From owner-chemistry@ccl.net Wed Jun 13 08:43:00 2018 From: "Mikael Johansson mikael.johansson*iki.fi" To: CCL Subject: CCL: Complete basis set extrapolation, but with DFT methods Message-Id: <-53353-180613012453-14880-Pta1e2IITl73tnqPI1HRBg=-=server.ccl.net> X-Original-From: Mikael Johansson Content-Language: en-US Content-Type: multipart/alternative; boundary="------------308D9368F8C6AAF107D2F45C" Date: Wed, 13 Jun 2018 08:24:38 +0300 MIME-Version: 1.0 Sent to CCL by: Mikael Johansson [mikael.johansson|*|iki.fi] This is a multi-part message in MIME format. --------------308D9368F8C6AAF107D2F45C Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 8bit Hello hello, Just a quick comment on extrapolating wf correlation energies: In general, DZ energies have such larger errors that including them in the extrapolation worsens the results. Thus, two-point [T,Q]Z schemes are preferred. See//for example Halkier /et al/, "Basis-set convergence in correlated calculations on Ne, N2, and H2O", https://doi.org/10.1016/S0009-2614(98)00111-0 As briefly already mentioned, what kind of scheme to use for mixes of "standard" GGA-DFT and wf beyond Hartree-Fock, like in double hybrids and RPA, is perhaps not so straightforward. Cheers, Mikael On 13.6.2018 06:33, Guzman, Francisco fg58(a)njit.edu wrote: > and extrapolated single point CCSD(T)/AV[D,T,Q]Z energies. > -- Dr. Mikael Johansson Academy Researcher Department of Chemistry University of Helsinki http://blogs.helsinki.fi/johansson-group --------------308D9368F8C6AAF107D2F45C Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: 8bit
Hello hello,

Just a quick comment on extrapolating wf correlation energies: In general, DZ energies have such larger errors that including them in the extrapolation worsens the results. Thus, two-point [T,Q]Z schemes are preferred. See for example Halkier et al, "Basis-set convergence in correlated calculations on Ne, N2, and H2O", https://doi.org/10.1016/S0009-2614(98)00111-0

As briefly already mentioned, what kind of scheme to use for mixes of "standard" GGA-DFT and wf beyond Hartree-Fock, like in double hybrids and RPA, is perhaps not so straightforward.

Cheers,
Mikael

On 13.6.2018 06:33, Guzman, Francisco fg58(a)njit.edu wrote:

and extrapolated single point CCSD(T)/AV[D,T,Q]Z energies.

-- 
Dr. Mikael Johansson
Academy Researcher
Department of Chemistry
University of Helsinki
http://blogs.helsinki.fi/johansson-group

--------------308D9368F8C6AAF107D2F45C-- From owner-chemistry@ccl.net Wed Jun 13 11:33:00 2018 From: "Susi Lehtola susi.lehtola]~[alumni.helsinki.fi" To: CCL Subject: CCL: Complete basis set extrapolation, but with DFT methods Message-Id: <-53354-180613113023-10161-4TrzJLB8RWfRgH6qPKwOpQ+/-server.ccl.net> X-Original-From: Susi Lehtola Content-Language: en-US Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=utf-8; format=flowed Date: Wed, 13 Jun 2018 18:30:12 +0300 MIME-Version: 1.0 Sent to CCL by: Susi Lehtola [susi.lehtola]|[alumni.helsinki.fi] On 06/13/2018 06:33 AM, Guzman, Francisco fg58(a)njit.edu wrote: > Thank you all for the responses and fruitful discussion. To give more of > a background, in our group we study heavy molecules so we are limited in > the basis sets available (No Jensen :( unfortunately). The Karlsruhe def2 sets are also very good and come with ECPs for heavy atoms which make them suitable for the majority of calculations. > From experience the CCSD(T)/AVnZ calculations are very good, but what > surprising was the behavior of the DFT/AVTZ method. For a given work > reaction scheme and target molecule, sometimes the the DFT/AVTZ > calculation would be within 0.5 kcal/mol of the extrapolated > CCSD(T)/AVnZ calculation, and sometimes the difference was as large as > 10 kcal/mol. > > In essence, the work reaction scheme allows one to save on CPU time and > allow the study of a diverse set of molecules. Achieving accurate and > reliable thermo with only DFT methods would be an ideal scenario. This sounds more like method error than basis set error. That is, the DFT/AVTZ result is probably still sufficiently close to the basis set limit, only the DFT functional (which you didn't specify) fails to give you the correct energetics. -- ------------------------------------------------------------------ Mr. Susi Lehtola, PhD Junior Fellow, Adjunct Professor susi.lehtola(~)alumni.helsinki.fi University of Helsinki http://www.helsinki.fi/~jzlehtol Finland ------------------------------------------------------------------ Susi Lehtola, dosentti, FT tutkijatohtori susi.lehtola(~)alumni.helsinki.fi Helsingin yliopisto http://www.helsinki.fi/~jzlehtol ------------------------------------------------------------------