that PLUMED will not work with old versions of LAMMPS.
>
> Cheers
> Paolo
>
> On 18/02/2011, at 9:43 AM, Fabio Pietrucci wrote:
>
> > Hi Charlie,
>
> >   a trivial question, if you run the same lammps input without
> > plumed, does it work?
>
> > Fabio
>

> >> For more options, visit this group athttp://groups.google.com/group/plumed-users?hl=en.

[Paolo Raiteri]

Hi Charlie,

It works on my iMac, MacBook and on a few thousands cores cluster with the gcc and intel compilers (versions 4.6.0 and 10.1 respectively).

Did you try to run LAMMPS with the same configuration before patching it with PLUMED, i.e. as downloaded from the Sandia website? Does it work?

Cheers
Paolo

[Hongyi Liu]

Hey Paolo,

 

Yes, I did with the same Makefile before pactching and it worked very well without the plumed patch. After pacthing, the compiling is smooth too. But when running the example from the plumed package, the error came out.

 

I will try the intel compiler. Thanks. will let you know what happen.

 

Charlie

--

HONGYI LIU, Ph.D.

Postdoctoral Associate, Earth & Atmospheric Science

KAUST-Cornell Center for Energy and Sustainability
Cornell University

[Giovanni Bussi]

Hi all,

I have no experience with LAMMPS but it seems a problem with the
compiler optimization.

In call "MPI_Allgather(&meatoms,1,MPI_INT,my_atoms,1,MPI_INT,world);",
file fix_plumed.cpp,
meatoms and myatoms should have different addresses in memory, but
maybe the compiler is removing the temporary "meatoms=my_atoms[me]"
introduced two lines up.

You can try the following replacement:
MPI_Allgather(&meatoms,1,MPI_INT,my_atoms,1,MPI_INT,world);
into:
int *tmp=new int;
*tmp=meatoms;
MPI_Allgather(tmp,1,MPI_INT,my_atoms,1,MPI_INT,world);
delete tmp;

Let me know if it works,

Giovanni

[Hongyi Liu]

Dear Paolo,

Sorry for this lagging reply. I just tried the intel compiler for Lammps, intel compxe fortran and c++, both of them are the last version. The compiling was going very well. But when I run the test with do_regtest_lammps.sh, I still got the error as below. I think Giovanni is right, and I will try that soon.

[charlie@hl373 lammps]$ ./do_regtest_lammps.sh
*************************** testing started ******************************
 started on  Tue Mar 1 10:11:41 EST 2011
-------------------------regtesting lammps----------------------------------
Starting tests in /home/charlie/software/plumed/tests/lammps/torsion

Fatal error in PMPI_Allgather: Invalid buffer pointer, error stack:

PMPI_Allgather(896): MPI_Allgather(sbuf=0x7f239710, scount=1, MPI_INT, rbuf=0x7f239710, rcount=1, MPI_INT, MPI_COMM_WORLD) failed

PMPI_Allgather(869): Buffers must not be aliased

tail: cannot open `/home/charlie/software/plumed/tests/lammps/TEST-lammps/torsion/input.out' for reading: No such file or directory

Fatal error in PMPI_Allgather: Invalid buffer pointer, error stack:

PMPI_Allgather(896): MPI_Allgather(sbuf=0x7812a710, scount=1, MPI_INT, rbuf=0x7812a710, rcount=1, MPI_INT, MPI_COMM_WORLD) failed

PMPI_Allgather(869): Buffers must not be aliased

tail: cannot open `/home/charlie/software/plumed/tests/lammps/TEST-lammps/torsion/input.out' for reading: No such file or directory
>>>>>>>>>>>>>>>>> /home/charlie/software/plumed/tests/lammps/torsion
                                     torsion/input          RUNTIME FAIL               TYPE
                                     torsion/input          RUNTIME FAIL               TYPE
<<<<<<<<<<<<<<<<< /home/charlie/software/plumed/tests/lammps/torsion
--------------------------------------------------------------------------
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
/home/charlie/software/plumed/tests/lammps/TEST-lammps/torsion/input.out
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
/home/charlie/software/plumed/tests/lammps/TEST-lammps/torsion/input.out
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
--------------------------------- summary --------------------------------
number of FAILED  tests 2
number of WRONG   tests 0
number of CORRECT tests 0
number of NEW     tests 0
number of         tests 2
--------------------------------------------------------------------------
Tue Mar  1 10:11:42 EST 2011
*************************** testing ended ********************************

Thanks a lot.

Best regards,

Charlie

On Tue, Mar 1, 2011 at 12:57 AM, Paolo Raiteri <paolo....@gmail.com> wrote:

Hey,

Have you tried the example in .../plumed/tests/lammps/torsion ??

Does it work?

If it does work can you send us all the input files we need to reproduce the error?

Cheers,

Paolo

[Hongyi Liu]

Dear Giovanni,

The compilation was aborted after modifing the fix_plumed.cpp file:

icc -DLAMMPS_PLUMED     -O -DLAMMPS_GZIP -I../../lib/atc -I../../lib/reax -I../../lib/poems -I../../lib/meam  -DMPICH_SKIP_MPICXX -I/usr/local/mpich2/include  -DFFT_FFTW -I/usr/local/fftw/2.1.5/include -c fix_plumed.cpp
fix_plumed.cpp(297): error: identifier "meatoms" is undefined
   *tmp=meatoms;
        ^
compilation aborted for fix_plumed.cpp (code 2)
make[1]: *** [fix_plumed.o] Error 2

Thanks a lot and hope to get your advice further.

Best regards,

Charlie

[Hongyi Liu]

Dear Giovanni,

"meatoms" was replaced by "my_atoms[me]" in your code and now we can sucessfully run the test of plumed. Thank you so much~

[charlie@hl373 lammps]$ ./do_regtest_lammps.sh
*************************** testing started ******************************

 started on  Tue Mar 1 11:35:05 EST 2011

-------------------------regtesting lammps----------------------------------
Starting tests in /home/charlie/software/plumed/tests/lammps/torsion

NOW COPY /home/charlie/software/plumed/tests/lammps/TEST-lammps/torsion/std_out INTO  /home/charlie/software/plumed/tests/lammps/LAST-lammps/torsion/std_out
NOW COPY /home/charlie/software/plumed/tests/lammps/TEST-lammps/torsion/HILLS.out INTO  /home/charlie/software/plumed/tests/lammps/LAST-lammps/torsion/HILLS.out
>>>>>>>>>>>>>>>>> /home/charlie/software/plumed/tests/lammps/torsion
                                     torsion/input     NEW  (  0.02 sec)               TYPE
                                     torsion/input     NEW  (  0.02 sec)               TYPE
<<<<<<<<<<<<<<<<< /home/charlie/software/plumed/tests/lammps/torsion
--------------------------------------------------------------------------
--------------------------------- summary --------------------------------
number of FAILED  tests 0

number of WRONG   tests 0
number of CORRECT tests 0

number of NEW     tests 2

number of         tests 2
--------------------------------------------------------------------------

Tue Mar  1 11:35:06 EST 2011
*************************** testing ended ********************************


Best regards,

Charlie