Hello again,
I have tested the script for the poiseuille flow with the parameters
proposed by Georg Rempfer. The simulation was speeded up considerably by
that. Anyway, still at least one question arises.
Attached you will find the flow profile after it has saturated. As you
may notice, the maximum velocity is smaller than the expected one. As it
was written in the bachelor thesis of Georg, this is due to the grid
(“the walls can only be determined as precise as the resolution of the
LB lattice”, p. 17). I have also tested it with a grid spacing of 0.5
which resulted in a maximum velocity closer to the expected value but
still too low.
Is there any way to avoid that problem in espresso or is it due to the
nature of the lattice Boltzmann implementation and do I have to accept
that. I am asking, since I have seen some other MD-simulations, which
also used the LB method but did not show that problem.
Greets and thanks a lot again!
Markus Wink
*Von:address@hidden
[mailto:address@hidden
*Im Auftrag von *Wink, Markus
*Gesendet:* Montag, 27. Januar 2014 18:06
*An:* 'Georg Rempfer'; address@hidden
*Betreff:* Re: [ESPResSo-users] Simulation of fluid using LB-method
Hello,
thanks a lot for the fast reply and your help. I will test the script
with the new parameters and will give a feedback of the performance.
Anyway, I still have two questions left:
1) You wrote, that the dynamic viscosity nu should be one. In the user’s
guide it is written (page 158), that one inserts the kinematic viscosity
nu/rho for the lbfluid command. Sure, since rho equals unity in MD
units, that makes no difference, I am just asking to make that point clear.
2) How do you get such an “odd” number for the force density? In SI
units, it is 64.000kg/m^2/s^2. If, as you proposed, [m] = 10^-15kg,
[x]=10^-6m and [t]=10^-6s, shouldn’t the force density be 6.4*10^-5 in
MD units? Sorry, for being so confused.
3) When talking about force in the lbfluid command, one is really
talking about the force density and not about the force on every node of
the LB-grid. That is at least, what I understood while reading the user
guide. Am I right with that point?
Thanks again!
Greets,
Markus
*Von:address@hidden
<mailto:address@hidden>
[mailto:address@hidden
*Im Auftrag von *Georg Rempfer
*Gesendet:* Montag, 27. Januar 2014 17:07
*An:* address@hidden <mailto:address@hidden>
*Betreff:* Re: [ESPResSo-users] Simulation of fluid using LB-method
Forgot to attach the file. Here you go.
2014-01-27 Georg Rempfer <address@hidden
<mailto:address@hidden>>
Sorry, I had missed the attachment in the first email. A remark
unrelated to the performance issue:
I would not feel comfortable using such extreme parameters
(density~10^10 and viscosity~10^-3) with lattice-Boltzmann. LB only
produced proper hydrodynamic behavior in a limited parameter range. By
choosing values around 1, you should be on the safe side. Also I got a
force density of 52896 kg/m^2/s^2 using your units. Attached you find a
derivation of a more suitable unit system. In that system, your force
density would be 5.29e-5. Independently of those performance issues, I
think that should speed up your calculations considerably (looking at
the flow in calculation time makes it seem much lighter and more viscous
now, so it will reach its stationary state much faster).
2014-01-27 Wink, Markus <address@hidden
<mailto:address@hidden>>
>
> Hello everyone,
>
>
>
> First of all: thanks for your support.
>
>
>
> I thought that I sent the *.tcl script with my first post. Anyway.
Attached you will find the scripts and the config.log.
>
>
>
> Greetings,
>
>
>
> Markus
>
>
>
>
>
> Von:
address@hidden
<mailto:address@hidden>
[mailto:espressomd-users-bounces+markus.wink
<mailto:espressomd-users-bounces%2Bmarkus.wink>address@hidden
<mailto:address@hidden>] Im Auftrag von Olaf Lenz
> Gesendet: Montag, 27. Januar 2014 15:05
> An: Georg Rempfer
> Cc: address@hidden <mailto:address@hidden>
>
>
> Betreff: Re: [ESPResSo-users] Simulation of fluid using LB-method
>
>
>
> Maybe you can also send your config.log? Then we can see how the
program is compiled.
>
>
>
> Olaf
>
>
>
> 2014-01-27 Georg Rempfer <address@hidden
<mailto:address@hidden>>
>
> Hello Markus,
>
>
>
> the friction parameter won't have an influence on your simulation as
long as you don't use any explicit particles. Can you provide your
simulation script? If you give it to me, I will run and time it here.
>
>
>
> Greetings,
>
> Georg
>
>
>
> 2014-01-27 Ulf Schiller <address@hidden
<mailto:address@hidden>>
>
>
>
> On 01/27/2014 02:07 PM, Wink, Markus wrote:
>
> Hello,
>
> I hope I understood your point correctly. I have run the script with
two calls of "integrate 1000", having a box of 50^3 and an agrid of 1,
resulting in 125*10^6 site updates.
> It took 22 min 2s for the simulation resulting in roughly 19k SUPS.
This is far below your expected 10M. I have no clue what the reason for
that is.
>
> The binary of Espresso I have created only had "LB", "LB_BOUNDARIES"
and "EXTERNAL_FORCES" activated. Running the testscript "lb_planar.tcl"
(box of length 12, total integrate of 5000) takes about 14 sec,
resulting in ~0.6M SUPS.
>
>
>
> Hi,
>
> well, Espresso has of course some overhead since it's not a pure LB
code, but ~200k (that's what you get I think) resp. 600k seems pretty
slow. What CPU are you running on?
>
> Does anyone else have an estimate of how many site updates per second
they get?
>
> Cheers,
> Ulf
>
>
>
> -----Ursprüngliche Nachricht-----
> Von:
address@hidden
<mailto:address@hidden>
[mailto:espressomd-users-bounces+markus.wink
<mailto:espressomd-users-bounces%2Bmarkus.wink>address@hidden
<mailto:address@hidden>] Im Auftrag von Ulf Schiller
> Gesendet: Montag, 27. Januar 2014 12:16
> An: address@hidden <mailto:address@hidden>
> Betreff: Re: [ESPResSo-users] Simulation of fluid using LB-method
>
> On 01/27/2014 11:54 AM, Wink, Markus wrote:
>
> 3) Even though, the velocity profile saturates, it takes pretty long
> (time scale of days). The saturation sets in at about 350 times the
> "integration 1000" command. This corresponds to a time scale of ms in Si
> units. Besides using OpenMpi or the GPU impementation, is there anyway
> to speeden up the simulation? For example by adjusting the parameter tau
> in the lbfluid command?
>
>
> Hi,
>
> that sounds a bit strange to me. If your system is 50^3 and you
> integrate 1000 steps, that gives 125*10^6 site updates. On a reasonably
> modern CPU, a reasonable LB code should get to around 10M SUPS (million
> site updates per second), that is, one call of "integrate 1000" should
> not take much more than 10 seconds, so 350 calls roughly one hour not
> days. Can you check how many site updates per second you get on your
system?
>
> Cheers,
> Ulf
smime.p7s