One of the main "feature" of Kohn-Sham DFT (KS-DFT) is the approximation of the exact exchange-correlation functional, leading to a zoo of functionals with many flaws. One such flaw is the self-interaction error (SIE), i.e. an electron can interact with itself! Reducing SIE is one of the goal when "designing" a new modern functional (by… Continue reading Minnesota MN15 functional (and the others) – SIE!
It's sunny today, so I decided to perform just a small test (and spend the rest of my day staying outside) again with MN15, SCAN0, and ωB97M-V. The molecule is small, Fe(CO)4, inspired from the work of Harvey et al. (here and here). In these works, the authors studied the singlet-triplet gap ΔE = E(LS)… Continue reading Minnesota MN15 functional (and the others) – part 3
Thanks to the author of Erkale: The bug with SCAN0 is fixed. My calculation with Erkale is now faster. I should have included the computational details with Erkale: e.g. DFTGrid(75,302), or default Cholesky decomposition for the two-electron integrals (produce an error of ~1 kcal/mol). So here, this is a small update of my previous post (remember I… Continue reading Minnesota MN15 functional (and the others) – an update
Let's try to answer a question. How faster a calculation can be by exploiting molecular symmetry? Suppose we want to do a CCSD(T) calculation of benzene C6H6. The symmetry point group is D6h but for correlated methods, we can only employ at most D2h (the highest Abelian group). Then, how will the calculation be slower… Continue reading Symmetry, or not symmetry, that is the question
See a small update here. Last year I was at the 8th MQM conference and attended a talk of D. Truhlar advertising his latest Minnesota functional, the MN15 functional. According to his talk, this functional is good at describing everything, including multireference problems. I was quite skeptical of that. Last week, one of my colleagues… Continue reading Minnesota MN15 functional (and the others)