espressomd-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [ESPResSo-devel] One question about MPI


From: Ivan Cimrak
Subject: Re: [ESPResSo-devel] One question about MPI
Date: Fri, 15 Nov 2013 08:25:05 +0100
User-agent: Mozilla/5.0 (Windows NT 6.2; WOW64; rv:24.0) Gecko/20100101 Thunderbird/24.1.0

Thanks Peter,

You are right.

In fact, I have some MPI related problems with the OIF implementation and the previous example was my attempt to simplify the problem to a very simple script. Apparently, I simplified it too much:-)

I will try to define the problem once more and I will let you know.

Thanks,
Ivan

Dňa 14.11.2013 16:29 Peter Košovan wrote / napísal(a):
Hi Ivan,

you have chosen

inter 0 fene 10.0 1.0

i.e. your FENE potential diverges at r=1.0 and is udnefined for r>1.0. Then you put your particles to separation r=10, which results in the "bond broken" message.

Iguess what you actually wanted was a fene potential with the minimum around r=10 with a maximul amplitude of +/-1, i.e.

inter 0 fene 10.0 1.0 10.0;

first parameter is the stiffness constant, second is the maximum amplitude and third is the location of the minimum (optional, 0.0 by default)

With regards,

peter

P.S.: pozdravujem do Žiliny :-)


2013/11/14 Ivan Cimrak <address@hidden>
Hi,

I tested some MPI code (see the script below) and found out the following:

Define simulation box with dimensions 20x20x20. Define 8 particles located in the vertices of a smaller cube with an edge of length 2. The smaller cube is located such that its center has coordinates 10 10 10 and the smaller cube's faces are parallel to the faces of the simulation box.

Define 12 bonds between the particles along the edges of the cube, e.g. with fene interactions.
Then run the parallel code on 8 precessors. The simulation box will be divided into 8 subcubes with dimension 10x10x10. Unfortunately, in each of these subcubes you have just one particle and you get the error

background_errors 0 {083 bond broken between particles 0 and 1 (particles not stored on the same node)} etc....

How can I avoid these errors?  Can anybody give some explanation?

One can play with variables A and B by decreasing the size of smaller cube and once the smaller cube has the edge smaller than 1, everything is fine.

Thanks,

Ivan

Script:
setmd time_step 0.001
setmd skin 0.2

thermostat off

setmd box_l 20 20 20

set A 9
set B 11
part 0 pos $A $A $A type 0
part 1 pos $B $A $A type 0
part 2 pos $B $B $A type 0
part 3 pos $A $B $A type 0
part 4 pos $A $A $B type 0
part 5 pos $B $A $B type 0
part 6 pos $B $B $B type 0
part 7 pos $A $B $B type 0

inter 0 fene 10.0 1.0

part 0 bond 0 1
part 1 bond 0 2
part 2 bond 0 3
part 3 bond 0 0
part 4 bond 0 5
part 5 bond 0 6
part 6 bond 0 7
part 7 bond 0 4
part 0 bond 0 4
part 1 bond 0 5
part 2 bond 0 6
part 3 bond 0 7

integrate 1





--
Dr. Peter Košovan

Departmtent of Physical and Macromolecular Chemistry
Faculty of Science, Charles University in Prague, Czech Republic

Katedra fyzikální a makromolekulární chemie
Přírodovědecká fakulta Univerzity Karlovy v Praze

www.natur.cuni.cz/chemistry/fyzchem/
Tel. +420221951290
Fax +420224919752


reply via email to

[Prev in Thread] Current Thread [Next in Thread]