Hey Christoph, "> I guess if you would have an infinitely long trajectory and hence > infinitely long blocks the ends should go to 0. > We implemented this after https://doi.org/10.1007/978-94-010-0173-1_7, > so you can have a look at that as well. "
I agree, but only in a scenario where the potential form of choice (pairwise for example) is indeed such that allows you to reproduce the forces. Maybe some systems require multi-body terms or some things like that. In any case, sorry, because my sentence " I mean, how do they the value of (F_il,model - Fil,ref)^2, which should ideally be 0, at each point?" was not well written. it should be "I mean, how do *these errors relate to* the value of (F_il,model - Fil,ref)^2, (...), at each point?" What I mean is: if you have for example more than one superatom type, you will have minimum 3 pairwise potentials: 1-1, 1-2, 2-2. Each of them will be a separate .force output file and have its own reported "error". But if you look at force-matching's formalism as develop by Voth (which is the one implemented in VOTCA), they work with equations for the global, total force acting on each atom i. So in fact I was curious to know what are these individual errors of 1-1, 1-2 and 2-2 (i.e., how are they calculated by VOTCA?). "> The paper is from 2009, let me see if I can still find it. Please send > me your github username in an email." I sent you an email :) If I somehow emailed the wrong Christoph Junghans and you didnt receive anything, you can feel free to email me instead: [email protected] Em sexta-feira, 16 de junho de 2023 às 15:28:52 UTC+2, Christoph Junghans escreveu: > On Fri, Jun 16, 2023 at 4:21 AM Cecília Álvares > <[email protected]> wrote: > > > > Hello, > > > > I understand that in the .force output files that are generated by VOTCA > after force-matching the third column is the force error (including the > block-averaging procedure, in case more I am using several > overdetermined-systems of equations). > > > > I was wondering exactly these errors are (mathematically speaking). I > mean, how do they the value of (F_il,model - Fil,ref)^2, which should > ideally be 0, at each point? > > I guess if you would have an infinitely long trajectory and hence > infinitely long blocks the ends should go to 0. > We implemented this after https://doi.org/10.1007/978-94-010-0173-1_7, > so you can have a look at that as well. > > > > > Also, if it is not much to ask, would someone have ready-to-go the > magnitude ( +/- ) of these errors in the force-fields derived via FM for > propane in the original VOTCA publication? ( > https://doi.org/10.1021/ct900369w) I know I can run this myself and > check, so it is just in case someone happens to have this information > readily. It would be quite helpful. I am asking just to have a reference on > how small these errors should be: I want to be sure that the values in my > case are not too large (which would indicate that maybe my system is not > well described by pairwise potentials for example). Given that the FM > force-fields for hexane look quite beautiful, I wanted to use the > underlying error magnitude in the potential as a "reference" of how large > they should be. > > The paper is from 2009, let me see if I can still find it. Please send > me your github username in an email. > > Christoph > > > > > Thanks a lot! > > Cecilia > > > > -- > > Join us on Slack: https://join.slack.com/t/votca/signup > > --- > > 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 view this discussion on the web visit > https://groups.google.com/d/msgid/votca/eb34d925-de3c-4468-bb01-c07078abdc75n%40googlegroups.com > . > > > > -- > Christoph Junghans > Web: http://www.compphys.de > -- Join us on Slack: https://join.slack.com/t/votca/signup --- 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 view this discussion on the web visit https://groups.google.com/d/msgid/votca/45f93ad4-a0d5-4dfc-a3a3-5bb2d779ab2bn%40googlegroups.com.
