Re: [gmx-users] Normalization in g_rdf
Ah, now I see what you meant. Then it is fine. That is a separate issue and is not really a problem for me. Thank you, Ondrej On Tue, Nov 24, 2009 at 17:39, Berk Hess wrote: > Hi, > > Sure g_rdf takes care of pressure coupling, but the cut-off of the rdf plot > is set from the initial frame. > So if your volume fluctuates a lot, you might have incorrect normalization > very close to the cut-off. > But in "normal" liquid simulations the box length will never change more > than 1%. > > Berk > >> Date: Tue, 24 Nov 2009 17:12:44 +0100 >> Subject: Re: [gmx-users] Normalization in g_rdf >> From: ondrej.marsa...@gmail.com >> To: gmx-users@gromacs.org >> >> Hi again, >> >> I thought pressure coupling is handled by g_rdf. Looking at the code, >> it seems that the block starting with: >> >> /* Must init pbc every step because of pressure coupling */ >> >> takes care of that, so even with a relatively big volume change at the >> beginning, it should not be a problem. Is that not the case? >> >> Ondrej >> >> >> On Tue, Nov 24, 2009 at 16:12, Berk Hess wrote: >> > Hi, >> > >> > g_rdf limits the distance to half the minimum periodic image distance. >> > So the volume normalization is always simply 4 pi r^2. >> > It uses the starting unit-cell with a factor of 0.99. >> > So you would only see strange effects when the box length reduces >> > more than 1% due to pressure coupling. >> > >> > Berk >> -- >> 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 > > > New Windows 7: Simplify what you do everyday. Find the right PC for you. > -- > gmx-users mailing list gmx-us...@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 > -- gmx-users mailing listgmx-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
RE: [gmx-users] Normalization in g_rdf
Hi, Sure g_rdf takes care of pressure coupling, but the cut-off of the rdf plot is set from the initial frame. So if your volume fluctuates a lot, you might have incorrect normalization very close to the cut-off. But in "normal" liquid simulations the box length will never change more than 1%. Berk > Date: Tue, 24 Nov 2009 17:12:44 +0100 > Subject: Re: [gmx-users] Normalization in g_rdf > From: ondrej.marsa...@gmail.com > To: gmx-users@gromacs.org > > Hi again, > > I thought pressure coupling is handled by g_rdf. Looking at the code, > it seems that the block starting with: > > /* Must init pbc every step because of pressure coupling */ > > takes care of that, so even with a relatively big volume change at the > beginning, it should not be a problem. Is that not the case? > > Ondrej > > > On Tue, Nov 24, 2009 at 16:12, Berk Hess wrote: > > Hi, > > > > g_rdf limits the distance to half the minimum periodic image distance. > > So the volume normalization is always simply 4 pi r^2. > > It uses the starting unit-cell with a factor of 0.99. > > So you would only see strange effects when the box length reduces > > more than 1% due to pressure coupling. > > > > Berk > -- > gmx-users mailing listgmx-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 _ New Windows 7: Simplify what you do everyday. Find the right PC for you. http://windows.microsoft.com/shop-- gmx-users mailing listgmx-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
Re: [gmx-users] Normalization in g_rdf
Hi again, I thought pressure coupling is handled by g_rdf. Looking at the code, it seems that the block starting with: /* Must init pbc every step because of pressure coupling */ takes care of that, so even with a relatively big volume change at the beginning, it should not be a problem. Is that not the case? Ondrej On Tue, Nov 24, 2009 at 16:12, Berk Hess wrote: > Hi, > > g_rdf limits the distance to half the minimum periodic image distance. > So the volume normalization is always simply 4 pi r^2. > It uses the starting unit-cell with a factor of 0.99. > So you would only see strange effects when the box length reduces > more than 1% due to pressure coupling. > > Berk -- gmx-users mailing listgmx-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
RE: [gmx-users] Normalization in g_rdf
Hi, g_rdf limits the distance to half the minimum periodic image distance. So the volume normalization is always simply 4 pi r^2. It uses the starting unit-cell with a factor of 0.99. So you would only see strange effects when the box length reduces more than 1% due to pressure coupling. Berk > Date: Tue, 24 Nov 2009 13:23:30 +0100 > Subject: Re: [gmx-users] Normalization in g_rdf > From: ondrej.marsa...@gmail.com > To: gmx-users@gromacs.org > > On Tue, Nov 24, 2009 at 13:01, Omer Markovitch wrote: > > > > On Tue, Nov 24, 2009 at 02:27, Ondrej Marsalek > > wrote: > >> > >> Dear all, > >> > >> I would like to understand better the way g_rdf performs > >> normalization. I have two unexpected results: > >> > >> 1) In a simple simulation of atomic ions in water in a cubic box, I > >> get RDFs that clearly reach a constant value at large enough > >> distances, but that value is somewhat lower than one. The simulation > >> is NpT, could that be a problem for the normalization? > >> > > g(r) should fluctuate around 1 at large distances, say 9 Angstroms for pure > > water. > > Indeed. However, see also Berk's remark. It seems to match very well > for my system. > > > > >> > >> 2) In a simulation in a dodecahedron, I get an unexpected decrease in > >> the RDF at larger distances (for free ions in solution). Is there some > >> know problem with normalization in triclinic cells? Is the RDF perhaps > >> not truncated "soon" enough? > > > > Try calculating it for much larger distances, if you have PBC this should > > not be a problem even with the current docecahedron. > > Maybe I am missing something, but the range that g_rdf considers is > hardwired - half the smallest box length with PBC, 3x the largest one > without PBC. Is there a way to include contributions from multiple > images of the unit cell? > > BTW, if g_rdf really only considers the box edge lengths and > disregards angles, it would make sense that normalization is broken in > general cells at larger distances, no? > > Ondrej > > > > --Omer. > > > > > > -- > > gmx-users mailing listgmx-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 > > > -- > gmx-users mailing listgmx-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 _ Express yourself instantly with MSN Messenger! Download today it's FREE! http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/-- gmx-users mailing listgmx-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
Re: [gmx-users] Normalization in g_rdf
On Tue, Nov 24, 2009 at 11:12, Berk Hess wrote: > Hi, > > If you have N molecules with total volume which is significantly less than > the unit-cell, > as is the case for ions in solvent, the RDF of the N molecules againt > themselves > will converge to (N-1)/N. > > Berk Hi, thanks for the reminder. It matches well for the system I mentioned. Still, I don't quite see why I get the same behavior for self RDF as well as cross RDF, but I will have a look at this later. Ondrej -- gmx-users mailing listgmx-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
Re: [gmx-users] Normalization in g_rdf
On Tue, Nov 24, 2009 at 13:01, Omer Markovitch wrote: > > On Tue, Nov 24, 2009 at 02:27, Ondrej Marsalek > wrote: >> >> Dear all, >> >> I would like to understand better the way g_rdf performs >> normalization. I have two unexpected results: >> >> 1) In a simple simulation of atomic ions in water in a cubic box, I >> get RDFs that clearly reach a constant value at large enough >> distances, but that value is somewhat lower than one. The simulation >> is NpT, could that be a problem for the normalization? >> > g(r) should fluctuate around 1 at large distances, say 9 Angstroms for pure > water. Indeed. However, see also Berk's remark. It seems to match very well for my system. > >> >> 2) In a simulation in a dodecahedron, I get an unexpected decrease in >> the RDF at larger distances (for free ions in solution). Is there some >> know problem with normalization in triclinic cells? Is the RDF perhaps >> not truncated "soon" enough? > > Try calculating it for much larger distances, if you have PBC this should > not be a problem even with the current docecahedron. Maybe I am missing something, but the range that g_rdf considers is hardwired - half the smallest box length with PBC, 3x the largest one without PBC. Is there a way to include contributions from multiple images of the unit cell? BTW, if g_rdf really only considers the box edge lengths and disregards angles, it would make sense that normalization is broken in general cells at larger distances, no? Ondrej > --Omer. > > > -- > gmx-users mailing list gmx-us...@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 > -- gmx-users mailing listgmx-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
Re: [gmx-users] Normalization in g_rdf
On Tue, Nov 24, 2009 at 02:27, Ondrej Marsalek wrote: > Dear all, > > I would like to understand better the way g_rdf performs > normalization. I have two unexpected results: > > 1) In a simple simulation of atomic ions in water in a cubic box, I > get RDFs that clearly reach a constant value at large enough > distances, but that value is somewhat lower than one. The simulation > is NpT, could that be a problem for the normalization? > > g(r) should fluctuate around 1 at large distances, say 9 Angstroms for pure water. > 2) In a simulation in a dodecahedron, I get an unexpected decrease in > the RDF at larger distances (for free ions in solution). Is there some > know problem with normalization in triclinic cells? Is the RDF perhaps > not truncated "soon" enough? > Try calculating it for much larger distances, if you have PBC this should not be a problem even with the current docecahedron. --Omer. -- gmx-users mailing listgmx-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
RE: [gmx-users] Normalization in g_rdf
Hi, If you have N molecules with total volume which is significantly less than the unit-cell, as is the case for ions in solvent, the RDF of the N molecules againt themselves will converge to (N-1)/N. Berk > Date: Tue, 24 Nov 2009 01:27:27 +0100 > From: ondrej.marsa...@gmail.com > To: gmx-users@gromacs.org > Subject: [gmx-users] Normalization in g_rdf > > Dear all, > > I would like to understand better the way g_rdf performs > normalization. I have two unexpected results: > > 1) In a simple simulation of atomic ions in water in a cubic box, I > get RDFs that clearly reach a constant value at large enough > distances, but that value is somewhat lower than one. The simulation > is NpT, could that be a problem for the normalization? > > 2) In a simulation in a dodecahedron, I get an unexpected decrease in > the RDF at larger distances (for free ions in solution). Is there some > know problem with normalization in triclinic cells? Is the RDF perhaps > not truncated "soon" enough? > > If these should work better than they do for me, I will of course > check my simulations again. I would just like to check first if this > is to be expected. > > Thanks, > Ondrej > -- > gmx-users mailing listgmx-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 _ New Windows 7: Simplify what you do everyday. Find the right PC for you. http://windows.microsoft.com/shop-- gmx-users mailing listgmx-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
[gmx-users] Normalization in g_rdf
Dear all, I would like to understand better the way g_rdf performs normalization. I have two unexpected results: 1) In a simple simulation of atomic ions in water in a cubic box, I get RDFs that clearly reach a constant value at large enough distances, but that value is somewhat lower than one. The simulation is NpT, could that be a problem for the normalization? 2) In a simulation in a dodecahedron, I get an unexpected decrease in the RDF at larger distances (for free ions in solution). Is there some know problem with normalization in triclinic cells? Is the RDF perhaps not truncated "soon" enough? If these should work better than they do for me, I will of course check my simulations again. I would just like to check first if this is to be expected. Thanks, Ondrej -- gmx-users mailing listgmx-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