Berk Hess wrote: > > > > Date: Mon, 1 Mar 2010 15:44:28 +0100 > > From: r.fried...@bioc.uzh.ch > > To: gmx-users@gromacs.org > > Subject: Re: [gmx-users] NVE of water > > > > Mark Abraham wrote: > > > On 2/03/2010 12:39 AM, Ran Friedman wrote: > > >> Hi, > > >> > > >> I would still argue that double precision is important. > > > > > > Oh? The discussion of Table 4 of > > > http://pubs.acs.org/doi/abs/10.1021/ct700301q (2008 GROMACS 4 JCTC > > > paper) suggested to me that single-precision NVE could be done well in > > > GROMACS. Am I missing something? > > > > > > Mark > > "The presented benchmarks were performed in the NVT ensemble" > (section IX). > > Or am I missing something? > > No, but everything that affects energy conservation in NVE also > affects it in NVT, > in addition the thermostat affect the integration accuracy in NVT (and > in NVT > you do not measure energy conservation from the total energy, but from the > conserved energy quantity). > > Double precision can be important for energy conservation, but often other > factors deteriorate the energy conservation orders of magnitude from what > can be reached in single precision already. Double precision is only > required > for testing or when you really need to generate an NVE ensemble. > > Berk Thanks for clarifying Berk - I was referring to the latter (i.e., when you really need to generate an NVE ensemble). I guess this is the main reason to run with NVE. Since most people are not running NVE with large systems or for production simulations, the added accuracy often justifies the use of double precision.
Best, Ran
-- gmx-users mailing list gmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/mailing_lists/users.php