From owner-chemistry@ccl.net Mon Jun 1 07:51:01 2015 From: "Susi Lehtola susi.lehtola[#]alumni.helsinki.fi" To: CCL Subject: CCL: Measuring Instantaneous Correlation of Individual Orbitals Message-Id: <-51416-150531225844-15948-4VSIo1fF277FLEsR2INiTg() server.ccl.net> X-Original-From: Susi Lehtola Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=utf-8; format=flowed Date: Sun, 31 May 2015 19:58:30 -0700 MIME-Version: 1.0 Sent to CCL by: Susi Lehtola [susi.lehtola[#]alumni.helsinki.fi] On 05/31/2015 08:16 AM, Billy McCann thebillywayne-,-gmail.com wrote: > I'd like to, for now, leave aside density functional theory because I > don't have much experience or insight into the nature of the > exchange-correlation operators; I can't seem to get a systematic > understanding of that particular operator in its various formulations. > And it's this correlation energy which I'm curious about. That the > operator contains both exchange, correlation, plus a correction to the > kinetic energies of the Kohn-Sham orbitals confounds me even more when > trying to understand it, not even mentioning double-hybrid DFA's. I > know that brilliant scientists have worked on various density > functional approximations, and I do not at all want to belittle their > work. DFA is a great tools for physicists and chemists. There is no "exchange-correlation operator"; exchange and correlation just happen to be names that we call certain types of matrix elements of the Coulomb operator. Now, DFT is nothing more exotic than the realization that one can exactly write the Schrödinger equation as something that only depends on the density, which yields a huge reduction in the degrees of freedom. The problem is we don't know the exact functional, and there's no way to approach the exact solution systematically. There's only a very rough partitioning of the approximate functionals onto different levels of theory (Jacob's ladder)... and going up in the ladder, for instance from GGA functionals to hybrid GGA functionals, may give you much worse results. > If I understand correctly, all methods begin from the HF approximation and correct for > dynamical correlation by making a linear combination of Slater > determinants by different methods. Yes, except for many-component wave functions like in MCSCF, where the orbitals are optimized including some extent of (strong) correlation. > But what I'm wondering about is the correlation energy of a *single* > atomic or molecular orbital. Is it that comparing the HF orbital > energy to, say, a corresponding orbital resulting from a CCSD(T) > calculation would yield such an energy? I've pondered this question, > but I've read others who say that this isn't entirely the case because > HF does indeed account for some small degree of electron correlation, > but only in an averaged way. (I think I remember reading this in > Cramer's text.) Perhaps MC-SCF may provide such an answer, by > measuring the coefficients of each determinant? Atomic or molecular orbitals aren't observables, so that notion doesn't make sense. (Also, as already pointed out by Robert Molt, the notion of correlation is also a bit ill defined. Richard Feynman said that instead of correlation energy one should really talk about dumbness energy, since we talk of correlation only because we're too dumb to solve the many-electron Schrödinger equation directly.) > So my question is two-fold: > > 1. How can the dynamical electron correlation energy of a single > atomic or molecular orbital be measured? Can it even be done? You can't measure it. But in calculations you *can* define it by calculating the difference E(full) - E(frozen orbital). This is, of course, often used because the frozen core approximation is popular in post-HF calculations. > 2. Is it possible to make a generalized statement such as, "Core > electrons experience a greater degree of correlation because they are > surrounded by more electrons," or "Valence electrons experience a > greater degree of electron correlation because they are bound more > loosely to the system, allowing their wavefunctions to fluctuate more > freely,"? You'd need to separate strong and dynamic correlation. Strong correlation only affects valence orbitals. For dynamic correlation IIRC both core and valence orbitals usually have similar correlation energies, which is not what one might expect. Namely, for core orbitals one can make the argument you gave above, that the correlation energy should be larger than for valence electrons, because the electron density is much larger in the core region. However, it is also true that the velocity of core orbitals is much higher than of valence orbitals, so the electrons spend much less time in close contact. -- ----------------------------------------------------------------------- Mr. Susi Lehtola, PhD Chemist Postdoctoral Fellow susi.lehtola\a/alumni.helsinki.fi Lawrence Berkeley National Laboratory http://www.helsinki.fi/~jzlehtol USA ----------------------------------------------------------------------- From owner-chemistry@ccl.net Mon Jun 1 08:26:01 2015 From: "Jan H Jensen jhjensen++chem.ku.dk" To: CCL Subject: CCL: Computational Chemistry Highlights: May issue Message-Id: <-51417-150601013053-25400-8xAWJwsEBvsuSWVyvbANRQ]-[server.ccl.net> X-Original-From: "Jan H Jensen" Date: Mon, 1 Jun 2015 01:30:52 -0400 Sent to CCL by: "Jan H Jensen" [jhjensen(0)chem.ku.dk] The May issue of Computational Chemistry Highlights is out. http://www.compchemhighlights.org/2015_05_01_archive.html CCH is an overlay journal that identifies the most important papers in computational and theoretical chemistry published in the last 1-2 years. CCH is not affiliated with any publisher: it is a free resource run by scientists for scientists. Table of content for this issue features contributions from CCH editors Steven Bachrach and Jan Jensen: Exploring the Accuracy Limits of Local Pair Natural Orbital Coupled-Cluster Theory http://www.compchemhighlights.org/2015/05/exploring-accuracy-limits-of-local-pair.html Using dispersion-corrected density functional theory to understand supramolecular binding thermodynamics http://x.vindicosuite.com/click/fbfpc=1;v=5;m=3;l=401071;c=776283;b=3368032;dct=http%3A/ /www.compchemhighlights.org/2015/05/using-dispersion-corrected-density.html Uthrene, a radically new molecule? http://www.compchemhighlights.org/2015/05/uthrene-radically-new-molecule.html Interested in more? There are many ways to subscribe to CCH updates. http://www.compchemhighlights.org/p/get-cch-updates.html Also, for your daily computational chemistry fix subscribe to Computational Chemistry Daily https://paper.li/janhjensen/1416314690 From owner-chemistry@ccl.net Mon Jun 1 09:01:01 2015 From: "Tian Lu sobereva^^^sina.com" To: CCL Subject: CCL:G: transition density file generation Message-Id: <-51418-150601062015-27837-PTndPmApLonIYdcypqJStw###server.ccl.net> X-Original-From: "Tian Lu" Date: Mon, 1 Jun 2015 06:20:11 -0400 Sent to CCL by: "Tian Lu" [sobereva a sina.com] Dear Xing, AFAIK, Gaussian itself is unable to generate cube file for transition density. However, you can use Multiwfn (multiwfn.codeplex.com) to obtain it and directly visualize corresponding isosurface based on Gaussian TD or CIS calculation data. An example is given in Section 4.18.1 of the Multiwfn manual (this example also extensively discussed how to analyze charge transfer during electron excitation) -------------------------------- Best wishes, Tian Lu Beijing Kein Research Center for Natural Sciences, Beijing, P. R. China website: www.keinsci.com ----- Original Message ----- > From: "Xing Yin xiy726_-_gmail.com" To: "Lu, Tian " Subject: CCL:G: transition density file generation Date: 2015-06-01 02:55 Sent to CCL by: Xing Yin [xiy726[-]gmail.com] Hi Abbey, I was doing some related research. Just curious how did you make Gaussian generate the transition density files? The "cubegen" in G09 won't work with any density=transition keywords, If I simply use the "cube" keyword in g09, what I obtained is nothing but total SCF density. The output of density=scf and density=transition=1 is exactly the same. Thanks! On 5/30/2015 8:54 AM, Abbey Meprathu Philip abbeyphilip88 . gmail.com wrote: > Sent to CCL by: "Abbey Meprathu Philip" [abbeyphilip88[-]gmail.com] > Dear CCL members, > I am trying to calculate the transition density of a molecule. Using > Gaussian I am trying to generate the .cube file which could be used for > further calculation. I have input the co-ordinates of the molecule from the > crystal structure of the molecule and am using the following command to > generate the .cube file. > > %chk=NI-absorption_cube.chk > %mem=1GB > # td=(singlets,nstates=12) b3lyp/6-311++g(2d,2p) scrf= > (solvent=acetonitrile,pcm) geom=connectivity > scf=(convergence=6,maxcycle=512) > > As I am new to this type of calculation I am not much aware of the > commands to use. I want to calculate the transition density from the solid- > state crystal co-ordinates and don't want to use the solvent model. So what > command should I be using in order to calculate the transition density in > the solid state? Should I leave the scrf= (solvent= ) blank? > > Thank and regards, > Abbey> > -- Best Wishes, Xing From owner-chemistry@ccl.net Mon Jun 1 19:30:00 2015 From: "Sergio Emanuel Galembeck segalemb]|[usp.br" To: CCL Subject: CCL: ADF Computational Chemistry Workshop - USP/RP Message-Id: <-51419-150601191914-21049-P0GL1OO8xxROfCAbCpILZA||server.ccl.net> X-Original-From: "Sergio Emanuel Galembeck" Date: Mon, 1 Jun 2015 19:19:13 -0400 Sent to CCL by: "Sergio Emanuel Galembeck" [segalemb!A!usp.br] > From November 16 to 20, 2015, the Department of Chemistry of Ribeiro Preto campus of University of So Paulo, Brazil, will host a hands-on ADF Computational Chemistry Workshop, for 25 participants. During the mornings, Prof. Dr. F. Matthias Bickelhaupt, FRSC, and Dr. Celia Fonseca Guerra, will give lectures about some aspects of electronic structure methods, and ways to obtain chemical information from those methods. In the afternoons, hands-on exercises using ADF software will explore different possibilities of the use of this software. All interested advanced undergraduate, graduate students, pos-doc, faculty or some people that use, or are planning to use electronic structure calculations are invited to do the inscription for this event, from June 1 to 30. Each participant needs to bring your own notebook. It is not necessary to have a previous experience on ADF software. More information could be find in: http://3qc.iqm.unicamp.br/ADFCCW The Organizing Committee, Nelson Henrique Morgon (IQ-UNICAMP) Ricardo Vessecchi (FFCLRP-USP) Srgio Emanuel Galembeck, (FFCLRP-USP) (coord.)