From owner-chemistry@ccl.net Tue Oct 18 03:00:00 2016 From: "Marcel Swart marcel.swart-#-icrea.cat" To: CCL Subject: CCL: CCL #DFT16poll results are out Message-Id: <-52405-161018025904-6694-zulzl8MTDuZti79HHAeZdw(0)server.ccl.net> X-Original-From: "Marcel Swart" Date: Tue, 18 Oct 2016 02:59:02 -0400 Sent to CCL by: "Marcel Swart" [marcel.swart---icrea.cat] >> On 17 Oct 2016, at 20:28 , Susi Lehtola susi.lehtola ~ alumni.helsinki.fi >> wrote: >>>> Sent to CCL by: "Jim Kress" [jimkress35%x%gmail.com] >>>> Marcel, Matthias, and Miquel; >>>> Great work as usual. Thanks for doing this. It is immensely helpful. Thanks Jim. >> I would tend to disagree. What the poll tells you is a very limited picture. Democracy has no say on >> what is the truth... Especially when you can't vote for what you'd like to, but are presented with a >> choice between what are essentially bad and worse functionals. [..] >> What represents the state of the art, for instance, the wB97X-V, B97M-V and wB97M-V functionals, >> have not been included in the poll, even though they beat the crap out of basically anything on the >> list. Thanks for the feedback Susi. I gather this is your way of saying that you would like these functionals to be included in next year's edition? Note that as stated on the website, rules and news-item the normal procedure is to send me a mail with your suggestions (as indeed people have done in the past). Note that there are "only" 10 slots available for new functionals, of which two are already taken for next year. With these three suggestions it goes to five. >> The poll seems to be centered on getting a picture of the functionals that people use. You could >> get a much better picture just by tabulating the number of citations to the articles describing these >> functionals. Finally, I would like to remind the blog entry on Nature Chemistry's The Sceptical Chymist: http://blogs.nature.com/thescepticalchymist/2014/11/five-years-of-polling-the-computational- chemistry-community.html where many arguments pro and contra have been put forward. To quote one of the godfathers of DFT: The DFT popularity poll is somewhat like citation analysis: It measures (but in a different way) how well a functional has been received by a set of readers and users. There are many reasons why some functionals are received better than others: accuracy, reliability, wide applicability, computational efficiency, well-founded construction, availability in standard codes, reputation of the functional and its authors, historical priority, novelty, and even hype. The poll has to be seen as measuring all these things, and perhaps more. To the extent that the polled scientists use rational criteria, the results of the poll can point other scientists toward good or interesting functionals (John Perdew, 2014) From owner-chemistry@ccl.net Tue Oct 18 09:14:00 2016 From: "Brian Skinn bskinn!A!alum.mit.edu" To: CCL Subject: CCL: CCL #DFT16poll results are out Message-Id: <-52406-161018091306-19097-5s0h/w8IAthsSxOVBjp/GQ()server.ccl.net> X-Original-From: Brian Skinn Content-Type: multipart/alternative; boundary=001a113952226f6532053f2373ac Date: Tue, 18 Oct 2016 09:12:36 -0400 MIME-Version: 1.0 Sent to CCL by: Brian Skinn [bskinn(!)alum.mit.edu] --001a113952226f6532053f2373ac Content-Type: text/plain; charset=UTF-8 Susi, >> You could >> get a much better picture just by tabulating the number of citations to the articles describing these >> functionals. I disagree with your assertion that this would provide a better picture. What fraction of the time do articles that use a given functional *actually cite* (and cite correctly!) the original method paper for that functional? Especially B3LYP -- the method is so standard for 'casual computation' that I would not be surprised if it is not cited more frequently than it is! So, yes, as Dr. Perdew says in Dr. Swart's quote, a citation analysis would give a different picture, providing different information -- but it would in no substantial sense be "more authoritative" or "higher quality" as a representation of usage patterns than is DFTpoll. Best regards, Brian On Tue, Oct 18, 2016 at 2:59 AM, Marcel Swart marcel.swart-#-icrea.cat < owner-chemistry(~)ccl.net> wrote: > > Sent to CCL by: "Marcel Swart" [marcel.swart---icrea.cat] > >> On 17 Oct 2016, at 20:28 , Susi Lehtola susi.lehtola ~ > alumni.helsinki.fi > >> wrote: > > >>>> Sent to CCL by: "Jim Kress" [jimkress35%x%gmail.com] > >>>> Marcel, Matthias, and Miquel; > > >>>> Great work as usual. Thanks for doing this. It is immensely helpful. > > Thanks Jim. > > >> I would tend to disagree. What the poll tells you is a very limited > picture. Democracy has no say on > >> what is the truth... Especially when you can't vote for what you'd like > to, but are presented with a > >> choice between what are essentially bad and worse functionals. > [..] > >> What represents the state of the art, for instance, the wB97X-V, B97M-V > and wB97M-V > functionals, > >> have not been included in the poll, even though they beat the crap out > of basically anything on the > >> list. > > Thanks for the feedback Susi. I gather this is your way of saying that you > would like these functionals > to be included in next year's edition? Note that as stated on the website, > rules and news-item the > normal procedure is to send me a mail with your suggestions (as indeed > people have done in the past). > Note that there are "only" 10 slots available for new functionals, of > which two are already taken for > next year. With these three suggestions it goes to five. > > >> The poll seems to be centered on getting a picture of the functionals > that people use. You could > >> get a much better picture just by tabulating the number of citations to > the articles describing these > >> functionals. > > Finally, I would like to remind the blog entry on Nature Chemistry's The > Sceptical Chymist: > http://blogs.nature.com/thescepticalchymist/2014/11/ > five-years-of-polling-the-computational- > chemistry-community.html > where many arguments pro and contra have been put forward. To quote one of > the godfathers of DFT: > > The DFT popularity poll is somewhat like citation analysis: It measures > (but in a different way) how > well a functional has been received by a set of readers and users. There > are many reasons why some > functionals are received better than others: accuracy, reliability, wide > applicability, computational > efficiency, well-founded construction, availability in standard codes, > reputation of the functional and > its authors, historical priority, novelty, and even hype. The poll has to > be seen as measuring all these > things, and perhaps more. To the extent that the polled scientists use > rational criteria, the results of > the poll can point other scientists toward good or interesting functionals > (John Perdew, 2014)> > > --001a113952226f6532053f2373ac Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable
S= usi,
<= span style=3D"color:rgb(80,0,80);font-size:12.8px">
>>= ; You could
>> get a much better pic= ture just by tabulating the number of citations to the articles describing = these
>> functionals.

I disagree with your = assertion that this would provide a better picture. What fraction of the ti= me do articles that use a given functional *actually cite* (and cite correc= tly!) the original method paper for that functional? Especially B3LYP -- th= e method is so standard for 'casual computation' that I would not b= e surprised if it is not cited more frequently than it is!
= So, yes, as Dr. Perdew = says in Dr. Swart's quote, a citation analysis would give a different p= icture, providing different information -- but it would in no substantial s= ense be "more authoritative" or "higher quality" as a r= epresentation of usage patterns than is DFTpoll.


Best regards,
Brian

On Tue, Oct 18, 201= 6 at 2:59 AM, Marcel Swart marcel.swart-#-icre= a.cat <owner-chemistry(~)ccl.net> wrote:

Sent to CCL by: "Marcel=C2=A0 Swart" [marcel.swart---icrea.= cat]
>> On 17 Oct 2016, at 20:28 , Susi Lehtola susi.lehtola ~ alumni.hels= inki.fi <owner-chemistry|ccl.net>
>> wrote:

>>>> Sent to CCL by: "Jim Kress" [jimkress35%x%gmail.com]<= br> >>>> Marcel, Matthias, and Miquel;

>>>> Great work as usual.=C2=A0 Thanks for doing this.=C2=A0 It= is immensely helpful.

Thanks Jim.

>> I would tend to disagree. What the poll tells you is a very limite= d picture. Democracy has no say on
>> what is the truth... Especially when you can't vote for what y= ou'd like to, but are presented with a
>> choice between what are essentially bad and worse functionals.
[..]
>> What represents the state of the art, for instanc= e, the wB97X-V, B97M-V and wB97M-V
functionals,
>> have not been included in the poll, even though they beat the crap= out of basically anything on the
>>=C2=A0 list.

Thanks for the feedback Susi. I gather this is your way of saying th= at you would like these functionals
to be included in next year's edition? Note that as stated on the websi= te, rules and news-item the
normal procedure is to send me a mail with your suggestions (as indeed peop= le have done in the past).
Note that there are "only" 10 slots available for new functionals= , of which two are already taken for
next year. With these three suggestions it goes to five.

>> The poll seems to be centered on getting a picture of the function= als that people use. You could
>> get a much better picture just by tabulating the number of citatio= ns to the articles describing these
>> functionals.

Finally, I would like to remind the blog entry on Nature Chemistry&#= 39;s The Sceptical Chymist:
http://blogs= .nature.com/thescepticalchymist/2014/11/five-years-of-polling-the= -computational-
chemistry-community.html
where many arguments pro and contra have been put forward. To quote one of = the godfathers of DFT:

The DFT popularity poll is somewhat like citation analysis: It measures (bu= t in a different way) how
well a functional has been received by a set of readers and users.=C2=A0 Th= ere are many reasons why some
functionals are received better than others: accuracy, reliability, wide ap= plicability, computational
efficiency, well-founded construction, availability in standard codes, repu= tation of the functional and
its authors, historical priority, novelty, and even hype.=C2=A0 The poll ha= s to be seen as measuring all these
things, and perhaps more. To the extent that the polled scientists use rati= onal criteria, the results of
the poll can point other scientists toward good or interesting functionals<= br> (John Perdew, 2014)



-=3D This is automatically added to each message by the mailing script =3D-=
E-mail to subscribers: CHEMISTRY(~)ccl.n= et or use:
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/cgi-bin/= ccl/send_ccl_message

E-mail to administrators: CHEM= ISTRY-REQUEST(~)ccl.net or use
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/cgi-bin/= ccl/send_ccl_message

Subscribe/Unsubscribe:
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/chemistry/s= ub_unsub.shtml

Before posting, check wait time at: http://www.ccl.net

Job: http://www.ccl.net/jobs
Conferences: http://server.ccl.net/ch= emistry/announcements/conferences/

Search Messages: http://www.ccl.net/chemistry/searchccl/index.shtml
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutcc= l/instructions/



--001a113952226f6532053f2373ac-- From owner-chemistry@ccl.net Tue Oct 18 10:11:00 2016 From: "Igors Mihailovs igorsm|,|cfi.lu.lv" To: CCL Subject: CCL: CCL #DFT16poll results are out Message-Id: <-52407-161018100920-17879-CrBb7T2Hj1q2gv+hhBcqgg^server.ccl.net> X-Original-From: Igors Mihailovs Content-Type: multipart/alternative; boundary="------------A289B4F97ADF07D17E2CA49C" Date: Tue, 18 Oct 2016 17:09:09 +0300 MIME-Version: 1.0 Sent to CCL by: Igors Mihailovs [igorsm^cfi.lu.lv] This is a multi-part message in MIME format. --------------A289B4F97ADF07D17E2CA49C Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit Yes, but if these indeed highly renowned experts discover from the poll than many people are using creepy functionals, they can possibly write another review in easy language (as they occasionally do, deserving even more respect) to explain those many people that there are good alternatives with small additional computational cost. Even better if some of these reviews are published open-access to reach the researchers in less prosperous countries. So I would like to thank the organizers of this poll! On 18.10.2016 01:54, Dr. Robert Molt Jr. r.molt.chemical.physics(!)gmail.com wrote: > > Sent to CCL by: "Dr. Robert Molt Jr." > [r.molt.chemical.physics!A!gmail.com] > I 100% agree with with Dr. Lehtola; no useful information can be > extracted from this poll. Science has nothing to do with democracy. I > am interested in reading review papers from experts, like Grimme, > Truhlar, Head-Gordon, Gordon, Perdew, Burke, Yang, etc. The evidence > in such review papers should guide action. We have reviewers for > journal articles because popularity is not a substitute for evidence. > > I don't revere Maxwell, Boltzmann, Einstein, etc. because they won > high school popularity contests. > > > On 10/17/16 2:28 PM, Susi Lehtola susi.lehtola ~ alumni.helsinki.fi > wrote: >> >> Sent to CCL by: Susi Lehtola [susi.lehtola~!~alumni.helsinki.fi] >> On 10/17/2016 07:34 AM, Jim Kress jimkress35+/-gmail.com wrote: >>> >>> Sent to CCL by: "Jim Kress" [jimkress35%x%gmail.com] >>> Marcel, Matthias, and Miquel; >>> >>> Great work as usual. Thanks for doing this. It is immensely helpful. >> >> I would tend to disagree. What the poll tells you is a very limited >> picture. Democracy has no say on what is the truth... Especially when >> you can't vote for what you'd like to, but are presented with a >> choice between what are essentially bad and worse functionals. >> >> The poll is on the popularity - *not* the accuracy - of OLD >> functionals, which is furthermore limited in its participants. For >> instance, the first division participants were >> >> LDA 1965 >> BP86 1988 >> PW91 1991 >> B3PW91 1993 >> B3LYP 1994 >> PBE 1996 (PBE0 1999) >> B97-D 1997 >> revPBE 1999 >> RPBE 1999 >> HSE 2003 >> CAM-B3LYP 2004 >> B2PLYP 2006 >> LC-wPBE 2006 >> B3LYP-D 2007? >> optB88-vdW 2007 >> wB97X-D 2008 >> M06-2X 2008 >> M06 2008 >> PWPB95-D3 2010 >> >> which I would say is really old. 13 out of 19 functionals are over 10 >> years old. What represents the state of the art, for instance, the >> wB97X-V, B97M-V and wB97M-V functionals, have not been included in >> the poll, even though they beat the crap out of basically anything on >> the list. >> >> The poll seems to be centered on getting a picture of the functionals >> that people use. You could get a much better picture just by >> tabulating the number of citations to the articles describing these >> functionals. > -- Ar cieņu, Igors Mihailovs inženieris/doktorants LU CFI Organisko materiālu laboratorija --------------A289B4F97ADF07D17E2CA49C Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: 8bit

Yes, but if these indeed highly renowned experts discover from the poll than many people are using creepy functionals, they can possibly write another review in easy language (as they occasionally do, deserving even more respect) to explain those many people that there are good alternatives with small additional computational cost. Even better if some of these reviews are published open-access to reach the researchers in less prosperous countries.

So I would like to thank the organizers of this poll!


On 18.10.2016 01:54, Dr. Robert Molt Jr. r.molt.chemical.physics(!)gmail.com wrote:

Sent to CCL by: "Dr. Robert Molt Jr." [r.molt.chemical.physics!A!gmail.com]
I 100% agree with with Dr. Lehtola; no useful information can be extracted from this poll. Science has nothing to do with democracy. I am interested in reading review papers from experts, like Grimme, Truhlar, Head-Gordon, Gordon, Perdew, Burke, Yang, etc. The evidence in such review papers should guide action. We have reviewers for journal articles because popularity is not a substitute for evidence.

I don't revere Maxwell, Boltzmann, Einstein, etc. because they won high school popularity contests.


On 10/17/16 2:28 PM, Susi Lehtola susi.lehtola ~ alumni.helsinki.fi wrote:

Sent to CCL by: Susi Lehtola [susi.lehtola~!~alumni.helsinki.fi]
On 10/17/2016 07:34 AM, Jim Kress jimkress35+/-gmail.com wrote:

Sent to CCL by: "Jim Kress" [jimkress35%x%gmail.com]
Marcel, Matthias, and Miquel;

Great work as usual.  Thanks for doing this.  It is immensely helpful.

I would tend to disagree. What the poll tells you is a very limited picture. Democracy has no say on what is the truth... Especially when you can't vote for what you'd like to, but are presented with a choice between what are essentially bad and worse functionals.

The poll is on the popularity - *not* the accuracy - of OLD functionals, which is furthermore limited in its participants. For instance, the first division participants were

LDA        1965
BP86        1988
PW91        1991
B3PW91        1993
B3LYP        1994
PBE        1996 (PBE0 1999)
B97-D        1997
revPBE        1999
RPBE        1999
HSE        2003
CAM-B3LYP    2004
B2PLYP        2006
LC-wPBE        2006
B3LYP-D        2007?
optB88-vdW    2007
wB97X-D        2008
M06-2X        2008
M06        2008
PWPB95-D3    2010

which I would say is really old. 13 out of 19 functionals are over 10 years old. What represents the state of the art, for instance, the wB97X-V, B97M-V and wB97M-V functionals, have not been included in the poll, even though they beat the crap out of basically anything on the list.

The poll seems to be centered on getting a picture of the functionals that people use. You could get a much better picture just by tabulating the number of citations to the articles describing these functionals. 


-- 
Ar cieņu,
Igors Mihailovs
inženieris/doktorants
LU CFI Organisko materiālu laboratorija
--------------A289B4F97ADF07D17E2CA49C-- From owner-chemistry@ccl.net Tue Oct 18 10:46:00 2016 From: "Igors Mihailovs igorsm[]cfi.lu.lv" To: CCL Subject: CCL:G: HOMO-LUMO molecular orbitals Message-Id: <-52408-161018101236-19617-TKJJIFISb9WCHI4sikbp0Q[]server.ccl.net> X-Original-From: Igors Mihailovs Content-Type: multipart/alternative; boundary="------------2F6AA0827A196CB96B66748C" Date: Tue, 18 Oct 2016 17:12:26 +0300 MIME-Version: 1.0 Sent to CCL by: Igors Mihailovs [igorsm[A]cfi.lu.lv] This is a multi-part message in MIME format. --------------2F6AA0827A196CB96B66748C Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 8bit Dear Tymofii and Safiya, Yes, Tymofii is right about Kohn–Sham orbitals. We can even recall that the variational principle does not hold for approximate functionals... But are HF orbitals required to be better if the electron correlation is not accounted for in the computation? And electrons are again moving in some averaged potential of their counterparts, like in KS-DFT. I am not a real computational chemist, but as far as I can judge, there are quite severe approximations in both cases, although whereas HF and PT are continuously improving the model by adding new terms to it, KS-DFT uses alternative approach of building the structure "from scratch". There is also continuous evidence that the form of KS-DFT orbitals is usually quite rigorous (see classical paper on this topic by Stowasser and Hoffmann, 1999 http://dx.doi.org/10.1021/ja9826892, Chong, Gritsenko & Baerends, 2002 http://dx.doi.org/10.1063/1.1430255 – several-small-molecules study that concludes B3LYP (sic!) orbitals are closer to Dyson orbitals than HF ones). If I do not mess something up, what KS-DFT really can have problems with are degenerate states and, of course, multireference wavefunctions. As Safiya is studying physisorption, electron correlation can be quite important for description of the system (due to non-covalent interactions). Could this be the reason of HF and B3LYP orbital shape differences? Sincerely, I am just making conjectures. With best regards, Igors Mihailovs engineer/PhD student ISSP, University of Latvia On 18.10.2016 00:12, Tymofii Nikolaienko tim_mail-,-ukr.net wrote: > > Sent to CCL by: Tymofii Nikolaienko [tim_mail##ukr.net] > Dear Safiya Amer, > > sorry for saying such not-popular thing, but it is a big question > whether any orbitals obtained from DFT do > have any physical or chemical sense. > Just recall that most of the widely DFT functionals are based on > Kohn-Sham formalist - that is, the orbitals > you get are the orbitals for 'auxiliary' system of non-interacting > electrons and, strictly speaking, the only > linkage between this system and your real system is that they have the > same electron density. > However, having the same density does not imply having the set of the > same orbitals. > Thus, it you might want to consider an alternative of just ignoring > the shapes of DFT orbitals. > > Best regards, > Tymofii > > > > On 16.10.2016 22:01, Safiya Ess Amer amersaf85%%yahoo.com wrote: >> Sent to CCL by: "Safiya Ess Amer" [amersaf85 ~ yahoo.com] >> Hello for all CCL subscribers >> I'm studying a system of hydrogen molecule adsorbed on an ionic >> surface. I >> used Gaussian 03W programme with two methods HF and DFT-b3lyp to obtain >> molecular orbitals by using GaussView programme. this rout section is >> used for >> that: >> >> %Chk=Th90 b3lyp >> #T b3lyp nosymm gen pseudo=read pop=full >> >> Now I'm looking forward to explain the difference between what I >> got by using >> HF and DFT. in other words, I got a certain shape of molecular >> orbitals using >> HF and other different shape using DFT, how can I explain this >> difference! >> >> Thanks in advance >> >> >> Safiya Amer >> Graduate student >> Misrata University >> amersaf85^yahoo.com> --------------2F6AA0827A196CB96B66748C Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: 8bit Dear Tymofii and Safiya,


Yes, Tymofii is right about Kohn–Sham orbitals. We can even recall that the variational principle does not hold for approximate functionals... But are HF orbitals required to be better if the electron correlation is not accounted for in the computation? And electrons are again moving in some averaged potential of their counterparts, like in KS-DFT. I am not a real computational chemist, but as far as I can judge, there are quite severe approximations in both cases, although whereas HF and PT are continuously improving the model by adding new terms to it, KS-DFT uses alternative approach of building the structure "from scratch".

There is also continuous evidence that the form of KS-DFT orbitals is usually quite rigorous (see classical paper on this topic by Stowasser and Hoffmann, 1999 http://dx.doi.org/10.1021/ja9826892, Chong, Gritsenko & Baerends, 2002 http://dx.doi.org/10.1063/1.1430255 – several-small-molecules study that concludes B3LYP (sic!) orbitals are closer to Dyson orbitals than HF ones). If I do not mess something up, what KS-DFT really can have problems with are degenerate states and, of course, multireference wavefunctions.


As Safiya is studying physisorption, electron correlation can be quite important for description of the system (due to non-covalent interactions). Could this be the reason of HF and B3LYP orbital shape differences? Sincerely, I am just making conjectures.


With best regards,
Igors Mihailovs
engineer/PhD student
ISSP, University of Latvia

On 18.10.2016 00:12, Tymofii Nikolaienko tim_mail-,-ukr.net wrote:

Sent to CCL by: Tymofii Nikolaienko [tim_mail##ukr.net]
Dear Safiya Amer,

sorry for saying such not-popular thing, but it is a big question whether any orbitals obtained from DFT do
have any physical or chemical sense.
Just recall that most of the widely DFT functionals are based on Kohn-Sham formalist - that is, the orbitals
you get are the orbitals for 'auxiliary' system of non-interacting electrons and, strictly speaking, the only
linkage between this system and your real system is that they have the same electron density.
However, having the same density does not imply having the set of the same orbitals.
Thus, it you might want to consider an alternative of just ignoring the shapes of DFT orbitals.

Best regards,
Tymofii



On 16.10.2016 22:01, Safiya Ess Amer amersaf85%%yahoo.com wrote:
Sent to CCL by: "Safiya Ess Amer" [amersaf85 ~ yahoo.com]
Hello for all CCL subscribers
I'm studying a system of hydrogen molecule adsorbed on an ionic surface. I
used Gaussian 03W programme with two methods HF and DFT-b3lyp to obtain
molecular orbitals by using GaussView programme. this rout section is used for
that:

%Chk=Th90 b3lyp
#T b3lyp nosymm  gen pseudo=read pop=full

  Now I'm looking forward to explain the difference between what I got by using
HF and DFT. in other words, I got a certain shape of molecular orbitals using
HF and other different shape using DFT, how can I explain this difference!

Thanks in advance


Safiya Amer
Graduate student
Misrata University
amersaf85^yahoo.com>

--------------2F6AA0827A196CB96B66748C-- From owner-chemistry@ccl.net Tue Oct 18 11:52:00 2016 From: "Jim Kress jimkress35{:}gmail.com" To: CCL Subject: CCL: CCL #DFT16poll results are out Message-Id: <-52409-161018114923-9201-N3gC89FqmjWMyyvuFOK/xg(0)server.ccl.net> X-Original-From: "Jim Kress" Content-Language: en-us Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset="us-ascii" Date: Tue, 18 Oct 2016 11:49:10 -0400 MIME-Version: 1.0 Sent to CCL by: "Jim Kress" [jimkress35^_^gmail.com] Lehtola, Molt et al: Most rational people use the poll as a STARTING POINT and then investigate > from there. If you actually read the poll in depth it does provide good insight for further exploration, especially as it is divided into various categories of interest. I stated that the poll is helpful, not the final answer. It at least provides a starting point for those who are bewildered by the plethora of functionals. If you don't want to use it, fine. Don't belittle the efforts of others to bring some order out of the chaos of density functional land. Your attempt to belittle the good work of others does not make you more "superior". People who have to put others down are doing it for the purpose of raising themselves up - which inevitably leads to their own diminution. If you have a better way of bringing order out of this chaos, then provide it. Otherwise your denigrating comments are unhelpful and without merit. Jim Kress From owner-chemistry@ccl.net Tue Oct 18 12:26:00 2016 From: "Susi Lehtola susi.lehtola%x%alumni.helsinki.fi" To: CCL Subject: CCL: CCL #DFT16poll results are out Message-Id: <-52410-161018112451-25893-DxIN5vXroNNfWZXCVuYnXw!A!server.ccl.net> X-Original-From: Susi Lehtola Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=utf-8; format=flowed Date: Tue, 18 Oct 2016 08:24:35 -0700 MIME-Version: 1.0 Sent to CCL by: Susi Lehtola [susi.lehtola*o*alumni.helsinki.fi] On 10/18/2016 06:12 AM, Brian Skinn bskinn!A!alum.mit.edu wrote: > Susi, > >>> You could >>> get a much better picture just by tabulating the number of citations to the articles describing these >>> functionals. > > I disagree with your assertion that this would provide a better picture. > What fraction of the time do articles that use a given functional > *actually cite* (and cite correctly!) the original method paper for that > functional? Especially B3LYP -- the method is so standard for 'casual > computation' that I would not be surprised if it is not cited more > frequently than it is! But then again, everyone knows that B3LYP in chemistry and PBE in physics are the most widely used methods. In contrast, if you look at anything newer then the citations should be there. Counting citations is still more scientific than polling a random bunch of people which functional(s) do they like. That's a self-feeding loop: B3LYP is popular, because B3LYP is popular, because ... -- ----------------------------------------------------------------------- Mr. Susi Lehtola, PhD Chemist Postdoctoral Fellow susi.lehtola^^alumni.helsinki.fi Lawrence Berkeley National Laboratory http://www.helsinki.fi/~jzlehtol USA ----------------------------------------------------------------------- From owner-chemistry@ccl.net Tue Oct 18 13:01:01 2016 From: "Susi Lehtola susi.lehtola _ alumni.helsinki.fi" To: CCL Subject: CCL: CCL #DFT16poll results are out Message-Id: <-52411-161018113308-30895-xJ1PJfb367JXajeX1wYM5w:-:server.ccl.net> X-Original-From: Susi Lehtola Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=windows-1252; format=flowed Date: Tue, 18 Oct 2016 08:32:50 -0700 MIME-Version: 1.0 Sent to CCL by: Susi Lehtola [susi.lehtola]^[alumni.helsinki.fi] On 10/17/2016 11:59 PM, Marcel Swart marcel.swart-#-icrea.cat wrote: > Finally, I would like to remind the blog entry on Nature Chemistry's The Sceptical Chymist: > http://blogs.nature.com/thescepticalchymist/2014/11/five-years-of-polling-the-computational- > chemistry-community.html > where many arguments pro and contra have been put forward. To quote one of the godfathers of DFT: > > The DFT popularity poll is somewhat like citation analysis: It measures (but in a different way) how > well a functional has been received by a set of readers and users. There are many reasons why some > functionals are received better than others: accuracy, reliability, wide applicability, computational > efficiency, well-founded construction, availability in standard codes, reputation of the functional and > its authors, historical priority, novelty, and even hype. The poll has to be seen as measuring all these > things, and perhaps more. To the extent that the polled scientists use rational criteria, the results of > the poll can point other scientists toward good or interesting functionals > (John Perdew, 2014) Well, I'd say it boils down to just popularity (hype) and most of all, availability. Especially since most codes only have a limited few functionals available. Compare that to e.g. libxc which now has close to 400 functionals. Note: libxc is LGPL licensed, so it can be used also in closed-source and commercial codes! It's also very hard for new functionals to break through, because most people don't follow up on the literature. New review articles don't help if they aren't read. Instead of the poll bringing in new information like review articles do, it's just reinforcing old preconceptions. -- ----------------------------------------------------------------------- Mr. Susi Lehtola, PhD Chemist Postdoctoral Fellow susi.lehtola[]alumni.helsinki.fi Lawrence Berkeley National Laboratory http://www.helsinki.fi/~jzlehtol USA ----------------------------------------------------------------------- From owner-chemistry@ccl.net Tue Oct 18 13:45:00 2016 From: "Dr.N Sukumar n.sukumar]_[snu.edu.in" To: CCL Subject: CCL: CCL #DFT16poll results are out Message-Id: <-52412-161018133233-23755-Da1lRH/V6BEsWLXv7NJeOw_+_server.ccl.net> X-Original-From: "Dr.N Sukumar" Content-Type: multipart/alternative; boundary=089e013c66ac515b99053f2713fb Date: Tue, 18 Oct 2016 23:02:23 +0530 MIME-Version: 1.0 Sent to CCL by: "Dr.N Sukumar" [n.sukumar()snu.edu.in] --089e013c66ac515b99053f2713fb Content-Type: text/plain; charset=UTF-8 I too would like to defend the principle behind such a poll, even if not all the specifics of this particular poll. There are different paradigms for making decisions, especially regarding complex systems. One is the expert systems approach, which would correspond in this case to studying the reviews of respected and acknowledged experts, and following their recommendations. The majority of casual users of DFT have neither the time nor the inclination to do this, and even if they did, they don't have the expertise themselves to analyze competing claims. Do all the expert reviews agree on what is THE best functional? A different approach is crowd-sourcing, which has been shown to be perform competitively to expert systems in many situations and superior in some. I posit that the majority of casual users of DFT behave like individual investors in the stock market. They may just continue using the functional that worked for them before, or what was used in the paper they just read. At best they may ask their local computational chemist which functional to use. Individual investors too follow all these various investment strategies. The efficient markets hypothesis in economics assumes that the market has all the relevant information, even though individual investors may not. The "wisdom of the market" percolates down through the network, because experts like Warren Buffet and John Perdew are also the bigger influencers (through their reviews, lectures, *etc*), acting as hubs in their respective networks, whereby their opinions percolate down to the retail investors/casual users far more efficiently than those of others. Now one may question whether the assumptions underlying the "efficient markets hypothesis" and network theory are valid in this case, whether the law of large numbers required for a statistical analysis is met, whether some of the procedures and questions employed in the poll should be tweaked, and so forth, but I think one should not dismiss the approach out of hand. *N. SukumarProfessor of ChemistryDirector, Center for Informatics**Shiv Nadar University, India* On Tue, Oct 18, 2016 at 6:42 PM, Brian Skinn bskinn!A!alum.mit.edu < owner-chemistry\a/ccl.net> wrote: > Susi, > > >> You could > >> get a much better picture just by tabulating the number of citations to > the articles describing these > >> functionals. > > I disagree with your assertion that this would provide a better picture. > What fraction of the time do articles that use a given functional *actually > cite* (and cite correctly!) the original method paper for that functional? > Especially B3LYP -- the method is so standard for 'casual computation' that > I would not be surprised if it is not cited more frequently than it is! > > So, yes, as Dr. Perdew says in Dr. Swart's quote, a citation analysis > would give a different picture, providing different information -- but it > would in no substantial sense be "more authoritative" or "higher quality" > as a representation of usage patterns than is DFTpoll. > > > Best regards, > Brian > > On Tue, Oct 18, 2016 at 2:59 AM, Marcel Swart marcel.swart-#-icrea.cat < > owner-chemistry:ccl.net> wrote: > >> >> Sent to CCL by: "Marcel Swart" [marcel.swart---icrea.cat] >> >> On 17 Oct 2016, at 20:28 , Susi Lehtola susi.lehtola ~ >> alumni.helsinki.fi >> >> wrote: >> >> >>>> Sent to CCL by: "Jim Kress" [jimkress35%x%gmail.com] >> >>>> Marcel, Matthias, and Miquel; >> >> >>>> Great work as usual. Thanks for doing this. It is immensely >> helpful. >> >> Thanks Jim. >> >> >> I would tend to disagree. What the poll tells you is a very limited >> picture. Democracy has no say on >> >> what is the truth... Especially when you can't vote for what you'd >> like to, but are presented with a >> >> choice between what are essentially bad and worse functionals. >> [..] >> >> What represents the state of the art, for instance, the wB97X-V, >> B97M-V and wB97M-V >> functionals, >> >> have not been included in the poll, even though they beat the crap out >> of basically anything on the >> >> list. >> >> Thanks for the feedback Susi. I gather this is your way of saying that >> you would like these functionals >> to be included in next year's edition? Note that as stated on the >> website, rules and news-item the >> normal procedure is to send me a mail with your suggestions (as indeed >> people have done in the past). >> Note that there are "only" 10 slots available for new functionals, of >> which two are already taken for >> next year. With these three suggestions it goes to five. >> >> >> The poll seems to be centered on getting a picture of the functionals >> that people use. You could >> >> get a much better picture just by tabulating the number of citations >> to the articles describing these >> >> functionals. >> >> Finally, I would like to remind the blog entry on Nature Chemistry's The >> Sceptical Chymist: >> http://blogs.nature.com/thescepticalchymist/2014/11/five- >> years-of-polling-the-computational- >> chemistry-community.html >> >> where many arguments pro and contra have been put forward. To quote one >> of the godfathers of DFT: >> >> The DFT popularity poll is somewhat like citation analysis: It measures >> (but in a different way) how >> well a functional has been received by a set of readers and users. There >> are many reasons why some >> functionals are received better than others: accuracy, reliability, wide >> applicability, computational >> efficiency, well-founded construction, availability in standard codes, >> reputation of the functional and >> its authors, historical priority, novelty, and even hype. The poll has >> to be seen as measuring all these >> things, and perhaps more. To the extent that the polled scientists use >> rational criteria, the results of >> the poll can point other scientists toward good or interesting functionals >> (John Perdew, 2014)>> E-mail to subscribers: CHEMISTRY:ccl.net or use:>> >> E-mail to administrators: CHEMISTRY-REQUEST:ccl.net or use>> >> >> > --089e013c66ac515b99053f2713fb Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable
I too would like to defend the principle behind such a pol= l, even if not all the specifics of this particular poll. There are differe= nt paradigms for making decisions, especially regarding complex systems. On= e is the expert systems approach, which would correspond in this case to st= udying the reviews of respected and acknowledged experts, and following the= ir recommendations. The majority of casual users of DFT have neither the ti= me nor the inclination to do this, and even if they did, they don't hav= e the expertise themselves to analyze competing claims. Do all the expert r= eviews agree on what is THE best functional?

A different approach is= crowd-sourcing, which has been shown to be perform competitively to expert= systems in many situations and superior in some. I posit that the majority= of casual users of DFT behave like individual investors in the stock marke= t. They may just continue using the functional that worked for them before,= or what was used in the paper they just read. At best they may ask their l= ocal computational chemist which functional to use. Individual investors to= o follow all these various investment strategies. The efficient markets hyp= othesis in economics assumes that the market has all the relevant informati= on, even though individual investors may not. The "wisdom of the marke= t" percolates down through the network, because experts like Warren Bu= ffet and John Perdew are also the bigger influencers (through their reviews= , lectures, etc), acting as hubs in their respective networks, where= by their opinions percolate down to the retail investors/casual users far m= ore efficiently than those of others.

Now one may question whether t= he assumptions underlying the "efficient markets hypothesis" and = network theory are valid in this case, whether the law of large numbers req= uired for a statistical analysis is met, whether some of the procedures and= questions employed in the poll should be tweaked, and so forth, but I thin= k one should not dismiss the approach out of hand.

N. Sukumar
Professor of Chemistry
Director, Center for Informatic= s
Shiv Nadar University, India


On Tue, Oct 18, 2016 at 6:42 PM, Brian Skinn= bskinn!A!alum.mit.edu <owner-ch= emistry\a/ccl.net> wrote:
Susi= ,

>> You could
>> get a m= uch better picture just by tabulating the number of citations to the articl= es describing these
>> functionals.<= /span>

I = disagree with your assertion that this would provide a better picture. What= fraction of the time do articles that use a given functional *actually cit= e* (and cite correctly!) the original method paper for that functional? Esp= ecially B3LYP -- the method is so standard for 'casual computation'= that I would not be surprised if it is not cited more frequently than it i= s!
So, = yes, as Dr. Perdew says in Dr. Swart's quote, a citation analysis would= give a different picture, providing different information -- but it would = in no substantial sense be "more authoritative" or "higher q= uality" as a representation of usage patterns than is DFTpoll.<= /div>


Best regards,
Brian

= On Tue, Oct 18, 2016 at 2:59 AM, Marcel Swart marcel.swart= -#-icrea.cat <o= wner-chemistry:ccl.net> wrote:

Sent to CCL by: "Marcel=C2=A0 Swart" [marcel.swart---icrea.= cat]
 >> On 17 Oct 2016, at 20:28 , Susi Lehtola susi.lehtola ~ alumni.hels= inki.fi <owner-chemistry|ccl.net>
>> wrote:

>>>> Sent to CCL by: "Jim Kress" [jimkress35%x%gmail.com]<= br> >>>> Marcel, Matthias, and Miquel;

>>>> Great work as usual.=C2=A0 Thanks for doing this.=C2=A0 It= is immensely helpful.

Thanks Jim.

>> I would tend to disagree. What the poll tells you is a very limite= d picture. Democracy has no say on
>> what is the truth... Especially when you can't vote for what y= ou'd like to, but are presented with a
>> choice between what are essentially bad and worse functionals.
[..]
>> What represents the state of the art, for instance, the wB97= X-V, B97M-V and wB97M-V
functionals,
>> have not been included in the poll, even though they beat the crap= out of basically anything on the
>>=C2=A0 list.

Thanks for the feedback Susi. I gather this = is your way of saying that you would like these functionals
to be included in next year's edition? Note that as stated on the websi= te, rules and news-item the
normal procedure is to send me a mail with your suggestions (as indeed peop= le have done in the past).
Note that there are "only" 10 slots available for new functionals= , of which two are already taken for
next year. With these three suggestions it goes to five.

>> The poll seems to be centered on getting a picture of the function= als that people use. You could
>> get a much better picture just by tabulating the number of citatio= ns to the articles describing these
>> functionals.

Finally, I would like to remind the blog ent= ry on Nature Chemistry's The Sceptical Chymist:
http://blogs.nature.com/thescepticalchymist/2014/11/fi= ve-years-of-polling-the-computational-
chemistry-community.html
where many arguments pro and contra have been put forward. To quote one of = the godfathers of DFT:

The DFT popularity poll is somewhat like citation analysis: It measures (bu= t in a different way) how
well a functional has been received by a set of readers and users.=C2=A0 Th= ere are many reasons why some
functionals are received better than others: accuracy, reliability, wide ap= plicability, computational
efficiency, well-founded construction, availability in standard codes, repu= tation of the functional and
its authors, historical priority, novelty, and even hype.=C2=A0 The poll ha= s to be seen as measuring all these
things, and perhaps more. To the extent that the polled scientists use rati= onal criteria, the results of
the poll can point other scientists toward good or interesting functionals<= br> (John Perdew, 2014)



-=3D This is automatically added to each message by the mailing script =3D-=
E-mail to subscribers: CHEMISTRY:ccl.net or use:
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/cgi-bin/ccl/send_ccl_message

E-mail to administrators: CHEMISTRY-REQUEST:ccl.net or use
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/cgi-bin/ccl/send_ccl_message

Subscribe/Unsubscribe:
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/chemistry/s= ub_unsub.shtml

Before posting, check wait time at: http://www.ccl.net

Job: http://www.ccl.net/jobs
Conferences: http://server.ccl.net/chemist= ry/announcements/conferences/

Search Messages: http://www.ccl.net/chemistry/searchccl/index.shtml
 =C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutcc= l/instructions/




--089e013c66ac515b99053f2713fb-- From owner-chemistry@ccl.net Tue Oct 18 15:49:00 2016 From: "Dr. Robert Molt Jr. r.molt.chemical.physics{=}gmail.com" To: CCL Subject: CCL: CCL #DFT16poll results are out Message-Id: <-52413-161018154802-19375-/sJF2B0kYRkqBwnBe+4mmQ()server.ccl.net> X-Original-From: "Dr. Robert Molt Jr." Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=windows-1252; format=flowed Date: Tue, 18 Oct 2016 15:47:54 -0400 MIME-Version: 1.0 Sent to CCL by: "Dr. Robert Molt Jr." [r.molt.chemical.physics : gmail.com] Kress: I was quite clear in "bringing order out of this chaos": read review articles. That's the intelligent starting point. Doing a search in google or a journal for a review article is not hard. Isn't that the entire point of a review article? To give overviews of the grand complexity of a field? I have offered disagreement with the value of something. If you feel disagreement is equivalent to "denigration," I suggest you examine your own illustrious history of "disagreement"on CCL ;) On 10/18/16 11:49 AM, Jim Kress jimkress35{:}gmail.com wrote: > Sent to CCL by: "Jim Kress" [jimkress35^_^gmail.com] > Lehtola, Molt et al: > > Most rational people use the poll as a STARTING POINT and then investigate >> from there. If you actually read the poll in depth it does provide good > insight for further exploration, especially as it is divided into various > categories of interest. > > I stated that the poll is helpful, not the final answer. It at least > provides a starting point for those who are bewildered by the plethora of > functionals. > > If you don't want to use it, fine. Don't belittle the efforts of others to > bring some order out of the chaos of density functional land. Your attempt > to belittle the good work of others does not make you more "superior". > People who have to put others down are doing it for the purpose of raising > themselves up - which inevitably leads to their own diminution. > > If you have a better way of bringing order out of this chaos, then provide > it. Otherwise your denigrating comments are unhelpful and without merit. > > Jim Kress> > From owner-chemistry@ccl.net Tue Oct 18 16:23:00 2016 From: "Tymofii Nikolaienko tim_mail _ ukr.net" To: CCL Subject: CCL:G: HOMO-LUMO molecular orbitals Message-Id: <-52414-161018142322-21098-Rvk/GIof0+lKu4W83LdwyA__server.ccl.net> X-Original-From: Tymofii Nikolaienko Content-Type: multipart/alternative; boundary="------------4A6053CC9F29095AAC5879E0" Date: Tue, 18 Oct 2016 21:23:06 +0300 MIME-Version: 1.0 Sent to CCL by: Tymofii Nikolaienko [tim_mail^ukr.net] This is a multi-part message in MIME format. --------------4A6053CC9F29095AAC5879E0 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 8bit Hm... To be honest, "The nature of the chemical bond—1990: *There are no such things as orbitals*!" [ J. F. Ogilvie, J. Chem. Educ., 1990, 67 (4), p 280, DOI: 10.1021/ed067p280 , http://pubs.acs.org/doi/abs/10.1021/ed067p280 -- a rather long-lasting discussions for which I can provide more references, both 'pros' and 'cons' if necessary; of course, notwithstanding this discussions I do use the concept of orbitals myself - but mostly in order to understand my system in a 'chemical' terms (and get some heuristics), but perhaps _not_ in order to explain why certain interaction/bonding occurs ] :) Going seriously, my statement was merely that in Hartree-Fock (where all limitations of this theory are known in advance) we get (canonical) orbitals as eigenfunctions of the one-particle Fock operator and we can 'track' a clear and direct relationship of this operator to the real physical system under study. In contrast to this, in Kohn-Sham DFT we get the orbitals as eigenfunctions of an operator which was set-up for a 'completely' different system (a Kohn-Sham system) which is a priori (!) a different system. (note that in DFT we do know the electron density of the system but we never know the wavefunction!) Of course, I realize that in practice all these fundamental distinctions are often quite small (if not negligible), but nonetheless it's a pity that often researchers simply forget/ignore the fact that these distinctions exist -- for example, when a 2-nd order perturbational analysis within the NBO method is applied to the 'density matrix' obtained from DFT. Best regards, Tymofii On 18.10.2016 17:12, Igors Mihailovs igorsm[]cfi.lu.lv wrote: > Dear Tymofii and Safiya, > > > Yes, Tymofii is right about Kohn–Sham orbitals. We can even recall > that the variational principle does not hold for approximate > functionals... But are HF orbitals required to be better if the > electron correlation is not accounted for in the computation? And > electrons are again moving in some averaged potential of their > counterparts, like in KS-DFT. I am not a real computational chemist, > but as far as I can judge, there are quite severe approximations in > both cases, although whereas HF and PT are continuously improving the > model by adding new terms to it, KS-DFT uses alternative approach of > building the structure "from scratch". > > There is also continuous evidence that the form of KS-DFT orbitals is > usually quite rigorous (see classical paper on this topic by Stowasser > and Hoffmann, 1999 http://dx.doi.org/10.1021/ja9826892, Chong, > Gritsenko & Baerends, 2002 http://dx.doi.org/10.1063/1.1430255 – > several-small-molecules study that concludes B3LYP (sic!) orbitals are > closer to Dyson orbitals than HF ones). If I do not mess something up, > what KS-DFT really can have problems with are degenerate states and, > of course, multireference wavefunctions. > > > As Safiya is studying physisorption, electron correlation can be quite > important for description of the system (due to non-covalent > interactions). Could this be the reason of HF and B3LYP orbital shape > differences? Sincerely, I am just making conjectures. > > > With best regards, > Igors Mihailovs > engineer/PhD student > ISSP, University of Latvia > > On 18.10.2016 00:12, Tymofii Nikolaienko tim_mail-,-ukr.net wrote: >> >> Sent to CCL by: Tymofii Nikolaienko [tim_mail##ukr.net] >> Dear Safiya Amer, >> >> sorry for saying such not-popular thing, but it is a big question >> whether any orbitals obtained from DFT do >> have any physical or chemical sense. >> Just recall that most of the widely DFT functionals are based on >> Kohn-Sham formalist - that is, the orbitals >> you get are the orbitals for 'auxiliary' system of non-interacting >> electrons and, strictly speaking, the only >> linkage between this system and your real system is that they have >> the same electron density. >> However, having the same density does not imply having the set of the >> same orbitals. >> Thus, it you might want to consider an alternative of just ignoring >> the shapes of DFT orbitals. >> >> Best regards, >> Tymofii >> >> >> >> On 16.10.2016 22:01, Safiya Ess Amer amersaf85%%yahoo.com wrote: >>> Sent to CCL by: "Safiya Ess Amer" [amersaf85 ~ yahoo.com] >>> Hello for all CCL subscribers >>> I'm studying a system of hydrogen molecule adsorbed on an ionic >>> surface. I >>> used Gaussian 03W programme with two methods HF and DFT-b3lyp to obtain >>> molecular orbitals by using GaussView programme. this rout section >>> is used for >>> that: >>> >>> %Chk=Th90 b3lyp >>> #T b3lyp nosymm gen pseudo=read pop=full >>> >>> Now I'm looking forward to explain the difference between what I >>> got by using >>> HF and DFT. in other words, I got a certain shape of molecular >>> orbitals using >>> HF and other different shape using DFT, how can I explain this >>> difference! >>> >>> Thanks in advance >>> >>> >>> Safiya Amer >>> Graduate student >>> Misrata University >>> amersaf85^yahoo.com> > --------------4A6053CC9F29095AAC5879E0 Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: 8bit

Hm... To be honest,
"The nature of the chemical bond—1990: There are no such things as orbitals!" [ J. F. Ogilvie, J. Chem. Educ., 1990, 67 (4), p 280, DOI: 10.1021/ed067p280 ,
http://pubs.acs.org/doi/abs/10.1021/ed067p280  -- a rather long-lasting discussions for which I can provide more references, both
'pros' and 'cons' if necessary; of course, notwithstanding this discussions I do use the concept of orbitals myself - but mostly in order to
understand my system in a 'chemical' terms (and get some heuristics), but perhaps _not_ in order to explain why certain interaction/bonding occurs  ]
:)

Going seriously, my statement was merely that in Hartree-Fock (where all limitations of this theory are known in advance) we
get (canonical) orbitals as eigenfunctions of the one-particle Fock operator and we can 'track' a clear and direct relationship of
this operator to the real physical system under study. In contrast to this, in Kohn-Sham DFT we get the orbitals as eigenfunctions
of an operator which was set-up for a 'completely' different system (a Kohn-Sham system) which is a priori (!) a different system.
(note that in DFT we do know the electron density of the system but we never know the wavefunction!)
Of course, I realize that in practice all these fundamental distinctions are often quite small (if not negligible), but nonetheless
it's a pity that often researchers simply forget/ignore the fact that these distinctions exist -- for example, when a
2-nd order perturbational analysis within the NBO method is applied to the 'density matrix' obtained from DFT.


Best regards,
Tymofii



On 18.10.2016 17:12, Igors Mihailovs igorsm[]cfi.lu.lv wrote:
Dear Tymofii and Safiya,


Yes, Tymofii is right about Kohn–Sham orbitals. We can even recall that the variational principle does not hold for approximate functionals... But are HF orbitals required to be better if the electron correlation is not accounted for in the computation? And electrons are again moving in some averaged potential of their counterparts, like in KS-DFT. I am not a real computational chemist, but as far as I can judge, there are quite severe approximations in both cases, although whereas HF and PT are continuously improving the model by adding new terms to it, KS-DFT uses alternative approach of building the structure "from scratch".

There is also continuous evidence that the form of KS-DFT orbitals is usually quite rigorous (see classical paper on this topic by Stowasser and Hoffmann, 1999 http://dx.doi.org/10.1021/ja9826892, Chong, Gritsenko & Baerends, 2002 http://dx.doi.org/10.1063/1.1430255 – several-small-molecules study that concludes B3LYP (sic!) orbitals are closer to Dyson orbitals than HF ones). If I do not mess something up, what KS-DFT really can have problems with are degenerate states and, of course, multireference wavefunctions.


As Safiya is studying physisorption, electron correlation can be quite important for description of the system (due to non-covalent interactions). Could this be the reason of HF and B3LYP orbital shape differences? Sincerely, I am just making conjectures.


With best regards,
Igors Mihailovs
engineer/PhD student
ISSP, University of Latvia

On 18.10.2016 00:12, Tymofii Nikolaienko tim_mail-,-ukr.net wrote:

Sent to CCL by: Tymofii Nikolaienko [tim_mail##ukr.net]
Dear Safiya Amer,

sorry for saying such not-popular thing, but it is a big question whether any orbitals obtained from DFT do
have any physical or chemical sense.
Just recall that most of the widely DFT functionals are based on Kohn-Sham formalist - that is, the orbitals
you get are the orbitals for 'auxiliary' system of non-interacting electrons and, strictly speaking, the only
linkage between this system and your real system is that they have the same electron density.
However, having the same density does not imply having the set of the same orbitals.
Thus, it you might want to consider an alternative of just ignoring the shapes of DFT orbitals.

Best regards,
Tymofii



On 16.10.2016 22:01, Safiya Ess Amer amersaf85%%yahoo.com wrote:
Sent to CCL by: "Safiya Ess Amer" [amersaf85 ~ yahoo.com]
Hello for all CCL subscribers
I'm studying a system of hydrogen molecule adsorbed on an ionic surface. I
used Gaussian 03W programme with two methods HF and DFT-b3lyp to obtain
molecular orbitals by using GaussView programme. this rout section is used for
that:

%Chk=Th90 b3lyp
#T b3lyp nosymm  gen pseudo=read pop=full

  Now I'm looking forward to explain the difference between what I got by using
HF and DFT. in other words, I got a certain shape of molecular orbitals using
HF and other different shape using DFT, how can I explain this difference!

Thanks in advance


Safiya Amer
Graduate student
Misrata University
amersaf85^yahoo.com>


--------------4A6053CC9F29095AAC5879E0-- From owner-chemistry@ccl.net Tue Oct 18 16:59:00 2016 From: "Susi Lehtola susi.lehtola(-)alumni.helsinki.fi" To: CCL Subject: CCL: CCL #DFT16poll results are out Message-Id: <-52415-161018144139-21179-pXn/U8kCPHb06EGaeoAIZg#%#server.ccl.net> X-Original-From: Susi Lehtola Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=windows-1252; format=flowed Date: Tue, 18 Oct 2016 11:41:25 -0700 MIME-Version: 1.0 Sent to CCL by: Susi Lehtola [susi.lehtola-.-alumni.helsinki.fi] On 10/18/2016 08:49 AM, Jim Kress jimkress35{:}gmail.com wrote: > If you have a better way of bringing order out of this chaos, then provide > it. Otherwise your denigrating comments are unhelpful and without merit. I guess you didn't read my messages. I have given two suggestions: include newer functionals and/or don't prescreen the poll by only including old functionals. -- ----------------------------------------------------------------------- Mr. Susi Lehtola, PhD Chemist Postdoctoral Fellow susi.lehtola~!~alumni.helsinki.fi Lawrence Berkeley National Laboratory http://www.helsinki.fi/~jzlehtol USA ----------------------------------------------------------------------- From owner-chemistry@ccl.net Tue Oct 18 17:33:00 2016 From: "Andreas Klamt klamt]![cosmologic.de" To: CCL Subject: CCL: CCL #DFT16poll results are out Message-Id: <-52416-161018164252-9951-DxIN5vXroNNfWZXCVuYnXw]![server.ccl.net> X-Original-From: Andreas Klamt Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=windows-1252 Date: Tue, 18 Oct 2016 22:42:41 +0200 MIME-Version: 1.0 Sent to CCL by: Andreas Klamt [klamt-.-cosmologic.de] to Jim Kress: Jim, you may be right in your defense for the poll (although I do not like if science becomes a question of majority votes). But the way your are bashing the people who wrote critical comments on the poll is not acceptable. CCl is a discussion forum, and critical voices must be allowed, yes they should be welcome!!! We need more people in science who have the courage to raise critical arguments!!! If the majority has good scientific arguments, these arguments may be repeated. That is o.k. But a personal offense against people raising critical opinions is unacceptable. Best regards Andreas Am 18.10.2016 um 17:49 schrieb Jim Kress jimkress35{:}gmail.com: > Sent to CCL by: "Jim Kress" [jimkress35^_^gmail.com] > Lehtola, Molt et al: > > Most rational people use the poll as a STARTING POINT and then investigate >> from there. If you actually read the poll in depth it does provide good > insight for further exploration, especially as it is divided into various > categories of interest. > > I stated that the poll is helpful, not the final answer. It at least > provides a starting point for those who are bewildered by the plethora of > functionals. > > If you don't want to use it, fine. Don't belittle the efforts of others to > bring some order out of the chaos of density functional land. Your attempt > to belittle the good work of others does not make you more "superior". > People who have to put others down are doing it for the purpose of raising > themselves up - which inevitably leads to their own diminution. > > If you have a better way of bringing order out of this chaos, then provide > it. Otherwise your denigrating comments are unhelpful and without merit. > > Jim Kress> > > -- -------------------------------------------------- Prof. Dr. Andreas Klamt CEO / Geschftsfhrer COSMOlogic GmbH & Co. KG Imbacher Weg 46 D-51379 Leverkusen, Germany phone +49-2171-731681 fax +49-2171-731689 e-mail klamt!=!cosmologic.de web www.cosmologic.de [University address: Inst. of Physical and Theoretical Chemistry, University of Regensburg] HRA 20653 Amtsgericht Koeln, GF: Prof. Dr. Andreas Klamt Komplementaer: COSMOlogic Verwaltungs GmbH HRB 49501 Amtsgericht Koeln, GF: Prof. Dr. Andreas Klamt From owner-chemistry@ccl.net Tue Oct 18 18:09:00 2016 From: "Jim Kress jimkress35#gmail.com" To: CCL Subject: CCL: CCL #DFT16poll results are out Message-Id: <-52417-161018170341-20713-JogCPQK5nTkootXxHc970g++server.ccl.net> X-Original-From: "Jim Kress" Content-Language: en-us Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset="us-ascii" Date: Tue, 18 Oct 2016 17:03:19 -0400 MIME-Version: 1.0 Sent to CCL by: "Jim Kress" [jimkress35|-|gmail.com] Lehtola, It's easy to complain and point out flaws. I HAVE YET TO SEE YOUR SOLUTION. Jim Kress -----Original Message----- > From: owner-chemistry+jimkress35==gmail.com:ccl.net [mailto:owner-chemistry+jimkress35==gmail.com:ccl.net] On Behalf Of Susi Lehtola susi.lehtola _ alumni.helsinki.fi Sent: Tuesday, October 18, 2016 11:33 AM To: Kress, Jim Subject: CCL: CCL #DFT16poll results are out Sent to CCL by: Susi Lehtola [susi.lehtola]^[alumni.helsinki.fi] On 10/17/2016 11:59 PM, Marcel Swart marcel.swart-#-icrea.cat wrote: > Finally, I would like to remind the blog entry on Nature Chemistry's The Sceptical Chymist: > http://blogs.nature.com/thescepticalchymist/2014/11/five-years-of-poll > ing-the-computational- > chemistry-community.html > where many arguments pro and contra have been put forward. To quote one of the godfathers of DFT: > > The DFT popularity poll is somewhat like citation analysis: It > measures (but in a different way) how well a functional has been > received by a set of readers and users. There are many reasons why > some functionals are received better than others: accuracy, > reliability, wide applicability, computational efficiency, > well-founded construction, availability in standard codes, reputation > of the functional and its authors, historical priority, novelty, and > even hype. The poll has to be seen as measuring all these things, and > perhaps more. To the extent that the polled scientists use rational > criteria, the results of the poll can point other scientists toward > good or interesting functionals (John Perdew, 2014) Well, I'd say it boils down to just popularity (hype) and most of all, availability. Especially since most codes only have a limited few functionals available. Compare that to e.g. libxc which now has close to 400 functionals. Note: libxc is LGPL licensed, so it can be used also in closed-source and commercial codes! It's also very hard for new functionals to break through, because most people don't follow up on the literature. New review articles don't help if they aren't read. Instead of the poll bringing in new information like review articles do, it's just reinforcing old preconceptions. -- ----------------------------------------------------------------------- Mr. Susi Lehtola, PhD Chemist Postdoctoral Fellow susi.lehtola/a\alumni.helsinki.fi Lawrence Berkeley National Laboratory http://www.helsinki.fi/~jzlehtol USA -----------------------------------------------------------------------http://www.ccl.net/cgi-bin/ccl/send_ccl_messagehttp://www.ccl.net/chemistry/sub_unsub.shtmlhttp://www.ccl.net/spammers.txt