Dear Rahul,

your plots show that at 500K, energy terms (etot, econs, econt) are relatively stable, but from what you're describing, higher temperatures introduce drift. This is common issue in CPMD with Nose-Hoover thermostat; it requires careful parameter tuning.

> Key parameters:

emass=200 (electron mass for CPMD)

dt=1 (in a.u)

fnosep=135 (for termostat control)

>> At higher temperatures atoms move faster so you need smaller time steps to properly capture dynamics.

Try dt=0.5 or 0.25, especially if you go above 800-900 K.

>>> At higher temperatures, electrons might not stay adiabatically separated from ions. Lowering emass helps maintain adiabaticity.

Try emass=150 or 100, so that monitoring kinetic should be in meV or less!

>>>> For Nose-Hoover frequency, fnosep=135 should be not so aggressive, but for high temperatures you should!!! reduce it to 100 or 110.

>>>>> If your cp.x supports it, increasing number of chains for the Nose-Hoover thermostat can help at high T, so that nchain=4 or even 5

>>>>>> For very high temperatures, pre-equilibrate at lower T /e.g heat from 500 to 900 K gradually over 1-2 ps/ then switch to constant 900 K.

So try this:

emass=150

dt=0.5

fnosep=110

Tweak further e.g., reduce emass again if needed.

Hope, it helps :]

Kostadin Georgiev Thank you for the valuable advice. I’ll check this and get back once I observe a proper trend.