I am running a polymer in solvent simulation with enhanced sampling. To run the enhanced sampling along my collective variable of choice, radius of gyration, I am using the PLUMED software.
For a sanity check, I decided to run PLUMED along with LAMMPS to ensure they are computing things correctly. I ran a simple NVT simulation and I am seeing a slight discrepancy in the Rg value reported by LMP and PLUMED.
As a quick snippet, these are the two files COLVARS (from PLUMED) and rad_gyr.avg (from LAMMPS) combined horizontally?
#! FIELDS time rg|# TimeStep c_rg_pol[0]
0.200000 24.208169|100 24.038
0.202000 24.209314|101 24.0382
0.204000 24.209620|102 24.0382
0.206000 24.208946|103 24.038
0.208000 24.207323|104 24.0377
0.210000 24.204950|105 24.0372
0.212000 24.202167|106 24.0365
0.214000 24.199378|107 24.0357
0.216000 24.196896|108 24.0347
0.218000 24.194880|109 24.0336
0.220000 24.193277|110 24.0324
Though the deviation is not large, it exists.
As another sanity check, I decided to run my own calculation of Rg using MDAnalysis. Turns out, LAMMPS is in agreement with MDAnalysis.
My question is, is this just a minor floating point issue, or is there a deeper bug at play here that I am not catching? I have asked on the plumed groups but have not heard from them.
I have attached my plumed.dat file, lammps input file, datafile and settingsfile to this message.
plumed.dat (106 Bytes)
pnipam_tip4p-ice-water.settings (4.4 KB)
pnipam_tip4p-ice-water.data (6.2 MB)
in.plumed (3.2 KB)