Timeline for DFT Functional Selection Criteria
Current License: CC BY-SA 3.0
17 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 16, 2015 at 20:36 | comment | added | user1420303 | @Aug This scaling is VERY important because it determinate if the method can be used with larger systems/bases. It is not so much important if it take 2 or 3 hours, but for larger systems it can change from 1 week to a century. | |
| Mar 16, 2015 at 20:34 | comment | added | user1420303 | @Aug , M06 and M06-2X are hybrid functionals, they should have a scaling of N^4 with system size (the same as B3LYP), BUT, that doesn't mean that the computation time for a given system is the same for M06 and B3LYP. Roughly speaking it means that if you double the system size the computation time for one iteration will be 2^4 longer for both functionals. | |
| Mar 16, 2015 at 20:26 | comment | added | user1420303 | @LordStryker: Equilibrium B3LYP geometries for a large basis set are normally good enough to make post ab initio calculation (for example Gn and Wn methods, the later with sub chemical precision for equilibrium energies). Even if it can fail, more reliable alternatives are too expensive (In my experience, B3LYP is more reliable for geometries than MP2) | |
| Mar 16, 2015 at 18:17 | comment | added | Aug | @user1420303 Just a question about "Jacob's Ladder ". What is the computational cost of Minnesota Series ( especially M06 ) ? When I try a molecule with M06, it seem it takes forever to finish ! | |
| Mar 16, 2015 at 18:15 | comment | added | Aug | @user1420303 in Colorado library we had not this item ! I ordered it via inter-library system and should have it tomorrow. I am sure that is a great paper. | |
| Mar 16, 2015 at 14:37 | comment | added | LordStryker | @Martin Going by what I see published in something like "Journal of Medicinal Chemistry", some people think that you don't need anything other than Hartree-Fock. | |
| Mar 16, 2015 at 14:22 | comment | added | Martin - マーチン♦ | I have been a busy bee and obsessively compiled my two cents, not particularly to answer your comment, that was just a side effect.|| @LordStryker In what cases don't you need another functional ;) | |
| Mar 16, 2015 at 13:14 | comment | added | LordStryker | @user1420303 Is the study looking at minima? If so, is B3LYP yielding actual minima on the PES? Is this data reported? As a referee I would be demanding that sort of information. Is B3LYP doing a poor job describing the set of systems as minima? If so then another functional is needed. B3LYP opts in this case would not be suitable for publication... but perhaps only preliminary optimizations for more rigorous ones later. | |
| Mar 16, 2015 at 12:08 | comment | added | user1420303 | @Aug Did you take a look to the tables on the paper and in its supplementary information? There you can get objective and quantitative information of the performance of a large list of functionals in test sets designed to test a specific property. Descriptions based on design can be not reliable when you test it in real world. I suggest, for example, see the tables of the M-12 functionals paper. | |
| Mar 16, 2015 at 11:48 | comment | added | user1420303 | @Martin, Although we may complain, I think that most referee won't. Would you also complain about the B3LYP usage for geometry optimizations? If so, What functional do you suggest for that purpose? | |
| Mar 16, 2015 at 6:22 | history | edited | Martin - マーチン♦ | CC BY-SA 3.0 |
Just noticed a small spelling error in the name of the cited paper.
|
| Mar 16, 2015 at 6:06 | comment | added | Martin - マーチン♦ | I agree with what you said, except the part with the B3LYP functional, I would always complain about that, but that is because the magical functional is also the lazy functional. | |
| Mar 16, 2015 at 5:57 | history | edited | Martin - マーチン♦ | CC BY-SA 3.0 |
inserted link to publication for easier access.
|
| Mar 16, 2015 at 5:33 | comment | added | Aug | This answer really helps but I am looking for something more structured and practical ( as what is in here: en.wikipedia.org/wiki/Minnesota_Functionals ) that's why I decided to put a bounty . | |
| Mar 14, 2015 at 18:11 | history | edited | user1420303 | CC BY-SA 3.0 |
added 62 characters in body
|
| Mar 14, 2015 at 18:01 | comment | added | Geoff Hutchison | +10. Excellent response. I think the one caveat I'd add is for "accuracy" since different functionals are designed for different properties. So if you want an accurate functional for intermolecular interactions, it might be different than one for transition state barriers. | |
| Mar 14, 2015 at 17:44 | history | answered | user1420303 | CC BY-SA 3.0 |