I compute every particle’s total force using group function fcm for every timestep, and the total force should include drag force, random force and the repulsive force between the particles and polymer network’s atoms if particles diffusion in polymer networks.

I compute every particle's total force using group function fcm for every

timestep, and the total force should include drag force, random force and

the repulsive force between the particles and polymer network's atoms if

particles diffusion in polymer networks.

unless you have a force term adding a drift to your system, the sum of

all forces (aka the force on the center of mass) *should* be (on average)

zero and the langevin thermostat (or any other regular thermostat) should

not change that.

you are comparing apples and oranges when you compare this term to the MSD.

axel.

thank you for your reply. I understand you are talking about the total force of the whole system. But I focus on the single particle which can diffuse freely or diffuse in a polymer network. The polymer network will hinder the particle diffuse, so diffusive particles’ MSD (statistic from many particles) will different between with and without polymer network. Force is reason for changing the state of motion of the diffusive particle. So I compute the single diffusive particle’s total force (for statistic), and try to find their difference in the case of with and without polymer network. But the results obtained from compute diffusive particles are no difference, which confused me.

thank you for your reply. I understand you are talking about the total

force of the whole system. But I focus on the single particle which can

diffuse freely or diffuse in a polymer network. The polymer network will

hinder the particle diffuse, so diffusive particles' MSD (statistic from

many particles) will different between with and without polymer network.

Force is reason for changing the state of motion of the diffusive

particle. So I compute the single diffusive particle's total force (for

statistic), and try to find their difference in the case of with and

without polymer network. But the results obtained from compute diffusive

particles are no difference, which confused me.

again, if there is no drift imposed, e.g. through a fix or an

inhomogeneity of the system, the average of the force should be 0.

regardless of how many particles are in the group. the impact of fix

langevin should not change that. the einstein relation after which the

(self-)diffusion constant is computed from the MSD is based on such a

random walk.

let's consider the most extreme case of hard spheres (force is either 0 or

infinity): here the diffusivity is obviously not determined by the force,

but rather by the average time or distance between two collisions. so i

contend, that your interpretation of what determines the magnitude of

diffusion is not correct.

axel.

I’m really appreciate for your reply, which unraveled my confusion. I will look for more information and reference about it and understand it more comprehensively. Thanks again!