(General) Experiment 4: Make explicit single, aromatic, and single bonds for each of the interpolated parameters and compare how it fares
DLM: For all these parameters there may not be matches
TG: We can switch on the params when there is a match with any molecule
JM: Here are the plots from my by-molecule experiments
DLM: Some molecules don’t have the cosine behaviour. Smirks Total count at the bottom represents the occurrence of the same pattern in the same molecule. In biphenyl it may repeat 4 times.
TG: Check the inner two indices which will give an idea of how many times the parameter is applied
JM: What next after k Vs wbo plots?
DLM: Do a first pass with molecules that have smirks count of 4, and that have greater than 4. IF there are more molecules in greater-than-4 plots then try to subdivide into TIG1 or other subset of TIG0 parameters, and plot k Vs wbo. Not worth doing this if there are less number of molecules, say 10.
Another task is to do the same experiment with a Carbon-Nitrogen central bond.
TG: For filtering out strong sterics build an openmm system and evaluate the energy with and without VdW+electrostatic terms, by turning off forces, check out openff-tk for examples
DLM: Make a list of those molecules with strong sterics, we may not use them in fitting interpolated parameters but Simon can make use of them while fitting LJ parameters as a benchmark set.
Another task is to provide a new FF with interpolated parameters divided into three general torsions each.
0 Comments