While using "Fix Langevin" system to keep the source and sink temperature constant during MD simulation of Si and/or Ge to calculate thermal conductivity, the temperature profile is showing a non-linearity inside my source and sink region. I guess it has something to do with my system size since it is in angstrom level i.e. (source and sink both taken as 20-30 angstrom each). However, when using a larger system, i.e. (source and sink both taken as 20-30 nanometer each), the non-linearity almost disappears except at the very end of source region and the very beginning of the sink region. I took quantum size effect into consideration. In nanoscale, the scattering might me ballistic but still that does not answer the question that I have - what might be the source of this non-linearity? I can go on with the larger system which does not pose any non-linearity issue, but it requires substantial amount of simulation time (i.e. 2/3 days most cases). That is the first reason, I started with a smaller system and got that non-linear trend inside those regions while using "Fix Langevin".