Re: [ESPResSo] problem with n_nodes

[Axel Arnold]

Hi!

With increasing number of nodes, you will always get this message, if you have 
longer bonds. The reason is that bonds can only exist between particles that 
are a) on the same processor or b) at most one cell away. The cell size is 
determined by the maximal interaction distance. Now, if you use e.g.harmonic 
bonds, the bond can stretch longer than the cell size, and if particles happen 
to be on different processors, you get the error message. Naturally, the 
message appears more often the more processors you have.

To get rid of this problem, you need to specify a larger r_cut for the harmonic 
potential (or whatever else you are using). However, you should first figure 
out how much the two bonded particles were actually apart. That you can do by 
using the Tcl "catch" command to catch the error, read the identities two 
offending particles, and print their positions.

Axel 

----- Original Message -----
From: "Jacob Kirkensgaard" <address@hidden>
To: "ESPResSo User mailing list" <address@hidden>
Sent: Monday, November 16, 2009 1:23:31 PM
Subject: [ESPResSo] problem with n_nodes

Hi all

I am runnning Espresso via mbtools on 1 node with 8 cpus with no problems, but 
going for example to 16 cpus I run into trouble like:

bond broken etc (particles not stored on the same node)

Anyone know how to fix this... Do you need more info?

Thanks,
Jacob
_______________________________________________
ESPResSo mailing list
address@hidden
https://fias.uni-frankfurt.de/mailman/listinfo/espresso