[gpumd] Re: Different results of thermal conductivity with orthogonal system and triclinic system of graphene

  • From: Bruce Fan <brucenju@xxxxxxxxx>
  • To: gpumd@xxxxxxxxxxxxx
  • Date: Wed, 24 Feb 2021 10:48:44 +0200

Yes, smaller Fe gives larger fluctuations.

BTW, I suggest you use larger systems. Your systems look quite small, which
cannot fully use the power of  a GPU.

"소순성" <soonsung2001@xxxxxxxxxx> 于 2021年2月24日周三 04:58写道:

Hi, Fan.
I think I find the answer for previous question.
I set the F_e parameter 1e-6 but When I change it to 1e-5, it works!

It seemed that too low F_e increased the error for heat currents.
Am I right?

--------- Original Mail ---------

*Sender* : "소순성" <soonsung2001@xxxxxxxxxx>
*Recipient* : <gpumd@xxxxxxxxxxxxx>
*Received Date* : 2021/Feb/24(Wed) 10:15:09
*Subject* : [gpumd] Different results of thermal conductivity with
orthogonal system and triclinic system of graphene
Hi, Fan.
Thank you for GPUMD and today, I have questions.

I tested HNEMD for graphene monolayers of which one is in orthogonal
system(made with conventional unit cell, 6.4x6.4nm2)
and another is in triclinic system (made with primitive unitcell
(12.3x12.3nm2, 60 degree).

With orthogonal cells, results converged well to about 3000W/mK within
small error range,
however which primitive cell, I performed 5 simulations and get the value
of 3100W/mK(+-500W/mK).

How can I get the same results with triclinic system?

P.S I attached my input files for triclinic system.













Other related posts: