Thermostat choice and time constant effects for small Al clusters in MD
hamid zabihi
I’m running molecular dynamics simulations for a small aluminum cluster (Al₆) and tested two thermostats — CSVR and Nosé–Hoover — with different time constants: very small (5, 15 fs), moderate (200 fs), and large (400 fs).
A few questions came up during these tests:
-
Is there any preferred choice between CSVR and Nosé–Hoover when simulating small metallic clusters (e.g., 4–6 atoms)?
-
With CSVR, the system temperature shows large fluctuations (often rising up to ~300 K) instead of stabilizing near the target 100 K, even with small time constants.
-
With Nosé–Hoover, the temperature profile is smoother and closer to equilibrium, but using very small time constants seems to make the thermostat too restrictive, potentially altering the physical nature of the MD trajectory.
-
On the other hand, using larger time constants (200 or 400 fs) makes it difficult for the system to reach thermal equilibrium within a reasonable simulation time.
I have attached the resulting temperature evolution figures and the corresponding md.inp files used for each simulation for reference.
Do you have any recommendations for:
-
Choosing between CSVR and Nosé–Hoover for small cluster dynamics
-
Selecting an appropriate time constant range to balance thermal equilibration and realistic trajectory sampling
Any advice or best practices for such small systems would be greatly appreciated.
Thank you!
