Re: [AMBER] Langevin thermostat different gamma_ln giving systematically different energies?

From: Thomas Cheatham <>
Date: Sun, 17 Jan 2021 02:23:48 +0000

>From your e-mail it is difficult to interpret - your average energies seem to be the same (-75,000) but the standard deviations are different (which is not unexpected - if you add randomized changes more frequently, more likely more deviation) but perhaps I am misunderstanding.


From: Liao <>
Sent: Saturday, January 16, 2021 6:44:16 PM
To: AMBER Mailing List
Subject: [AMBER] Langevin thermostat different gamma_ln giving systematically different energies?

Dear Amber community,

Iíve been testing out the effect with different gamma_ln values (0.1, 2, 5, 10) using a Langevin thermostat, and what I found with my system (NPT, using ntf=7 only evaluating non-bonded forces, giving my protein a weight restraint of 100 so that itís as fixed as possible) was that different random seeds with the same gamma_ln give very close average energies from the MD output (-75000+-5), while different gamma_ln values give different average energies(-75000+-500). The runs are over 100 ns and the averages seemed to be converged pretty well within each gamma_ln value, while they are systematically different between different gamma_ln.

Is this difference in energy expected? Because itís undesired for my tests here. Could it be something specific with my system thatís causing this?

AMBER mailing list

AMBER mailing list
Received on Sat Jan 16 2021 - 18:30:02 PST
Custom Search