Toon, Tell us a little bit more about the method for the FORCE_EVAL. I work mostly with QS (DFT) and one has to work hard to get accurate pressures in general. An accuracy below 0.1 GPa, 1000 bar is non-trivial (probably not even realistic given other sources of systematic error). 1 bar is a tiny pressure. I would say that with the thermostats the most important thing to check are the averages of course. 5000 bar is 0.5 GPa. I think that's reasonable. If you have plenty of computer resources one thing you could check is how the fluctuations change with system size. In NVT for example, I believe it can be shown that the fluctuations in T should scale as 1/sqrt(N). Don't know if this is true for other ensembles. <div>On 5/10/07, Toon <<a href="mailto:Toon.Ver...@gmail.com">Toon.Ver...@gmail.com</a>> wrote:<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> Hi, When performing an MD simulation, using the NPT_I ensemble with the default settings and a target pressure of 1 bar, what is the order of magnitude of the oscillations in the instantaneous pressure that one should expect? (I have set PV_AVA to true in the input file to get the instantaneous pressure in the output file.) Since I'm new on this subject, I was performing a simple test run on a box of solid methane with 64 molecules. (low temperature, below 90K, zero diffusion) I see oscillations in the instantaneous pressure with an amplitude of 5000 bar. Is this normal? Maybe the answer is very system dependent? best regards, Toon 443-567-8328 (C) 410-306-0709 (O)