Thank you Andrian,
I will look into that paper and your suggestion is very helpful to me.
________________________________
From: Adrian Roitberg <roitberg.ufl.edu>
Sent: Wednesday, November 16, 2016 7:38:58 AM
To: AMBER Mailing List
Subject: Re: [AMBER] HMassRepartition
Hi
But your previous mdin had ntt=1, which means it would never use the
gamma_ln parameters.
You need to read a bit on what that parameter means.
There is a recent paper by Alexey Onufriev on how conformation the speed
of sampling depends on the value you choose.
Biophysical Journal 108, 1153-64 (2015)
Bottomline, if you are simulating equilibria (as opposed to caring about
the actual time scale of things), then it does not matter which value
you use, you will get the same answer. However, a smaller value will
sample the same space 'faster'. For a very small system, if you use a
small value of gamma_ln, you run the risk of having large temperature
fluctuations, which can mess things up.
If you care about the actual time scale of processes, many of us have
found that a value of ~50 seems to match the time scales as if you were
running inexplicit solvent.
Adrian
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
https://urldefense.proofpoint.com/v2/url?u=http-3A__lists.ambermd.org_mailman_listinfo_amber&d=DgICAg&c=y2w-uYmhgFWijp_IQN0DhA&r=0Hah93XWYYBgmROOVcaccg&m=L4f57wnYv8ikL-1Qs-51BCFOl6eLy2kzyH8QcK4VUo8&s=QwvgfJZQd0yZdoFvd_2vM9WItwh8_j6qKSU4Tk5Tx2Q&e=
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Nov 16 2016 - 11:30:03 PST
