Re: [ESPResSo-users] ICCP3M WITH MPI

[marco macis]

Hello,

your suggestions were very good Marcello, I think my worst error was on the initialization of the induced charges.....

Unfortunately, now I've a problem of segmentation when I use more than one cpu (with only one cpu the model runs ok).

This problem appears when the warm-up step is finished, but I don't understand why, I'm an espresso user only by a month and I don't know if it is an mpi error or a mistake that I made in the input file.

This is the error message

[s953-labzer01:01749] *** Process received signal ***

[s953-labzer01:01749] Signal: Segmentation fault (11)

[s953-labzer01:01749] Signal code: Address not mapped (1)

[s953-labzer01:01749] Failing at address: 0x6a1a9c0

[s953-labzer01:01749] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0x10340) [0x7f3086c12340]

[s953-labzer01:01749] [ 1] ../../../espresso-3.3.0/Espresso() [0x4bb3fb]

[s953-labzer01:01749] [ 2] ../../../espresso-3.3.0/Espresso() [0x4bbabd]

[s953-labzer01:01749] [ 3] ../../../espresso-3.3.0/Espresso() [0x465835]

[s953-labzer01:01749] [ 4] ../../../espresso-3.3.0/Espresso() [0x46652d]

[s953-labzer01:01749] [ 5] ../../../espresso-3.3.0/Espresso() [0x468735]

[s953-labzer01:01749] [ 6] ../../../espresso-3.3.0/Espresso() [0x442559]

[s953-labzer01:01749] [ 7] ../../../espresso-3.3.0/Espresso() [0x448940]

[s953-labzer01:01749] [ 8] ../../../espresso-3.3.0/Espresso() [0x40d4ed]

[s953-labzer01:01749] [ 9] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf5) [0x7f308685eec5]

[s953-labzer01:01749] [10] ../../../espresso-3.3.0/Espresso() [0x40db1a]

[s953-labzer01:01749] *** End of error message ***

I attached my input file,

Thanks all of you for your suggestons

2015-04-01 15:53 GMT+02:00 Marcello Sega <address@hidden>:

Hi,

it doesn't seem to be an mpi-related problem, I get it also with single cpu.

Try to use

lappend normals 0
lappend normals 0
lappend normals 1

instead of

lappend normals {0 0 1}


Also be sure to:

(a) add real (i.e. non-iccp3m) particles only *after* the iccp3m
charges , otherwise the algorithm won't work (this is also stated in
the manual)

(b) give an initial (small) random value to the induced charges,
otherwise you might have convergence problems ( e.g. [expr 0.01 * (
[t_random] - 0.5) ]  )

Cheers,

M.

On Wed, Apr 1, 2015 at 3:06 PM, Florian Weik <address@hidden> wrote:
> Hello,
> could you please post the complete script that produces the error?
>
> Cheers,
> Florian
>
> On Wed, Apr 1, 2015 at 1:51 PM, marco macis <address@hidden> wrote:
>>
>> Hello,
>>
>> I am trying to set up a model to simulate the behaviour of ions in
>> solution in presence of a permittivity discontinuity with iccp3m using mpi
>> but it doesn't work.
>>
>> My model consist in two walls in the center of the box and 602 ions in
>> solution.
>>
>> I made three different tests:
>>
>>
>> - Only Lennard-Jones parameters activated for ions and walls particles: it
>> runs ok with mpi.
>>
>>
>> - Lennard-Jones parameters plus inter coulomb p3m for the ions in
>> solutions: it runs ok with mpi
>>
>>
>> - Lennard-Jones parameters plus inter coulomb p3m for ions in solutions
>> plus iccp3m for highlight the discontinuity of permittivity between walls
>> and bulk: it runs ok without mpi but not when I activate it.
>>
>>
>> The error that appears it seems relative to the level of tcl list normals,
>> the programs say “couldn't understand 0 0 ..” when it tries to implement
>> iccp3m algorithm.
>>
>>
>> In attachment the portion of my input where I build the walls
>>
>> #walls
>>
>> set normals ""
>>
>> set areas ""
>>
>> set sigmas ""
>>
>> set epsilons ""
>>
>> set l 100
>>
>> set lx 100
>>
>> set ly 100
>>
>> set lz 100
>>
>>
>> setmd box $lx $ly $lz
>>
>>
>> set numero $l
>>
>> set contatore 0
>>
>> for {set i 1} {$i < $numero} {incr i} {
>>
>> for {set j 1} {$j < $numero} {incr j} {
>>
>> set x $j
>>
>> set y $i
>>
>> set z 52
>>
>> lappend normals {0 0 1}
>>
>> lappend areas 1
>>
>> lappend sigmas 0
>>
>> lappend epsilons 2000
>>
>> set type 2
>>
>> part $contatore pos $x $y $z type $type q 0 v 0.0 0.0 0.0 fix
>>
>> incr contatore
>>
>> }
>>
>> }
>>
>> for {set i 1} {$i < $numero} {incr i} {
>>
>> for {set j 1} {$j < $numero} {incr j} {
>>
>> set x $j
>>
>> set y $i
>>
>> set z 48
>>
>> lappend normals {0 0 -1}
>>
>> lappend areas 1
>>
>> lappend sigmas 0
>>
>> lappend epsilons 2000
>>
>> set type 2
>>
>> part $contatore pos $x $y $z type $type q 0 v 0.0 0.0 0.0 fix
>>
>> incr contatore
>>
>> }
>>
>> }
>>
>>
>> …
>>
>> …
>>
>>
>> and i attached the portion where I call the electrostatic
>>
>>
>> inter coulomb 7.1432034 p3m tunev2 accuracy 1e-3
>>
>> iccp3m 19602 convergence 1e-3 areas $areas normals $normals sigmas $sigmas
>> epsilons $epsilons
>>
>>
>> Do you have any suggestion for my problem?
>>
>>
>> Thank you very much
>>
>>
>> Marco
>
>
>
>
> --
> Florian Weik, Dipl.-Phys.,
> Institut für Computerphysik, Allmandring 3, D-70569 Stuttgart
> Phone: +49-711-685-67703
> Public Key 0x0562F11D Fingerprint 3294 6360 EC93 37A3 BD40  F597 0BAD 3AF8
> 0562 F11D
>

--
Institut für Computergestützte Biologische Chemie
University of Vienna
PGP public key on MIT server http://goo.gl/zlIZix

run_iccp3m.tcl
Description: Tcl script