Hello all,

In the manual section for temp/profile compute claims:

"After the spatially-averaged velocity field has been subtracted from

each atom, the temperature is calculated by the formula KE = (dim/2 N -

dim*Nx*Ny*Nz) k T, where KE = total kinetic energy of the group of atoms

(sum of 1/2 m v^2), dim = 2 or 3 = dimensionality of the simulation, N =

number of atoms in the group, k = Boltzmann constant, and T =

temperature. The dim*Nx*Ny*Nz term are degrees of freedom subtracted to

adjust for the removal of the center-of-mass velocity in each of

Nx*Ny*Nz bins, as discussed in the (Evans) paper."

But I think with N you mean number of atoms in a certain bin (bin style)

that belongs also to the group specified as group-ID. Is that correct?

If it's so, I think I should use the 'compute_modify comp_id dynamic

yes' if within my bins the number of atoms inside is fluctuating

although the number of atoms within the group is preserved. Is that

right?

P.S: Why the binning for temp/profile is limited in the sense of not

having the possibility of choosing the lower and upper bounds for a

certain coordinate ? I would expect the same capability as 'compute

bin/3d' for example. Would it be easy to implement modifying the

existing source code?

Many thanks,

Eduardo - (MSc student at Imperial College London)

yes, N = # of atoms in bin

you donâ€™t need compute_modify dynanmic, b/c

the compute re-bins atoms every time it

calcs a temperature

Re: binning options. Compute temp/profile is

designed more for calculating a temperature

after subtracting out a local flow velocity. So

it can be used for a thermostat. You can limit

what atoms are involved in the calculation by

the group you define, which can also be a dynamic

group with a region limiting what atoms are in the

group.

If you want to tabulate the temperature in a bunch

of bins, see the compute chunk/atom and compute temp/chunk

and fix ave/chunk commands. That would be the

place to add an option limiting the spatial extent

covered by bins.

Steve

Dear Steve,

Thanks for your quick reply. The fact is that I want to calculate the

temperature profile of a sheared liquid in confinement that change in

width during equilibration (two mobile walls due to the barostating

process). Then when the fluctuations are minimal I shear the liquid

inside till steady state is achieved. At that point is when I want to

compute the temperature profile and it is observed that the liquid is

disposed in layers in which local equilibrium is attained. Because of

that, within each layer, it is possible to calculate the temperature

separating the peculiar and "non-peculiar" velocities in the flow

direction just how 'fix temp/profile' does. This command, as I commented

before, does not give me the flexibility I want on how to bin the fluid.

So two questions:

1) Would it be possible to implement the behabiour of this command using

'fix ave/chunk' for velocities and then manually calculate the

temperatures inside each bin from it?

2) Is there currently any way of doing not uniform binning in one

direction?. In example, binning from x_lo to x_hi with bins of size x1,

x2, x3, .. where x_lo - x_hi = x1 + x2 + x3 ... . That is because the

layers in my simulation might be of different thicknesses.

Regards,

Eduardo - (MSc student at Imperial College London)