Many thanks for your videos. They are lucid and interesting. I can learn a lot. I am a bit puzzled by your description of Lennard-Jones potential as a nonpolar interaction and then an additional electrostatic interaction. I thought that the L-J potential describes induced dipole-induced dipole (van der Waals) interactions at r>r* and electron-electron repulsion at r
Indeed, this is more of a "culture of chemistry" issue than a physical one. After all, given 4 physical forces (strong nuclear, weak nuclear, electromagnetic, and gravity), there's obviously only one of those that applies to molecular chemistry. But, culturally, we tend to distinguish "steric" and "non-polar" interactions -- which are perforce electromagnetic in nature! -- as a convenient heuristic...
Is it common practice to re-optimize a force field (FF) based on only the interaction types you need? Prof makes it sound easy, maybe have a log barrier or L2-norm loss function for bond types you care about, and an L1-norm loss function for all others, and anytime you do a new simulation first re-optimize your FF. As long as the data to fit the FF is easily available? Then, I guess you wouldn't have to worry about one program, or one FF being right and another wrong, because your FF is optimized for your molecule. That is, if the FF optimization is a convex problem.
Dear Professor The lectures are in-depth and analytical. With regard to the concept of geometry optimisation what I understand is that the PES is not computed during the process. Instead the structure settles in nearby minima without going energy uphill. Is the understanding correct??
A straight geometry optimization is indeed designed only to reduce the energy (although if one takes too large a step in a downhill direction, it is certainly possible that one might overshoot the minimum and end up at a higher energy -- step size is an optimization parameter that can be controlled).
First thank you very much for these awesome videos. Secondly I have a little doubt. In the last video you mentioned that an unconstrained structure has zero energy and it is a sort of hypothetical configuration that is used to calculate the energy of a molecule with some degree of strain. My question is : Do geometry optimizations carry my guess molecule that I have drawn to that hypothetical structure? In other words, are both the hypothetical unconstrained structure and the minima in the PES the same? Thank you very much again
Typically, no, because it is impossible geometrically to arrange the atoms in space such that every bond length is simultaneously at its equilibrium distance, every angle at ITS equilibrium value, etc. Thus, the strain is INTRINSIC to you having connected the atoms in the fashion in which they are defined to be bonded. There will certainly be a minimum, but it may be quite strained, nevertheless (as is the case, for example, for cyclobutane.
Dear Professor, I am Naveena from India. I have listened to your lecture on PES and the basics of the force field. They were really in-depth. As a consequence, I felt I needed to discuss with you about refining the force field. will it be okay for you to share your email ID, so I could possibly drop all the questions I have over the same? with regards Naveena P.
Dear Professor,
Thank you for this series of videos. You have made learning these difficult topics much easier.
Always a pleasure -- thanks for letting me know you're finding the videos useful!
Many thanks for your videos. They are lucid and interesting. I can learn a lot. I am a bit puzzled by your description of Lennard-Jones potential as a nonpolar interaction and then an additional electrostatic interaction. I thought that the L-J potential describes induced dipole-induced dipole (van der Waals) interactions at r>r* and electron-electron repulsion at r
Indeed, this is more of a "culture of chemistry" issue than a physical one. After all, given 4 physical forces (strong nuclear, weak nuclear, electromagnetic, and gravity), there's obviously only one of those that applies to molecular chemistry. But, culturally, we tend to distinguish "steric" and "non-polar" interactions -- which are perforce electromagnetic in nature! -- as a convenient heuristic...
Is it common practice to re-optimize a force field (FF) based on only the interaction types you need? Prof makes it sound easy, maybe have a log barrier or L2-norm loss function for bond types you care about, and an L1-norm loss function for all others, and anytime you do a new simulation first re-optimize your FF. As long as the data to fit the FF is easily available? Then, I guess you wouldn't have to worry about one program, or one FF being right and another wrong, because your FF is optimized for your molecule. That is, if the FF optimization is a convex problem.
Dear Professor
The lectures are in-depth and analytical. With regard to the concept of geometry optimisation what I understand is that the PES is not computed during the process. Instead the structure settles in nearby minima without going energy uphill. Is the understanding correct??
A straight geometry optimization is indeed designed only to reduce the energy (although if one takes too large a step in a downhill direction, it is certainly possible that one might overshoot the minimum and end up at a higher energy -- step size is an optimization parameter that can be controlled).
First thank you very much for these awesome videos. Secondly I have a little doubt. In the last video you mentioned that an unconstrained structure has zero energy and it is a sort of hypothetical configuration that is used to calculate the energy of a molecule with some degree of strain. My question is : Do geometry optimizations carry my guess molecule that I have drawn to that hypothetical structure? In other words, are both the hypothetical unconstrained structure and the minima in the PES the same? Thank you very much again
Typically, no, because it is impossible geometrically to arrange the atoms in space such that every bond length is simultaneously at its equilibrium distance, every angle at ITS equilibrium value, etc. Thus, the strain is INTRINSIC to you having connected the atoms in the fashion in which they are defined to be bonded. There will certainly be a minimum, but it may be quite strained, nevertheless (as is the case, for example, for cyclobutane.
Dear Professor,
I am Naveena from India. I have listened to your lecture on PES and the basics of the force field. They were really in-depth. As a consequence, I felt I needed to discuss with you about refining the force field. will it be okay for you to share your email ID, so I could possibly drop all the questions I have over the same?
with regards
Naveena P.
I'm a big hot coco fan
mini-marshmallows #FTW!