Thread: [Apbs-users] question on solution of PB equation
Biomolecular electrostatics software
Brought to you by:
sobolevnrm
From: magdalena g. <mag...@wp...> - 2006-11-14 12:59:28
|
hello, i have a question concerninng the solution of the PB equation itself, or more precisly the definition of the problem to be solved. i understand, that pmg encompasses the (Fortran in general) routines that solve the general eliptic differential equation. so the nonlinearity (the function c(u) given in these Fortran routines) has to be defined before enetering given routine. where then, it happens? which procedure initiates the solver with the nonlinearity? in case of PB this is actually the form of mobile charge distribution. but where (in the source) to we actually tell the solver how this distribution looks like? with kind regards magdalena ---------------------------------------------------- Champions On Ice - Olimpijscy mistrzowie tańca na lodzie! Zobacz spektakularne widowisko - Kliknij: http://klik.wp.pl/?adr=http%3A%2F%2Fadv.reklama.wp.pl%2Fas%2Fd33.html&sid=927 |
From: Vyom S. <vysharma@MIT.EDU> - 2006-11-14 17:00:35
|
Hi- I have a few questions about how one gets rid of the self energy contribution from the electrostatic energy. From what I understand this is done by doing two calculations, the first for the system whose potential we want to determine(I will call it the test system) and second using the same system but this time suspended in a medium of same dielectric constant(which I call the reference system). One then subtracts the reference electrostatic energy from the test system electrostatic energy to get the electrostatic energy of the test system less the self energy contribution. 1. Is it assumed that the Kappa (or free ion concentration) for the reference medium is 0? And if so then this in principle means one is solving the 3D poisson equation? 2. On assuming periodic boundary conditions (where one doesn't specify the potential value anywhere) does the difference in energies turn out to be the same as in the case of Dirichlet boundary condition? 3. For interactions among different molecules in the same system, one would have to add the dispersion energy terms to the electrostatic energy as well (here I assume this includes only the VDW term and not the cavity term). In this case do the DLVO characteristics evolve as a function of salt concentration? [By DLVO characteristics I mean the eventual agglomeration of molecules on increasing salt concentration and the evolution of a secondary minima] Any anaswers/comments would be useful. Thanks Vyom |
From: Nathan B. <nat...@ma...> - 2006-11-16 12:33:19
|
Hi Vyom -- > I have a few questions about how one gets rid of the self energy > contribution from the electrostatic energy. From what I understand > this > is done by doing two calculations, the first for the system whose > potential we want to determine(I will call it the test system) and > second using the same system but this time suspended in a medium of > same dielectric constant(which I call the reference system). One then > subtracts the reference electrostatic energy from the test system > electrostatic energy to get the electrostatic energy of the test > system > less the self energy contribution. That's correct. > 1. Is it assumed that the Kappa (or free ion concentration) for the > reference medium is 0? Right. > And if so then this in principle means one is > solving the 3D poisson equation? Yes, but the discrete nature of the solution is important since we're interested in solving the 3D poisson equation for the same finite- discretization charge distribution (e.g., approximations of delta functions) used in the test system to consistently eliminate self- interaction energies. > 2. On assuming periodic boundary conditions (where one doesn't specify > the potential value anywhere) does the difference in energies turn out > to be the same as in the case of Dirichlet boundary condition? Not necessarily. If the periodic system is large enough, this approximation could work -- but it would definitely need to be tested. > 3. For interactions among different molecules in the same system, one > would have to add the dispersion energy terms to the electrostatic > energy as well (here I assume this includes only the VDW term and not > the cavity term). In this case do the DLVO characteristics evolve as a > function of salt concentration? [By DLVO characteristics I mean the > eventual agglomeration of molecules on increasing salt concentration > and the evolution of a secondary minima] I would guess the presence or absence of salt dependence would depend on what sorts of dispersion interactions you're thinking of (fluctuations in solution salt charges vs. fluctuations in charge distributions on the molecules vs. ...). Can you clarify? Thanks, Nathan -- Assistant Professor, Dept. of Biochemistry and Molecular Biophysics Center for Computational Biology, Washington University in St. Louis Web: http://cholla.wustl.edu/ |
From: Vyom S. <vysharma@MIT.EDU> - 2006-11-16 16:03:22
|
> > Yes, but the discrete nature of the solution is important since we're > interested in solving the 3D poisson equation for the same finite- > discretization charge distribution (e.g., approximations of delta > functions) used in the test system to consistently eliminate self- > interaction energies. So does it mean that the boundary grid points have their potentials fixed not by the absolute position of the charges but by the positions of these charges after they are distributed on the neighboring grid points?? As an example if a charge is at, say x=10.2 and I have grid points at 10.0 and 11.0 then I distribute this charge at 10.0 and 11.0 and use these distributed charge values to calculate the boundary potentials in the case of poisson solver (and of-course this is the same grid which I use to calculate the grid boundary potential values for solving the PBE as well and not the actual position of the charge). >> 3. For interactions among different molecules in the same system, one >> would have to add the dispersion energy terms to the electrostatic >> energy as well (here I assume this includes only the VDW term and not >> the cavity term). In this case do the DLVO characteristics evolve as a >> function of salt concentration? [By DLVO characteristics I mean the >> eventual agglomeration of molecules on increasing salt concentration >> and the evolution of a secondary minima] > > I would guess the presence or absence of salt dependence would depend > on what sorts of dispersion interactions you're thinking of > (fluctuations in solution salt charges vs. fluctuations in charge > distributions on the molecules vs. ...). Can you clarify? I am thinking of only fluctuation in salt concentration which could be thought of as a fluctuation in salt charges as well. However the charge distribution on the molecule remains fixed throughout. Thanks Vyom On Nov 16, 2006, at 7:33 AM, Nathan Baker wrote: > Hi Vyom -- > >> I have a few questions about how one gets rid of the self energy >> contribution from the electrostatic energy. From what I understand >> this >> is done by doing two calculations, the first for the system whose >> potential we want to determine(I will call it the test system) and >> second using the same system but this time suspended in a medium of >> same dielectric constant(which I call the reference system). One then >> subtracts the reference electrostatic energy from the test system >> electrostatic energy to get the electrostatic energy of the test >> system >> less the self energy contribution. > > That's correct. > >> 1. Is it assumed that the Kappa (or free ion concentration) for the >> reference medium is 0? > > Right. > >> And if so then this in principle means one is >> solving the 3D poisson equation? > > Yes, but the discrete nature of the solution is important since we're > interested in solving the 3D poisson equation for the same finite- > discretization charge distribution (e.g., approximations of delta > functions) used in the test system to consistently eliminate self- > interaction energies. > >> 2. On assuming periodic boundary conditions (where one doesn't specify >> the potential value anywhere) does the difference in energies turn out >> to be the same as in the case of Dirichlet boundary condition? > > Not necessarily. If the periodic system is large enough, this > approximation could work -- but it would definitely need to be tested. > >> 3. For interactions among different molecules in the same system, one >> would have to add the dispersion energy terms to the electrostatic >> energy as well (here I assume this includes only the VDW term and not >> the cavity term). In this case do the DLVO characteristics evolve as a >> function of salt concentration? [By DLVO characteristics I mean the >> eventual agglomeration of molecules on increasing salt concentration >> and the evolution of a secondary minima] > > I would guess the presence or absence of salt dependence would depend > on what sorts of dispersion interactions you're thinking of > (fluctuations in solution salt charges vs. fluctuations in charge > distributions on the molecules vs. ...). Can you clarify? > > Thanks, > > Nathan > > -- > Assistant Professor, Dept. of Biochemistry and Molecular Biophysics > Center for Computational Biology, Washington University in St. Louis > Web: http://cholla.wustl.edu/ > > > > ----------------------------------------------------------------------- > -- > Take Surveys. Earn Cash. Influence the Future of IT > Join SourceForge.net's Techsay panel and you'll get the chance to > share your > opinions on IT & business topics through brief surveys - and earn cash > http://www.techsay.com/default.php? > page=join.php&p=sourceforge&CID=DEVDEV > _______________________________________________ > apbs-users mailing list > apb...@li... > https://lists.sourceforge.net/lists/listinfo/apbs-users |
From: Nathan B. <nat...@ma...> - 2006-11-18 02:03:24
|
Hi Vyom -- > So does it mean that the boundary grid points have their potentials > fixed not by the absolute position of the charges but by the > positions of these charges after they are distributed on the > neighboring grid points?? No, the boundary point potentials are determined by the exact charge locations, not the interpolated charges. The idea is that since the boundaries need to be far away from charges anyway, there's no need to distinguish between exact and interpolated delta functions. > As an example if a charge is at, say x=10.2 and I have grid points > at 10.0 and 11.0 then I distribute this charge at 10.0 and 11.0 and > use these distributed charge values to calculate the boundary > potentials in the case of poisson solver (and of-course this is the > same grid which I use to calculate the grid boundary potential > values for solving the PBE as well and not the actual position of > the charge). That's not the way it's currently implemented in APBS but this could be changed. However, I'm not sure what you'd want to do this... > I am thinking of only fluctuation in salt concentration which could > be thought of as a fluctuation in salt charges as well. However the > charge distribution on the molecule remains fixed throughout. APBS uses only a mean field grand canonical version of the PB equation -- no salt fluctuations. Thanks, Nathan > On Nov 16, 2006, at 7:33 AM, Nathan Baker wrote: > >> Hi Vyom -- >> >>> I have a few questions about how one gets rid of the self energy >>> contribution from the electrostatic energy. From what I understand >>> this >>> is done by doing two calculations, the first for the system whose >>> potential we want to determine(I will call it the test system) and >>> second using the same system but this time suspended in a medium of >>> same dielectric constant(which I call the reference system). One >>> then >>> subtracts the reference electrostatic energy from the test system >>> electrostatic energy to get the electrostatic energy of the test >>> system >>> less the self energy contribution. >> >> That's correct. >> >>> 1. Is it assumed that the Kappa (or free ion concentration) for the >>> reference medium is 0? >> >> Right. >> >>> And if so then this in principle means one is >>> solving the 3D poisson equation? >> >> Yes, but the discrete nature of the solution is important since we're >> interested in solving the 3D poisson equation for the same finite- >> discretization charge distribution (e.g., approximations of delta >> functions) used in the test system to consistently eliminate self- >> interaction energies. >> >>> 2. On assuming periodic boundary conditions (where one doesn't >>> specify >>> the potential value anywhere) does the difference in energies >>> turn out >>> to be the same as in the case of Dirichlet boundary condition? >> >> Not necessarily. If the periodic system is large enough, this >> approximation could work -- but it would definitely need to be >> tested. >> >>> 3. For interactions among different molecules in the same system, >>> one >>> would have to add the dispersion energy terms to the electrostatic >>> energy as well (here I assume this includes only the VDW term and >>> not >>> the cavity term). In this case do the DLVO characteristics evolve >>> as a >>> function of salt concentration? [By DLVO characteristics I mean the >>> eventual agglomeration of molecules on increasing salt concentration >>> and the evolution of a secondary minima] >> >> I would guess the presence or absence of salt dependence would depend >> on what sorts of dispersion interactions you're thinking of >> (fluctuations in solution salt charges vs. fluctuations in charge >> distributions on the molecules vs. ...). Can you clarify? >> >> Thanks, >> >> Nathan >> >> -- >> Assistant Professor, Dept. of Biochemistry and Molecular Biophysics >> Center for Computational Biology, Washington University in St. Louis >> Web: http://cholla.wustl.edu/ >> >> >> >> --------------------------------------------------------------------- >> ---- >> Take Surveys. Earn Cash. Influence the Future of IT >> Join SourceForge.net's Techsay panel and you'll get the chance to >> share your >> opinions on IT & business topics through brief surveys - and earn >> cash >> http://www.techsay.com/default.php? >> page=join.php&p=sourceforge&CID=DEVDEV >> _______________________________________________ >> apbs-users mailing list >> apb...@li... >> https://lists.sourceforge.net/lists/listinfo/apbs-users > -- Assistant Professor, Dept. of Biochemistry and Molecular Biophysics Center for Computational Biology, Washington University in St. Louis Web: http://cholla.wustl.edu/ |
From: <vin...@dr...> - 2006-11-14 18:00:28
|
Magdalena --- The file I think you're looking for is in the 'mypde.f' routine in the MG= =20 subdirectory of the APBS source: apbs-0.4.0/src/mg/mypde.f In particular, the line(s) in the code of relevance are: line 384 in subroutine c_vec and line 517 in dc_vec where the non-linear=20 term its derivative are defined ... cheers, vince On Tue, 14 Nov 2006, magdalena gruziel wrote: > hello, > > i have a question concerninng the solution of the PB equation itself, > or more precisly the definition of the problem to be solved. > i understand, that pmg encompasses the (Fortran in general) routines that > solve the general eliptic differential equation. so the nonlinearity > (the function c(u) given in these Fortran routines) has to be defined b= efore enetering > given routine. > where then, it happens? which procedure initiates the solver > with the nonlinearity? in case of PB this is actually the form > of mobile charge distribution. but where (in the source) to we actually t= ell the solver how this distribution looks like? > > with kind regards > magdalena > > > ---------------------------------------------------- > Champions On Ice - Olimpijscy mistrzowie ta=F1ca na lodzie! > Zobacz spektakularne widowisko - Kliknij: > http://klik.wp.pl/?adr=3Dhttp%3A%2F%2Fadv.reklama.wp.pl%2Fas%2Fd33.html&s= id=3D927 > > > > ------------------------------------------------------------------------- > Using Tomcat but need to do more? Need to support web services, security? > Get stuff done quickly with pre-integrated technology to make your job ea= sier > Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronim= o > http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&dat= =3D121642 > _______________________________________________ > apbs-users mailing list > apb...@li... > https://lists.sourceforge.net/lists/listinfo/apbs-users > |
From: Nathan B. <nat...@ma...> - 2006-11-15 11:43:58
|
Hi Magdalena -- Vincent pointed you to the correct portion of the code for the =20 solver; the various charge distributions and problem coefficients are =20= specified in the Vpmg_fillco routines of vpmg.c. The details of the =20 solver methodology are also described in more detail in the following =20= references: Holst M, Saied F. Multigrid solution of the Poisson-Boltzmann =20 equation. Journal of Computational Chemistry. 14 (1), 105-13, 1993. Holst M, Kozack RE, Saied F, Subramaniam S. Protein electrostatics: =20 Rapid multigrid-based newton algorithm for solution of the full =20 nonlinear Poisson-Boltzmann equation. J Biomol Struct Dyn. 11 (6), =20 1437-45, 1994. Holst M, Kozack RE, Saied F, Subramaniam S. Treatment of =20 electrostatic effects in proteins: Multigrid-based newton iterative =20 method for solution of the full nonlinear Poisson-Boltzmann =20 equation. Proteins. 18 (3), 231-45, 1994. Holst MJ, Saied F. Numerical solution of the nonlinear Poisson-=20 Boltzmann equation: Developing more robust and efficient methods. =20 Journal of Computational Chemistry. 16 (3), 337-64, 1995. and in an excellent summary based on Mike Holst's thesis: http://=20 cam.ucsd.edu/~mholst/pubs/dist/Hols94d.pdf Thanks, Nathan On Nov 14, 2006, at 11:58 AM, vin...@dr... wrote: > > Magdalena --- > > The file I think you're looking for is in the 'mypde.f' routine in =20 > the MG subdirectory of the APBS source: > > apbs-0.4.0/src/mg/mypde.f > > In particular, the line(s) in the code of relevance are: > > line 384 in subroutine c_vec and line 517 in dc_vec where the non-=20 > linear term its derivative are defined ... > > cheers, > > vince > > > > On Tue, 14 Nov 2006, magdalena gruziel wrote: > >> hello, >> >> i have a question concerninng the solution of the PB equation itself, >> or more precisly the definition of the problem to be solved. >> i understand, that pmg encompasses the (Fortran in general) =20 >> routines that >> solve the general eliptic differential equation. so the nonlinearity >> (the function c(u) given in these Fortran routines) has to be =20 >> defined before enetering >> given routine. >> where then, it happens? which procedure initiates the solver >> with the nonlinearity? in case of PB this is actually the form >> of mobile charge distribution. but where (in the source) to we =20 >> actually tell the solver how this distribution looks like? >> >> with kind regards >> magdalena >> >> >> ---------------------------------------------------- >> Champions On Ice - Olimpijscy mistrzowie ta=C5=84ca na lodzie! >> Zobacz spektakularne widowisko - Kliknij: >> http://klik.wp.pl/?adr=3Dhttp%3A%2F%2Fadv.reklama.wp.pl%2Fas%=20 >> 2Fd33.html&sid=3D927 >> >> >> >> ---------------------------------------------------------------------=20= >> ---- >> Using Tomcat but need to do more? Need to support web services, =20 >> security? >> Get stuff done quickly with pre-integrated technology to make your =20= >> job easier >> Download IBM WebSphere Application Server v.1.0.1 based on Apache =20 >> Geronimo >> http://sel.as-us.falkag.net/sel?=20 >> cmd=3Dlnk&kid=3D120709&bid=3D263057&dat=3D121642 >> _______________________________________________ >> apbs-users mailing list >> apb...@li... >> https://lists.sourceforge.net/lists/listinfo/apbs-users > ----------------------------------------------------------------------=20= > --- > SF.net email is sponsored by: A Better Job is Waiting for You - =20 > Find it Now. > Check out Slashdot's new job board. Browse through tons of =20 > technical jobs > posted by companies looking to hire people just like you. > http://jobs.slashdot.org/=20 > _______________________________________________ > apbs-users mailing list > apb...@li... > https://lists.sourceforge.net/lists/listinfo/apbs-users -- Assistant Professor, Dept. of Biochemistry and Molecular Biophysics Center for Computational Biology, Washington University in St. Louis Web: http://cholla.wustl.edu/ |