[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [ESPResSo] analyze pressure and energy problems
From: |
Axel Arnold |
Subject: |
Re: [ESPResSo] analyze pressure and energy problems |
Date: |
Tue, 16 Jun 2009 11:19:13 +0200 |
User-agent: |
KMail/1.9.4 |
On Monday 15 June 2009 19:45, Markus Deserno wrote:
> Dear İbrahim,
>
> without having looked too much at your specific problem...
> ...but: In "natural" simulation units, assuming all masses are
> essentially the same, the integration time step should not
> be 1, rather maybe 0.01. How many seconds that represents
> is a whole different question. This requires mapping and
> depends not only on your masses but also the hard-to-guess
> speedup factor which you get from coarse-graining. It seems
> to me that your system might just blow up.
Hi,
just to add to that: if you look carefully at the error message, you see that
Tcl complains that the result of "analyze energy kinetic" is not a valid
floating point number. And indeed, if you would print the value directly,
without enclosing in with "expr", you would see that in your case the result
is "nan", i.e. "not a number", after the first step, i.e. that some particle
velocities are completely out of range. This actually applies to most
energies in your system, btw... So, Markus is definitely right that your
timestep is about 2 orders of magnitude too large.
This also explains the signal 7 you received: if particle velocities are NaN,
the positions also become NaN, and the domain decomposition runs into
trouble.
Finally, regarding the warning from P3M: it is a bit misleading: P3M cannot
calculate the pressure tensor, however, it can calculate the scalar pressure.
The warning is only regarding the tensorial pressure, which for
electrostatics is a bit more cumbersome to calculate.
Axel
--
Dr. Axel Arnold
Fraunhofer SCAI
Schloss Birlinghoven, 53754 Sankt Augustin, Germany
Tel: +49 2241 14 2575