# G-K viscosity method,

Yes, you need to increase both the correlation length and the amount of averaging done to get ACFs which decay to zero. If the three components are still different when this happens, you need to simulate for even longer. If your system is isotropic they should be the same.

Dear Jackson,

Thank you very much for your reply.

I’ve just studied some articles including rigid water models. They pointed out because of isotropicity we make sure we need to calculate components which we consider them in

green-kubo method and this makes viscosity calculation easier. I use flexible model for viscosity and I do not know it is possible to get justifiable results or not. If it is possible,

let me know how I can make sure increasing in correlation length and sampling can be good. I mean how often I should do sampling for better results.

Dear Jackson,

Thank you very much for your reply.

I've just studied some articles including rigid water models. They pointed
out because of isotropicity we make sure we need to calculate components
which we consider them in

green-kubo method and this makes viscosity calculation easier. I use
flexible model for viscosity and I do not know it is possible to get
justifiable results or not. If it is possible,

let me know how I can make sure increasing in correlation length and
sampling can be good. I mean how often I should do sampling for better
results.

​ali,

there is no simple "do this and it will be correct" answer. you will have
to make 'experiments' and see for yourself. starting this kind of
calculation with a complex system (and molecules are more complex than
atoms) without having any experience is a massive waste of your time.

have a look at the existing examples in the LAMMPS distribution and try out
different settings (for simulation length, sampling frequency and
correlation length) and explore which help to improve the result and which
make it worse and by how much. those are examples for which the results are
known, so you can see, if you can find the optimal set of settings by
yourself. once you have mastered this challenge, you have a better sense
for how to tell, if something is converged or not and if a result is
trustworthy or not and then (and *only* then) you should continue with your
water simulations.

as niall was telling you earlier, there have been extensive discussions on
the subject and you should carefully(!) review them and extract as much
helpful information from them as possible and try not to ask anything that
has not been asked before.

axel.

Axel,

Yes, you are right.

I knew and no wonder you reply to me. I asked because I have recently no access to a supercomputer and I should perform my runs on my PC. It walks like a turtle, calmly.

Okay, that’s my job.

Axel,

Yes, you are right.

I knew and no wonder you reply to me. I asked because I have recently no
access to a supercomputer and I should perform my runs on my PC. It walks
like a turtle, calmly.

​if you know, why do you still do it the wrong way? haven't you learned by
now, that there are no shortcuts? that you always have to pay your dues?
be grateful, i am not your adviser, since i would have fried your backside
for wasting valuable CPU time.

axel.​

Oh my God!

Thank God!!