I work mostly with glasses, and starting initial configuration randomly leads to the highly non-equilibrium description of the system. In most of the cases BFGS have hard time converging the geometry, so I run FPMD at high temperature with slightly lower convergence criteria and wait until energies converge. Doing so generally solves my problem.
Hello, thank you for the videos. May I ask; The issue getting the wrong electrons (in a given irrep) does that also apply if you run without symmetry (or C1 lets say), is it a good way to double check ? Or am I missing something. I've enjoyed PSI4 tremendously in the last year to check my answers. I have now written my measly ab -initio package as a result. Still going months later (after a 30 year hiatus ). Rusty at the start, but things are beginning to flow once again, thanks to COVID, stuck at home, found something to do :D How missed it. Now I am hooked on electronic structure theory. My background was in spectroscopy and rotation vibration theory originally (with Prof Ian Mills OBE) you may well have met once upon a time. Now i understand why not many write their own integral code, the second derivatives were tough, but the most rewarding (analytically). Not as fast as PSI4, but respectable ( I think MY NVIDIA card may come in handy to through some CUDA at it one day) More cats please :)
Yes, good question... if there's a problem with a calculation because the wrong number of electrons is assigned to each irrep, it *may* help to run the computation in C1. However, there's not a guarantee that this will fix it, because even if the program isn't told to utilize the symmetry, the underlying symmetry of the molecule is still there in the computation. You can try to deliberately break the symmetry of the molecule as a way to get around this (some programs have a keyword to mix up the guess orbitals), but then there is a small chance that you might land on what we call an "artifactual symmetry-broken solution" (i.e., the wavefunction fails to eventually recover the correct symmetry of the molecule, and when this happens, all sorts of strange results are possible)
I work mostly with glasses, and starting initial configuration randomly leads to the highly non-equilibrium description of the system. In most of the cases BFGS have hard time converging the geometry, so I run FPMD at high temperature with slightly lower convergence criteria and wait until energies converge. Doing so generally solves my problem.
Respected sir highly informative veiod
Really helpful
Thank you this is amaIng
It is true for what you said about Vd W complexes. Please add more videos.
Thank you very much, a very interesting topic.
Thank you, Dr. Sherrill, this is helpful
Hello, thank you for the videos. May I ask; The issue getting the wrong electrons (in a given irrep) does that also apply if you run without symmetry (or C1 lets say), is it a good way to double check ? Or am I missing something.
I've enjoyed PSI4 tremendously in the last year to check my answers. I have now written my measly ab -initio package as a result. Still going months later (after a 30 year hiatus ).
Rusty at the start, but things are beginning to flow once again, thanks to COVID, stuck at home, found something to do :D
How missed it. Now I am hooked on electronic structure theory. My background was in spectroscopy and rotation vibration theory originally (with Prof Ian Mills OBE) you may well have met once upon a time.
Now i understand why not many write their own integral code, the second derivatives were tough, but the most rewarding (analytically).
Not as fast as PSI4, but respectable ( I think MY NVIDIA card may come in handy to through some CUDA at it one day)
More cats please :)
Yes, good question... if there's a problem with a calculation because the wrong number of electrons is assigned to each irrep, it *may* help to run the computation in C1. However, there's not a guarantee that this will fix it, because even if the program isn't told to utilize the symmetry, the underlying symmetry of the molecule is still there in the computation. You can try to deliberately break the symmetry of the molecule as a way to get around this (some programs have a keyword to mix up the guess orbitals), but then there is a small chance that you might land on what we call an "artifactual symmetry-broken solution" (i.e., the wavefunction fails to eventually recover the correct symmetry of the molecule, and when this happens, all sorts of strange results are possible)
@@DavidSherrill1 Thank you, that makes sense ... thinking about it some more.
There's some evidence that diffuse functions are not (always) necessary for anions: doi: 10.1021/jp312755z