Dear Steven, On Jun 4, 2013, at 10:26 AM, [email protected] wrote:
> > > On Monday, June 3, 2013 4:26:29 PM UTC+1, Christoph Junghans wrote: > 2013/6/3 <[email protected]>: > > Dear Votca Users, > > > > I wish to reproduce my atomistic simulation angles of my protein with CG > > between beads of residues. However I wish not to produce tables as an input > > to gromacs mdrun but get the equilibrium value and then adjust the spring > > constant to make it faster. > Tabulated vs functional form is at most 50% faster, but the bonded > interactions usually take 10% or less of the total simulation time, so > you will not gain much in the end. > > > > For this purpose shall I just create distributions or histograms normalized > > over the volume? Then place my spring where the distribution is max so set > > angle in gromacs to this value and adjust the force constant to match this > > distribution? Please, advise whether I am right. > You will have to use the volume normalize histogram, but csg_boltzmann > can do most of the calculation for you. > > Why don't you just take the table and fit k*(x-x_0)^2 to it? > > That is what I am planning to do.But which table? Shall I take the table > with distributions of bonds (or angles?) of the potential or volume > normalized histogram to fit k*(x-x_0)^2 ? > In the Gromacs topology file you write the parameters of the potential functions, therefore you need to fit the potential curve. > > Cheers, > > Christoph > > > Thanks for your help. I got gaussian distribution for my bonds so I can fit > it into k*(x-x_0)^2 and multiple gaussian distribution of my angles... Is > that normal? My beads involve 7 - 22 atoms, maybe that is why? Shall I use a > tables as an input to mdrun? > > Steven It is quite normal for CG distributions to not be gaussian. Since the CG system has less degrees of freedom compared to the atomistic one, a single CG bonded interaction represent by its nature several atomistic intramolecular interactions (bonds, angles, dihedrals, and intramolecular non bonded interaction). Even more a simple Boltzmann inversion of a distribution can cast into the potentials contributions coming from parts of the molecule (or of the system) that do not belong to the group of atoms that you are directly considering. Therefore the distributions you obtain are a sum of contributions each one with its own weight. It is extremely difficult to say a priory something about the shape of the final distribution. As it has been said the use of a table normally is not too expensive, therefore it is normally a good idea to use them. Anyhow the best thing you can do is to run two test system: one where you use tabulated potential and one where you use your "best" fitting. In this way you can see the difference between their performances and their results in your specific case. Cheers, Emiliano > > > > > Steven > > > > -- > > You received this message because you are subscribed to the Google Groups > > "votca" group. > > To unsubscribe from this group and stop receiving emails from it, send an > > email to [email protected]. > > To post to this group, send email to [email protected]. > > Visit this group at http://groups.google.com/group/votca?hl=en. > > For more options, visit https://groups.google.com/groups/opt_out. > > > > > > > > -- > Christoph Junghans > Web: http://www.compphys.de > > -- > You received this message because you are subscribed to the Google Groups > "votca" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To post to this group, send email to [email protected]. > Visit this group at http://groups.google.com/group/votca?hl=en. > For more options, visit https://groups.google.com/groups/opt_out. > > -- You received this message because you are subscribed to the Google Groups "votca" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/votca?hl=en. For more options, visit https://groups.google.com/groups/opt_out.
