I have found that if your initial configuration is distributed well enough, then a classical MD thermalization isn't necessary....but you'll need a good program that does that for you. If you already have a classical MD code you're familiar with, then use that. Otherwise, the skew start method of MOLDY (http://www.ccp5.ac.uk/librar.shtml#MOLDY) works really well. MOLDY also has MD capability, but like I said the skew start method for generating an initial configuration has been sufficient for me in SIESTA (although I haven't used it for water).
On Mon, Dec 7, 2009 at 8:27 AM, N H <[email protected]> wrote: > Another thing you cann try is equilibrating the system with a classical MD > run at first ... It should work. > > > On Mon, Dec 7, 2009 at 5:10 PM, Szyja, B.M. <[email protected]> wrote: > >> Hi Sufian, >> >> In general, if the random distribution of molecules gives us the system >> close enough to it's optimal geometry (not necessarily in its minimum), you >> could start MD without prior optimization, and you only need the >> equilibration run to "take care" of the proper geometry. But if you observe >> such jumps in temperature, it usually means too much or too close contacts. >> I would suggest you to do at least a couple of optimization steps and then >> try again. From my experience I can say that water is quite tricky solvent, >> it is not only due to the density of the simulated cell, but also the >> relative orientation of molecules (presence of hydrogen bonds network) can >> play an important role. >> >> Success! >> Bartek >> >> ________________________________________ >> From: [email protected] [[email protected]] >> Sent: Monday, December 07, 2009 11:32 AM >> To: [email protected] >> Cc: [email protected] >> Subject: RE: [SIESTA-L] Energy flauctuation in MD >> >> Hi Bartek >> i am planning to optimize the net structure (16H2O + Formaamide) after i >> "shake" the unit cell with those water molecules only >> random distribution of water molecules were chosen alone firstly, >> I really do not know if i need to optimize H2O molecules before MD run? >> >> yours, >> sufian >> > >> > Hi Sufian, >> > >> > My guess is the geometry being too far from optimal. Have you optimized >> > the system (atomic coordinates and/or cell size) before the MD run? >> > >> > Cheers, >> > Bartek >> > >> > ________________________________________ >> > From: [email protected] [[email protected]] >> > Sent: Monday, December 07, 2009 10:58 AM >> > To: [email protected] >> > Cc: [email protected] >> > Subject: Re: [SIESTA-L] Energy flauctuation in MD >> > >> > Hi Andrey >> > >> > I have 16-H2O molecule in my unit cell, i am trying to cool the system >> > down from 1000K to 300K to be more realistic. The problem still there, >> > the >> > temperature goes very high and won't stop at the target temperature, i >> > tried to use annealing, Nose,... all those options give the same >> problem, >> > i am wondering if there is any unseen option that can be used to fix >> this >> > problem? >> > >> > yours, >> > sufain >> >> >> >> >> >>> Hi Sufian, >> >>> >> >>> How many molecules does your "bunch" contain? It's normal to have >> >>> temperature fluctuations of ~T/sqrt(N), where N is the number of >> >>> particles. Also I encountered non-physical temperature fluctuations >> >>> when initial temperature of run was not equal to final temperature. >> >>> Seems that Nose thermostat in Siesta is not capable to implement >> >>> temperature gradients. And the final remark - if you want to model >> >>> gas-liquid phase transition, Siesta might not be the right tool to do >> >>> it, since it needs a big supercell and large simulation time. >> >>> >> >>> Cheers, >> >>> Andrey. >> >>> >> >>>> Dear Siesta Users >> >>>> >> >>>> I am trying to run a molecular dynamics simulation for a bunch of H2O >> >>>> molecules,i chose the Tini =800 K and Tfin= 300 K my run starts >> >>>> normally for the first one or two steps. but after that a huge >> >>>> fluctuations in temperature appear, it reached a couple of thousands >> >>>> of kelvins i,tried many options but i still has the same problem, >> >>>> can any body assist me in this problem? >> >>>> Thank you all, >> >>>> >> >>>> sufian >> >>>> >> >>> >> >> >> >> >> > >> > >> > >> >> >> >
