In the examples of our papers we used blocksizes of the order of 5-20 frames, 4000 frames in total i think for the newest ones with 1000 molecules (error in force ~1%). It also depends how you choose your spline grid (in our case ~30 segments per non-bonded interaction). At some point the direct solver is inefficient, then you'll have to decrease block size or make spline grid more coarse.
2011/3/11 Sikandar Mashayak <[email protected]>: > yeah, sounds interesting .. so i will do two test cases, in first - taking > full trajectory from gromacs, I will do mdrun -rerun and exclude any CG1-CG1 > interactions then use the output trajectory to FMatch CG1-CG2, CG2-CG2 > simultaneously. > In second, will remove CG1-CG2 interactions as well to get CG2-CG2 and > vice-versa. Will get back to you once i get the results. > Also, on another thought, I understand this varies from case to case, but > how many frames are usually ok for Fmatch to converge and what value of > tolerance that is limit on error (3rd colmn of cg-cg.force) is recommended > to decide convergence... > thanks > sikandar > > On Fri, Mar 11, 2011 at 4:06 AM, Victor Ruehle <[email protected]> wrote: >> >> FM tries to fit the total force, if you have some fixed components, >> you'll have to subtract them first, or exlcude in a rerun. so in your >> case you would have to exclude CG1-CG1 interactions or subtract >> corresponding forces (csg_fmatch --force-trj, documentation will >> follow). So csg_fmatch only does the mapping of total forces in >> trajectory + fit of all specified interactions at once. >> >> So separation is an issue, and you'll have to decide what works best >> for you particular system. I don't think there is a general rule how >> to proceed. So you can exclude/subtract CG1-CG1, then fit CG1-CG2 + >> CG2-CG2 simultaneously or try to exclude one of em as well and do the >> fit individually. In any case I'm interested how it works out and can >> give advise on how to set this up in VOTCA/gromacs. >> >> 2011/3/11 Sikandar Mashayak <[email protected]>: >> > I am kinda confused here. Lets say I have mixture of two molecules, one >> > with >> > more than one atoms and second one is just simple one atom molecule, >> > call it >> > as CG1. >> > Then I wish to coarse-grain multi-atom molecule to its COM site, call it >> > as >> > CG2 and now determine what is the effective CG potentials between >> > CG2-CG2, >> > CG1-CG2, I dont have to worry about CG1-CG1 which I know already. >> > The trajectory contains the force information due to all possible >> > interactions. So to determine CG2-CG2, do I have to eliminate any >> > interaction between CG2-CG1, and same for CG1-CG2 i.e. eliminate >> > CG2-CG2,CG1-CG1. Or FM takes of these by itself. That is to say when >> > computing CG2-CG2 it only considers force acting on CG2 site only >> > because of >> > CG2-CG2 interaction and not any other. >> > Yeah, it would be great if you could forward me the paper you submitted >> > in >> > regard to this topic. >> > thanks >> > sikandar >> > On Thu, Mar 10, 2011 at 5:34 PM, Victor Ruehle <[email protected]> wrote: >> >> >> >> Hey, >> >> >> >> how do you want do describe interactions between coarse-grained and >> >> atomistic molecules? Each atom with coarse-grained site? How many >> >> different atom types and with that interactions will you have then? >> >> Also be aware that mixing resolutions might lead to additional >> >> problems. >> >> >> >> Currently the only way to keep one molecule at atomistic resolution in >> >> force matching is by specifying a 1:1 mapping files, which maps each >> >> atom to one coarse-grained bead. However, what you plan to do is not >> >> trivial since your trajectory has the full forces, also the ones which >> >> you don't want to parametrize. If you can do a clean splitting, either >> >> by excluding interactions, or by subtracting already known ones (this >> >> feature was added in 1.1) this might work. We recently submitted a >> >> paper on this, I can send you the draft if you want. >> >> >> >> Victor >> >> >> >> 2011/3/10 Sikandar Mashayak <[email protected]>: >> >> > Hi >> >> > I want to perform a Force Matching on a reference system which >> >> > contains >> >> > two >> >> > types of molecules. I am planning to CG one of it and keep the other >> >> > in >> >> > all-atom description. >> >> > So I am wondering how do I specify options in fmatch.xml file? Will I >> >> > have >> >> > to specify non-bonded interactions for CG-CG and each pair of >> >> > CG-atoms >> >> > of >> >> > other molecule? >> >> > thanks >> >> > sikandar >> >> > >> >> > -- >> >> > You received this message because you are subscribed to the Google >> >> > Groups >> >> > "votca" group. >> >> > To post to this group, send email to [email protected]. >> >> > To unsubscribe from this group, send email to >> >> > [email protected]. >> >> > For more options, visit this group at >> >> > http://groups.google.com/group/votca?hl=en. >> >> > >> >> >> >> -- >> >> You received this message because you are subscribed to the Google >> >> Groups >> >> "votca" group. >> >> To post to this group, send email to [email protected]. >> >> To unsubscribe from this group, send email to >> >> [email protected]. >> >> For more options, visit this group at >> >> http://groups.google.com/group/votca?hl=en. >> >> >> > >> > -- >> > You received this message because you are subscribed to the Google >> > Groups >> > "votca" group. >> > To post to this group, send email to [email protected]. >> > To unsubscribe from this group, send email to >> > [email protected]. >> > For more options, visit this group at >> > http://groups.google.com/group/votca?hl=en. >> > >> >> -- >> You received this message because you are subscribed to the Google Groups >> "votca" group. >> To post to this group, send email to [email protected]. >> To unsubscribe from this group, send email to >> [email protected]. >> For more options, visit this group at >> http://groups.google.com/group/votca?hl=en. >> > > -- > You received this message because you are subscribed to the Google Groups > "votca" group. > To post to this group, send email to [email protected]. > To unsubscribe from this group, send email to > [email protected]. > For more options, visit this group at > http://groups.google.com/group/votca?hl=en. > -- You received this message because you are subscribed to the Google Groups "votca" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/votca?hl=en.
