question and advice about" fix ave/spatial command"

Dear Steve:
I am a post graduate student from china,and I am studying the thermal conductivity of CNT using lammps through the command fix thermal conductivity and fix ave/spatial. Recently I found that the temperature keeps going on as the simulation progressing.Below is what i think about this problem.

In this command(fix ave/spatial command), there’s a “delta” to define the thickness of spatial bins in dim, which is defined all by myself when I write the input script.But what bothers me is that the box size will change in the simulation,and after the change maybe the old delta is not suitable anymore. However,as we all know that we can’t change the delta when a simulation is going on. I wander if this is the reason why the teperature couldn’t keep stable. So,what can I do in this condition?
How about we change the delta into another N, just like the N in the fix thermal conductivity command?

Any help will be appreciated,wish to hear from you soon!

My guess to your question is no. That is probably not the reason your temperature is not stable

My question to you is why is your box size changing when you are running your thermal conductivity simulation? Which ensemble are you using?

The fix ave/spatial command has a “units reduced” option

to allow for changing box size with a static delta.

But Vikas’s question is good - generally people compute
kappa for an NVE ensemble using Muller-Plathe,

so the box size shouldn’t be changing.


Thanks for your opinion!
Now the temperature is not keep going on any more, it was some careless mistake, but the temperature gradient is not good especially at the ends.
Of course I am using an NVE ensemble in the simulation. It seems that you think that it is wrong when the box size change in asimulation like this.But I don’t think so.
First of all,in my simulation, I use read data to “creat a box”,and the dim of z’s size is from 1.054755 to 98.20548, and after the simulation it may change to from 1.0411
to 97.9084.Actually it is not a big change, but I think it maybe the reason why my temperature gradient is not good, especially at the ends. Because I am using a periodical boundary condition in the MP method, and a little change in z direction may results in untrustable temperature in the ends of my CNT, when the delta is not changed. I’ve tried the “units reduced” ,but it does’t work. And in my simulation now, delta is 4.858. Which is calculated from (98.20548-1.054755)/20.
I really don’t know what to do now.

Could you care to explain why your box size is changing in NVE ensemble?

Second, what you mean by “temperature gradient is not good”? This description does not help anyone.