I am simulating aqueous uranyl nitrate solution confined inside zeolite NaY having 15 uranyl ion, 30 nitrate ion and 1536 water molecules to maintain the concentration of 0.5M. Since uranyl ion is much heavier than nitrate and water I want to implement rRESPA to integrate uranyl ion for the longer time step (10fs or more) and time step for the rest of the molecules is 1fs. Without rRESPA my simulation is running perfectly fine. I am struggling to implement rRESPA. I have attached my input. Kindly help me.
Any help will be highly appreciated.
With Best Regards
Indian Institute of Science.
in.unzeowater (6.79 KB)
you are misunderstanding how r-RESPA works. it is not about how fast individual particles move but about how quickly forces change.
so while you may have heavy atoms moving slowly, they are interacting with much faster moving particles and thus their forces change correspondingly quickly, which means you cannot apply r-RESPA to those interactions.
however, you are wasting a lot of time by using ewald instead of pppm, and there is no point in using ewald/disp, if you are not using pair styles that require it. it will just be the same as plain ewald.
that doesn’t mean that you cannot use r-RESPA, e.g. with multiple cutoff or between bonded, non-bonded, and kspace interactions.
how this is done is documented in the documentation for the run_style command.
using r-RESPA requires very careful testing and observation of energy conservation.
i also strongly recommend to use run style verlet for (initial) equilibration and then do energy conservation tests with that fist to set a benchmark for what you can reach with r-RESPA.