From chemistry-request@server.ccl.net Sat Nov 17 12:00:37 2001 Received: from gandalf.cber.nih.gov ([128.231.52.5]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id fAHH0aB29449 for ; Sat, 17 Nov 2001 12:00:36 -0500 Received: from localhost (rvenable@localhost) by gandalf.cber.nih.gov (980427.SGI.8.8.8/980728.SGI.AUTOCF) via ESMTP id LAA27714; Sat, 17 Nov 2001 11:55:26 -0500 (EST) Date: Sat, 17 Nov 2001 11:55:26 -0500 From: Rick Venable To: Ibrahim Moustafa cc: chemistry@ccl.net Subject: Re: CCL:Velocity re-assignment In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII There are several methods and reasons for modifying MD velocities: + periodic assignment during equilibration for NVE dynamics + periodic scaling during equilibration for NVE dynamics + changed every timestep, coupled to a piston for NVT dynamics NVE and NVT refer to constant volume ensembles, with explicit conserved system Hamiltonians; velocities are not changed during NVE dynamics, but 100 ps or more of equilibration with velocity adjustment is usually needed to get a system w/o large temperature drift. One should not include the equilibration time period in the calculation of molecular properties. The use of a piston to "extend" the system (adding degrees of freedom) is often used to provide true NVT dynamics; the potential and kinetic energy of the piston are included in the system Hamiltonian. The same method can be applied to NPT (const T,P) and NPH (const P) ensembles in a rigorous way, from the viewpoint of statistical mechanics. I should note that a method termed "weak coupling", popularized by HC Berendsen's implementation, has also been used for constant T simulations; a theoretical issue with that approach is that the ensemble isn't known, and hence neither is the Hamiltonian. The extended system approach is preferred, due to the rigorous stat mech ensemble. There's a more complete discussion of these issues in: "Computer Simulation of Liquids", MP Allen and DJ Tildesley, Clarendon Press, Oxford UK (1987; pbk 1989,1990). Feller, Zhang, Pastor & Brooks, JCP, 103, 4613 (1995) The JCP paper discusses the extended system implementation for constant T and/or constant P in CHARMM, and compares it to the weak coupling algorithm. On Fri, 16 Nov 2001, Ibrahim Moustafa wrote: > I'm doing MD on my protein structure; in some recent work on > simulation using MD at const. temp. I've seen people re-assign velocities > every certain Intervals. > > I want to know excactly how to re-assign velocities periodically? > Is it by changing the seed value randomly periodically? or they use > another way to do that? > > What are the benefits for the simulation from velocity re-assignment? > Doesn't this drift the trajectory everytime velocities being > reassigned?? Is this good for simulation? > > I'd be appreciated if anyone calrify this point. > > I use Tripos software to do the simulation. =+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+= Rick Venable FDA/CBER/OVRR Biophysics Lab 1401 Rockville Pike HFM-419 Rockville, MD 20852-1448 U.S.A. (301) 496-1905 Rick_Venable@nih.gov ALT email: rvenable@speakeasy.org =+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+= From chemistry-request@server.ccl.net Sat Nov 17 14:17:52 2001 Received: from hermes.bmc.uu.se ([130.238.39.245]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id fAHJHpB31764 for ; Sat, 17 Nov 2001 14:17:52 -0500 Received: from XRAY.bmc.uu.se ([130.238.37.10]) by hermes.bmc.uu.se (Post.Office MTA v3.5.3 release 223 ID# 0-62649U1500L1000S0V35) with ESMTP id se; Sat, 17 Nov 2001 20:17:32 +0100 Received: from zorn.bmc.uu.se (IDENT:spoel@zorn.bmc.uu.se [130.238.37.95]) by XRAY.bmc.uu.se (8.9.3/8.8.7) with ESMTP id UAA10319; Sat, 17 Nov 2001 20:20:00 +0100 Date: Sat, 17 Nov 2001 20:17:26 +0100 (CET) From: David van der Spoel X-Sender: spoel@zorn.bmc.uu.se To: Rick Venable cc: Ibrahim Moustafa , chemistry@ccl.net Subject: Re: CCL:Velocity re-assignment In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII On Sat, 17 Nov 2001, Rick Venable wrote: Without wanting to start a holy war... The Berendsen coupling does not give a well known stat mech ensemble for intermediate values of the coupling parameter tau_T, however in the limiting cases (tau_T -> 0, or tau_T -> oo) it is identical to well known ensembles. The clear advatange of the algorithm is that it works with minimal effect on the dynamics of the system (velocity modification < 1%). >There are several methods and reasons for modifying MD velocities: > >+ periodic assignment during equilibration for NVE dynamics >+ periodic scaling during equilibration for NVE dynamics >+ changed every timestep, coupled to a piston for NVT dynamics These algorithme do influence the dynamics of the system considerably because the trajectories are not continuous. Besides, they dont yield a known stat mech ensemble of your system, but of your system plus a piston... So you can ask yourself what does that mean... The real thing to do is avoid having to use a temperature scaling algorithm. Equilibrate at the temperature you need using any algorithm, then take care to conserve energy, by avoiding cut-offs. In periodic systems you will have to use Ewald summation or PME for the Coulomb interactions, and a smooth (long) cut-off for the non bondeds. In non periodic systems you will have to include all interactions. Groeten, David. ________________________________________________________________________ Dr. David van der Spoel, Biomedical center, Dept. of Biochemistry Husargatan 3, Box 576, 75123 Uppsala, Sweden phone: 46 18 471 4205 fax: 46 18 511 755 spoel@xray.bmc.uu.se spoel@gromacs.org http://zorn.bmc.uu.se/~spoel ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ From chemistry-request@server.ccl.net Sat Nov 17 14:56:40 2001 Received: from gandalf.cber.nih.gov ([128.231.52.5]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id fAHJueB32319 for ; Sat, 17 Nov 2001 14:56:40 -0500 Received: from localhost (rvenable@localhost) by gandalf.cber.nih.gov (980427.SGI.8.8.8/980728.SGI.AUTOCF) via ESMTP id OAA36035; Sat, 17 Nov 2001 14:50:37 -0500 (EST) Date: Sat, 17 Nov 2001 14:50:36 -0500 From: Rick Venable To: David van der Spoel cc: Ibrahim Moustafa , chemistry@ccl.net Subject: Re: CCL:Velocity re-assignment In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII On Sat, 17 Nov 2001, David van der Spoel wrote: > Without wanting to start a holy war... > The Berendsen coupling does not give a well known stat mech ensemble > for intermediate values of the coupling parameter tau_T, however in > the limiting cases (tau_T -> 0, or tau_T -> oo) it is identical to > well known ensembles. The clear advatange of the algorithm is that > it works with minimal effect on the dynamics of the system (velocity > modification < 1%). However, one does not typically use values close to zero or infinity, so in general, the ensemble isn't known. By taking care with the piston parameters, the extended systems also have minimal effect on the dynamics of the system. The early work by people such as HC Andersen and WG Hoover, and later studies by ML Klein and others have shown this. > On Sat, 17 Nov 2001, Rick Venable wrote: > >There are several methods and reasons for modifying MD velocities: > > > >+ periodic assignment during equilibration for NVE dynamics > >+ periodic scaling during equilibration for NVE dynamics > >+ changed every timestep, coupled to a piston for NVT dynamics > These algorithme do influence the dynamics of the system considerably > because the trajectories are not continuous. Besides, they dont yield a > known stat mech ensemble of your system, but of your system plus a > piston... So you can ask yourself what does that mean... As I stated clearly and you have omitted, it is not appropriate to include the equlibration period when computing properties from MD simulations. The problem is, as you indicate, that the periodic velocity adjustment makes the dynamics discontinuous. The use of a piston for const T does correspond to the NVT ensemble; or are you claiming that Allen, Tildesely, Andersen, Hoover, Klein, etc. are wrong? > The real thing to do is avoid having to use a temperature scaling > algorithm. Equilibrate at the temperature you need using any algorithm, > then take care to conserve energy, by avoiding cut-offs. In periodic > systems you will have to use Ewald summation or PME for the Coulomb > interactions, and a smooth (long) cut-off for the non bondeds. > In non periodic systems you will have to include all interactions. I agree about equilibration (and said so in my orignal message), and also about the use of Ewald methods, especially for true periodic systems. However, a protein in a cube (etc.) of water with periodic images isn't really a true periodic system, because the effective concentration is much higher than typical solution experimantal work. In most cases, one doesn't want the field effect of the protein in the adjoining image cell, and a cutoff may be needed to exclude it. =+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+= Rick Venable FDA/CBER/OVRR Biophysics Lab 1401 Rockville Pike HFM-419 Rockville, MD 20852-1448 U.S.A. (301) 496-1905 Rick_Venable@nih.gov ALT email: rvenable@speakeasy.org =+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+= From chemistry-request@server.ccl.net Sat Nov 17 16:28:34 2001 Received: from bute.st-andrews.ac.uk (im17@[138.251.12.1]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id fAHLSYB01360 for ; Sat, 17 Nov 2001 16:28:34 -0500 Received: from localhost (im17@localhost) by bute.st-andrews.ac.uk (8.9.1a/8.9.1) with SMTP id VAA18433; Sat, 17 Nov 2001 21:17:14 GMT Date: Sat, 17 Nov 2001 21:17:14 +0000 (GMT) From: Ibrahim Moustafa X-Sender: im17@st-andrews.ac.uk To: Rick Venable cc: David van der Spoel , chemistry@ccl.net Subject: Re: CCL:Velocity re-assignment In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII The debate on which algorithm is better in MD (NTV) is still going; and the use of the extended method is more rigorous stast. mech. Actually, I have to use weak coupling method coz this algorithm is already implemented in SYBYL while the extended algorithm is not. I still need to know: given that i use weak coupling algorithm for MD-NVT with coupling factor 0.1 ps, how can I re-assign the velocity in a proper way such that not biasing the simulation results; is that by changing the seed value periodically by entering different seed values in the input for the simulation? if so, does this cause any harm to the simulation??!! Thanks for all contributed to calarify this point. Name :Ibrahim M.Moustafa Mail address :Center for Biomolecular Science, BIOMOLECULAR SCIENCE BUILDING, North Haugh,St-Andrews, Fife,KY16 9ST Scotland,U.K. Tel :+44(0)1334-467257 E-mail :im17@st-andrews.ac.uk On Sat, 17 Nov 2001, Rick Venable wrote: > On Sat, 17 Nov 2001, David van der Spoel wrote: > > Without wanting to start a holy war... > > The Berendsen coupling does not give a well known stat mech ensemble > > for intermediate values of the coupling parameter tau_T, however in > > the limiting cases (tau_T -> 0, or tau_T -> oo) it is identical to > > well known ensembles. The clear advatange of the algorithm is that > > it works with minimal effect on the dynamics of the system (velocity > > modification < 1%). > > However, one does not typically use values close to zero or infinity, so > in general, the ensemble isn't known. By taking care with the piston > parameters, the extended systems also have minimal effect on the > dynamics of the system. The early work by people such as HC Andersen > and WG Hoover, and later studies by ML Klein and others have shown this. > > > On Sat, 17 Nov 2001, Rick Venable wrote: > > >There are several methods and reasons for modifying MD velocities: > > > > > >+ periodic assignment during equilibration for NVE dynamics > > >+ periodic scaling during equilibration for NVE dynamics > > >+ changed every timestep, coupled to a piston for NVT dynamics > > These algorithme do influence the dynamics of the system considerably > > because the trajectories are not continuous. Besides, they dont yield a > > known stat mech ensemble of your system, but of your system plus a > > piston... So you can ask yourself what does that mean... > > As I stated clearly and you have omitted, it is not appropriate to > include the equlibration period when computing properties from MD > simulations. The problem is, as you indicate, that the periodic > velocity adjustment makes the dynamics discontinuous. > > The use of a piston for const T does correspond to the NVT ensemble; or > are you claiming that Allen, Tildesely, Andersen, Hoover, Klein, etc. > are wrong? > > > The real thing to do is avoid having to use a temperature scaling > > algorithm. Equilibrate at the temperature you need using any algorithm, > > then take care to conserve energy, by avoiding cut-offs. In periodic > > systems you will have to use Ewald summation or PME for the Coulomb > > interactions, and a smooth (long) cut-off for the non bondeds. > > In non periodic systems you will have to include all interactions. > > I agree about equilibration (and said so in my orignal message), and > also about the use of Ewald methods, especially for true periodic > systems. However, a protein in a cube (etc.) of water with periodic > images isn't really a true periodic system, because the effective > concentration is much higher than typical solution experimantal work. > In most cases, one doesn't want the field effect of the protein in the > adjoining image cell, and a cutoff may be needed to exclude it. > > =+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+= > Rick Venable > FDA/CBER/OVRR Biophysics Lab > 1401 Rockville Pike HFM-419 > Rockville, MD 20852-1448 U.S.A. > (301) 496-1905 Rick_Venable@nih.gov > ALT email: rvenable@speakeasy.org > =+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+= > > > > > -= This is automatically added to each message by mailing script =- > CHEMISTRY@ccl.net -- To Everybody | CHEMISTRY-REQUEST@ccl.net -- To Admins > MAILSERV@ccl.net -- HELP CHEMISTRY or HELP SEARCH > CHEMISTRY-SEARCH@ccl.net -- archive search | Gopher: gopher.ccl.net 70 > Ftp: ftp.ccl.net | WWW: http://www.ccl.net/chemistry/ | Jan: jkl@ccl.net > > > > > > From chemistry-request@server.ccl.net Sat Nov 17 16:55:22 2001 Received: from hermes.bmc.uu.se ([130.238.39.245]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id fAHLtMB01806 for ; Sat, 17 Nov 2001 16:55:22 -0500 Received: from XRAY.bmc.uu.se ([130.238.37.10]) by hermes.bmc.uu.se (Post.Office MTA v3.5.3 release 223 ID# 0-62649U1500L1000S0V35) with ESMTP id se; Sat, 17 Nov 2001 22:55:03 +0100 Received: from zorn.bmc.uu.se (IDENT:spoel@zorn.bmc.uu.se [130.238.37.95]) by XRAY.bmc.uu.se (8.9.3/8.8.7) with ESMTP id WAA11559; Sat, 17 Nov 2001 22:57:31 +0100 Date: Sat, 17 Nov 2001 22:54:57 +0100 (CET) From: David van der Spoel X-Sender: spoel@zorn.bmc.uu.se To: Rick Venable cc: Ibrahim Moustafa , chemistry@ccl.net Subject: Re: CCL:Velocity re-assignment In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII On Sat, 17 Nov 2001, Rick Venable wrote: > >However, one does not typically use values close to zero or infinity, so >in general, the ensemble isn't known. By taking care with the piston >parameters, the extended systems also have minimal effect on the >dynamics of the system. The early work by people such as HC Andersen >and WG Hoover, and later studies by ML Klein and others have shown this. I was basically recalling what is in my understanding in this paper: >@Article{Morishita2000a, > author = {T. Morishita}, > title = {Fluctuation formulas in molecular-dynamics simulations >with the weak coupling heat bath}, > journal = {J. Chem. Phys.}, > year = 2000, > volume = 113, > pages = {2976-2982} >} I am not claiming that the older papers by Klein, Hoover etc. are wrong, but if you use a piston to keep the temperature this becomes part of the system, and hence you are not simulating the system you think you do. As far as I know most coupling methods influence the temperature fluctuations (necessary for computing heat capacities etc.), which can be quite easily seen by taking the fourier transform of the temperature fluctuations. These usually have a peak corresponding to the coupling constant of your piston. >The use of a piston for const T does correspond to the NVT ensemble; or >are you claiming that Allen, Tildesely, Andersen, Hoover, Klein, etc. >are wrong? See above. >I agree about equilibration (and said so in my orignal message), and >also about the use of Ewald methods, especially for true periodic >systems. However, a protein in a cube (etc.) of water with periodic >images isn't really a true periodic system, because the effective >concentration is much higher than typical solution experimantal work. >In most cases, one doesn't want the field effect of the protein in the >adjoining image cell, and a cutoff may be needed to exclude it. One should always ask: "how do I minimise the artefacts on my system of interest." PME I think is less bad than cut-offs... One day when I am bored I will do the simulations with PME with ever increasing box size, and check the interaction between the protein and its image. See how far I have to go to simulate a protein at infinite dilution. On a side note, under physiological conditions (i.e. in a cell) the protein density is quite high. So why not simulate three or four proteins in a single simulation box with PME? The different proteins will average out any serious "crystal" artefacts due to PME and you can write four papers based on one extremely realistic simulation... Groeten, David. ________________________________________________________________________ Dr. David van der Spoel, Biomedical center, Dept. of Biochemistry Husargatan 3, Box 576, 75123 Uppsala, Sweden phone: 46 18 471 4205 fax: 46 18 511 755 spoel@xray.bmc.uu.se spoel@gromacs.org http://zorn.bmc.uu.se/~spoel ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ From chemistry-request@server.ccl.net Sat Nov 17 18:05:58 2001 Received: from hotmail.com ([216.32.181.118]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id fAHN5vB04292 for ; Sat, 17 Nov 2001 18:05:58 -0500 Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC; Sat, 17 Nov 2001 15:05:44 -0800 Received: from 164.68.50.20 by lw2fd.hotmail.msn.com with HTTP; Sat, 17 Nov 2001 23:05:44 GMT X-Originating-IP: [164.68.50.20] From: "Luke Sweet" To: chemistry@ccl.net Subject: g98 compilation using g77 Date: Sat, 17 Nov 2001 15:05:44 -0800 Mime-Version: 1.0 Content-Type: text/html Message-ID: X-OriginalArrivalTime: 17 Nov 2001 23:05:44.0780 (UTC) FILETIME=[62F458C0:01C16FBC]
I am trying to compile g98 rev A.11 source code running Linux 7.1 on a AMD k6-3 using the g77 and gcc compilers.  I have read that AMD processors are not the greatest for running g98 on but I just want to figure out how to compile the code without using the portland group compiler (I plan on getting a PIII later).  I have read that g98 has been compiled successfully with g77 and that most of the fixes are in the i386.make file.  I have make a couple changes but now I am stuck as to what to do next.  If anyone has any advise on other things that need to be changed or an i386.make file that would work for me, please respond.  Here is the end of the bldg98.log file that I get when trying to compile.
 
g77   -O2    -c dummy.f
rm -f dummy.f dummy.c
ld -shared -o util.so dummy.o --whole-archive util.a \
--no-whole-archive
gau-fsplit -e ml0 bsd/main.F
ml0.F already exists, over-written
make -f bsd/g98.make OPTFLAG=' -O2   '  ml0.o
make[1]: Entering directory `/luke/sweetle/g98/g98'
gau-cpp -I/luke/sweetle/g98/g98 -I/luke/sweetle/g98/g98s -DGAUSS_PAR -DGAUSS_THPAR  -DDEFMAXSHL=20000 -DDEFMAXATM=20000 -DDEFMAXNZ=20000 -DDEFNVDIM=257 -DDEFARCREC=1024 -DMERGE_LOOPS -DUSE_ESSL -D_I386_ -DLITTLE_END -DUSING_F2C -DDEFMAXIOP=100 -DDEFMAXCHR=1024 -DDEFLMAX=13 -DDEFN3MIN=10 -DDEFMAXHEV=2000 -DDEFCACHE=64 -DDEFMAXLECP=10 -DDEFMAXFUNIT=5 -DDEFMAXFFILE=10000 -DDEFMAXFPS=1300 -DDEFMAXINFO=200 -DDEFMAXOP=120 -DDEFMAXTIT=100 -DDEFMAXRTE=4000 -DDEFMAXOV=500 -D_ALIGN_CORE_  -DO_BKSPEF -DDEFMXTS=1500 -DDEFMXBAS=500 -DDEFMXOPT=50 -DDEFMXBOND=12 -DDEFMXSPH=250 -DDEFMXINV=1500   ml0.F ml0.f
g77  -O2    -c ml0.f
rm -f ml0.f ml0.c
make[1]: Leaving directory `/luke/sweetle/g98/g98'
g77   -O2    -o g98 ml0.o   util.so
ml0.o: In function `MAIN__':
ml0.o(.text+0x7c): undefined reference to `swflag_'
util.so: undefined reference to `gtshls_'
util.so: undefined reference to `swappt_'
util.so: undefined reference to `gtzcrp_'
util.so: undefined reference to `gtoeps_'
util.so: undefined reference to `gsqrt_'
util.so: undefined reference to `symnum_'
util.so: undefined reference to `symscl_'
util.so: undefined reference to `gtrssy_'
util.so: undefined reference to `symten_'
util.so: undefined reference to `symtop_'
util.so: undefined reference to `gord93_'
util.so: undefined reference to `gran_'
util.so: undefined reference to `wait_'
util.so: undefined reference to `grouptr_'
util.so: undefined reference to `gredun_'
util.so: undefined reference to `gtzcrd_'
util.so: undefined reference to `gvwrite_'
util.so: undefined reference to `gtpqil_'
util.so: undefined reference to `gredfc_'
util.so: undefined reference to `gsin_'
util.so: undefined reference to `fork_'
util.so: undefined reference to `symcor_'
util.so: undefined reference to `symant_'
util.so: undefined reference to `symmo_'
util.so: undefined reference to `gtinct_'
util.so: undefined reference to `gsetmp_'
util.so: undefined reference to `symtr6_'
util.so: undefined reference to `gsign_'
util.so: undefined reference to `gtzcrs_'
util.so: undefined reference to `gxmpy_'
collect2: ld returned 1 exit status
make: *** [g98] Error 1
endif


Luke Sweet
Lake Forest College
 
 

Get your FREE download of MSN Explorer at http://explorer.msn.com