dpd as a thermostat

Hi, all Lammps users.
I come across several questions about using dpd to simulate a poiseuille flow.Just a few days ago, I've asked the same question and got answer from Axel, Samuel and Steve, I appreciate their kindly help.However, I still have some questions, as follows:

1. dpd contains a thermostat, so I can't use pair_style dpd and pair_style dpd/tstat at the same time,right? { When I use dpd + nve, the flow temperature in y direction is not 1.0 as I expected. However, when I use dpd+nve +dpd/tstat (I guess I use twice dpd thermostat), the temperature in y direction is exactly 1.0, however,the flow velocity is beyond the analytical velocity. }
2. I am confused by compute temp/partial and compute temp/profile. Both computes remove velcity bias in some dimension before compute temperature, what's the difference? Accroding to answer from Axel, Samuel and Steve last time, I know that I needn't to exclude a velocity bias from a dpd flow, since dpd as a thermostat works on relative velocity of 2 atoms in a pair. Is this means that if I use a compute/partial to calculate temperature of the system, I will get the wrong result?
3. I want to set the flow temperature to 1.0, is this means that temperature in y direction is exactly 1.0 or some temperature fluctuation is allowed near the wall?
4. I feel dpd as a thermostat can't fulfill my target-- right flow velocity and right temperature. So I have to use another thermostat or I have to change the weight function w(r) in the dpd function?

These question seems too naive, but they have confused me for weeks. I hope if any one can give me some suggestions. Thank you very much!


Xiaojun Li
Biomechanics and Biomaterials Laboratory
Department of Applied Mechanics
School of Aerospace Engineering
Beijing Institute of Technology
Beijing 100081, P. R. China
Email: [email protected]...

in.smallmodel-nve8 (3.25 KB)

Comments below.


If you read the doc page, dpd/tstat is a subset of what dpd already does. I.e. dpd
already thermostats. So you don’t want to thermostat again.

Don’t thermostat twice.

Thermostatting and monitoring the temperature are 2 different things.
Dpd should thermostat your flowing system fine. To monitor the thermal
temp of your system you need to remove the flow velocity. There
are various ways to do that. Compute temp/partial and compute temp/profile
do it in different ways. The doc pages for the 2 commands explain
what they do.

What is flow temperature? I think of flow velocity and thermal temperature.
The latter two are independent.

Thank you, Steve.
“Flow temperature”,I mean, when I create velocity, I grouped “fluid particles” as flow,i.e. ,“flow” is the groupID.
velocity flow create 1.0 4568465 rot yes dist gaussian
There are upper and lower wall outside the flowdomain. When I create the temperature of the system, should I use “velocity all” instead of “velocity flow”?
I also think "rot yes " is not necessary, because dpd is linear and angular momentum conservation.


When I create the temperature of the system, should I use “velocity all” instead of “velocity flow”?

try it and see - I think you’re now asking Qs that can easily
be answered by reading the doc pages. The mail list
cannot formulate your model for you - that’s your job.