Re: Error of Particle node not found

[Jean-Noël Grad]

Hi,

Offline discussion with Jonas: deleted particles are indeed kept in the 
particle list for future recycling; it helps reducing the complexity of 
the C++ code. You could use a try...except RuntimeError to skip 
non-existent particles, unfortunately this is far from ideal, as the 
number of particles is an implementation detail.

In other words, `len(system.part)` and `len(system.part[:].id)` both 
return N_existing, but to iterate with a try...except you would need the 
quantity N_existing + N_deleted. Jonas suggested using the h5md output 
format directly, which should allow you to run queries on the particle 
population with h5md-related tools.

The documentation on Reaction Ensemble methods is currently undergoing 
changes (https://github.com/espressomd/espresso/pull/3254). As soon as 
this work is done, we'll update it to reflect the fact that particles 
are not iterable when they undergo a chemical reaction.

Thank you for reporting this issue.

Best,
JN

On 11/14/19 1:39 PM, Jean-Noël Grad wrote:
> Hi,
>
> I'm not too familiar with the constant pH method, maybe Jonas can help 
> here. From what I understand, the constant pH method will create and 
> delete particles, yet the list of particle ids doesn't shrink upon 
> particle deletion (maybe due to particle recycling?).
>
> The line `system.part[k].pos` will fail because the particle was 
> deleted. However `system.part[:].pos[k]` will work just fine, because 
> the ParticleSlice has a bitmask to skip deleted particles; when printing 
> debug messages in the core, I get a jump around values of k that refer 
> to deleted particles:
> ...
> get_particle_node(237)
> get_particle_node(238)
> get_particle_node(240)
> get_particle_node(241)
> ...
>
> The issue is reproducible in 4.1. @Jonas is this behavior documented? 
> The user guide section on constant pH doesn't mention it:
> http://espressomd.org/html/doc/advanced_methods.html#constant-ph-simulation-using-the-reaction-ensemble 
>
>
> Best,
> JN
>
> On 11/7/19 8:49 AM, Jiaxing Yuan wrote:
> > Dear all,
> >
> > I am simulating a system of electrolytes where charge regulation 
> > occurs so the total particle number is changing during the simulation. 
> > However, in the production run process of my script, an error of 
> > "Particle node not found" is observed.  I confirm if I remove the 
> > lines below, the script works well:
> >
> > for k in range(number_of_particles):
> >          print(system.part[k].type, file=f_config)
> >      print("position", file=f_config)
> >      for k in range(number_of_particles):
> >          print(system.part[k].pos, file=f_config)
> >
> > So I am confused about why these lines lead to an error. What is the 
> > correct way to output the configuration so that one can analyze the 
> > configuration? My script is attached below, thank you!
> >
> > Best,
> > Jiaxing