Bug in previous code, grid was offset from molecule causing asymmetry, this is an update with the correct grid
Slide 3 plots show RMSE in ESP compared to no virtual sites
MS/PB: how are vsites defined? Where does charge come from?
TG: 3 parent atoms, 1 N and 2 C
JW: When I think of VSites, I think ~0.5 A away from atom, but these are ~0.01 A away. Is that expected?
TG: There is a range of nearly equally preferable spots, could be further
BW – Agree, that’s almost on top of the atom. Could the charge of the atom be changed instead?
MS – For water models, the charges are moved 0.15 A from the nucleus. So 0.01 is pretty small
TG – Maybe BW was asking whether you could get the same using BCC fits?
MS – Possibly. Interesting to see how the minima cross through r=0 in some cases and not in others. That may indicate that BCCs could do the same job.
TG: Want to make VSite negative, analagous to adding (pos) charge to N, make N less negative and C a little less positive
Grid effect is much smaller now that unit conversion bug fixes grid symmetry
Scan over different charge partitions/removing charge from different atoms to see behavior (slide 5-7)
JW: Interesting that delta RMSE is similar in right 3 but q, r is so different (slide 5)
TG: Yeah, need to sit down and think about the implications of the scan
JW: did you re-evaluate Lily’s numbers with your scripts or is it possibly not an apples to apples comparison? (slide 8 )
TG: Lily calculated it, but I asked her for something that would be apples to apples
TG: Lily fit not only VSite but also all BCC’s so hard for me to evaluate that since I don’t have her BCCs
TG: Vsite definition only allows BCC on 3 atoms, used to allow on arbitrary number of atoms
JW: Is arbitrary BCC’s a bug?
TG: No, error was pretty clear that you can’t do it
JW: will look into it
JW: Oh, you’re talking about VSites, not charge increments? That’s intentional, can only change charge on tagged atom involved in VSite
TG: Charges related to VSite will likely affect charges on further away atoms
JW: Yeah…many ways to do that so I’ll leave it to Science team to decide what behavior they’d like and then y’all can tell me to implement it.