From owner-chemistry@ccl.net Thu Sep 5 02:28:00 2019 From: "Susi Lehtola susi.lehtola/a\helsinki.fi" To: CCL Subject: CCL:G: Structure breaks even after started with correct bond length Message-Id: <-53832-190905022553-20469-n1MHFmEV2EcKQBGe4PDqaA..server.ccl.net> X-Original-From: Susi Lehtola Content-Language: en-US Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=utf-8; format=flowed Date: Thu, 5 Sep 2019 09:25:43 +0300 MIME-Version: 1.0 Sent to CCL by: Susi Lehtola [susi.lehtola[-]helsinki.fi] On 9/4/19 8:46 PM, Nikhil Maroli scinikhil!=!gmail.com wrote: > Hi, > There are no studies available on this structure ( the structure also > not reported yet). The available calculations were performed using the > CP2K package with mixed Gaussian and plane-wave basis set, using the > molecularly optimized MOLOPT double f-valence polarized (mDZVP) basis > set implemented in CP2K code. Im not asking for a basis set for my > study. I have used B3LYP, CCSD and MP2 function with different basis set > to obtain an optimization geometry. I ended up with different errors and > the structure is broken at the end. I came to know the possibility of > freezing the coordinates, but at the same time, there is suggestion that > the freezing is unrealistic and may not be accurate. So in a nutshell, > information from anyone who had a similar situation which is resolved > would be helpful for me. > Thank you for the help Dear Nikhil, I'm afraid the method you are running, ump2=full/lanl2mb, makes very little sense due to several reasons. First, lanl2mb is a minimal basis set and as such gives unreliable results even at Hartree-Fock (HF) and density functional level. You need to use a bigger basis set, e.g. def2SVP. Second, post-Hartree-Fock methods like MP2 and CCSD require really large basis sets to reach sufficient convergence: usually one would use at least triple- or quadruple-zeta, and extrapolate to the complete basis set limit. At lanl2mb level you just get garbage. If you can't handle big basis sets, use DFT instead. Third, I'm not sure what the electronic structure of your perovskite is, but HF and post-HF methods yield unreliable results for metallic systems due to the small/vanishing band gap and significant spin contamination. You can see whether this happens by running DFT with different fractions of exact exchange, e.g. PBE vs PBE0. As Dr Cuffe already pointed out, the periodic and cluster structures will differ due to surface effects. If I was you, I'd start with the methods and geometry of the CP2K calculations. If you can reproduce the periodic calculation, this gives you a good starting point for studying how using a finite size cluster changes the results. -- ------------------------------------------------------------------ Mr. Susi Lehtola, PhD Junior Fellow, Adjunct Professor susi.lehtola_+_alumni.helsinki.fi University of Helsinki http://susilehtola.github.io/ Finland ------------------------------------------------------------------ Susi Lehtola, dosentti, FT tutkijatohtori susi.lehtola_+_alumni.helsinki.fi Helsingin yliopisto http://susilehtola.github.io/ ------------------------------------------------------------------ From owner-chemistry@ccl.net Thu Sep 5 12:09:01 2019 From: "Alex Mathew alexmathewmd*o*gmail.com" To: CCL Subject: CCL:G: Gaussian running with GPU Message-Id: <-53833-190905060346-28941-8Vuvwi5sgSpIit/qaND2ZQ[a]server.ccl.net> X-Original-From: "Alex Mathew" Date: Thu, 5 Sep 2019 06:03:45 -0400 Sent to CCL by: "Alex Mathew" [alexmathewmd%a%gmail.com] Dear all, I m having a workstation with 3 GTX 1070 card and 2 x 12 core processor (48 threads). Can anyone share me the keword/input file to use GPU+CPU in Gaussian-16. I have tried the following %CPU=0-36 !to use 36 thread for processor %GPUCPU=0,1,2=37,38,39 ! three gpu with id 0,1,2 But it returned syntax error. From owner-chemistry@ccl.net Thu Sep 5 13:48:00 2019 From: "Uche Anene uche a uconn.edu" To: CCL Subject: CCL:G: Structure breaks even after started with correct bond length Message-Id: <-53834-190905134230-31362-x4GkV8HXvqmDowXtnTLnQQ() server.ccl.net> X-Original-From: Uche Anene Content-Type: multipart/alternative; boundary="Apple-Mail=_ED054F72-D563-47D9-931E-C22B6A0B0EC9" Date: Thu, 5 Sep 2019 13:42:20 -0400 Mime-Version: 1.0 (Mac OS X Mail 12.4 \(3445.104.11\)) Sent to CCL by: Uche Anene [uche^_^uconn.edu] --Apple-Mail=_ED054F72-D563-47D9-931E-C22B6A0B0EC9 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=utf-8 Hi Nikhil, When I have seen a Gaussian error message like the one you mentioned=20 "Largest valence mixing into a core orbital is 4.37D-01 Largest core mixing into a valence orbital is 3.40D-01 Excessive mixing of frozen core and valence orbitals.=E2=80=9D It usually means basis set is too big or too small for some of the = atoms. You could try mixing the basis set by using "GEN=E2=80=9D = command. You could apply cc-pVDZ to metal atoms and 6-311G(d,p) for = non-metals. SCF=3DQC will probably not help you in this case. It=E2=80=99= s a way of taking smaller steps to find a stationary point. Placing the = =E2=80=9Cu=E2=80=9D in front of your functional is just if you want = unrestricted. Gaussian's default is to use (r) restricted for singlet = and (u) unrestricted for higher multiplicities, unless you specify it. = So, If all you atoms are paired and your multiplicity is 1, you don=E2=80=99= t need to define it. Gaussian=E2=80=99s default is (r). In this example I will use PBE; It has worked well for me. You could use = another DFT functional or another flavor of PBE. I increased my SCF = cycle to 1000. #P OPT FREQ LC-wPBE/GEN GEO=3Dconnectivity SCF=3DMaxcycle=3D1000 Title 0 1 Geometry Cu 0 cc-pVDZ **** C O H 0 6-311G(d,p) **** Cu 0 cc-pVDZ **** Also, If you have multiple metals, they need to be described separately = as shown above and separated by a space and (****) Gaussian will give you an error message if it is not correct. You can = reference here https://gaussian.com/gen/ =20 --- Uchenna A. Anene Ph.D. Student, Chemical Engineering =20 University of Connecticut Innovation Partnership Building=20 159 Discovery Drive Storrs, CT 06269 uche##uconn.edu https://alpay.ims.uconn.edu/uche-anene/# www.linkedin.com/in/uche-anene =20= https://www.researchgate.net/profile/Uchenna_Anene = > On Sep 5, 2019, at 2:25 AM, Susi Lehtola susi.lehtola/ahelsinki.fi = wrote: >=20 >=20 > Sent to CCL by: Susi Lehtola [susi.lehtola[-]helsinki.fi] > On 9/4/19 8:46 PM, Nikhil Maroli scinikhil!=3D!gmail.com wrote: >> Hi, >> There are no studies available on this structure ( the structure also = not reported yet). The available calculations were performed using the = CP2K package with mixed Gaussian and plane-wave basis set, using the = molecularly optimized MOLOPT double f-valence polarized (mDZVP) basis = set implemented in CP2K code. Im not asking for a basis set for my = study. I have used B3LYP, CCSD and MP2 function with different basis set = to obtain an optimization geometry. I ended up with different errors and = the structure is broken at the end. I came to know the possibility of = freezing the coordinates, but at the same time, there is suggestion that = the freezing is unrealistic and may not be accurate. So in a nutshell, = information from anyone who had a similar situation which is resolved = would be helpful for me. >> Thank you for the help >=20 > Dear Nikhil, >=20 >=20 > I'm afraid the method you are running, ump2=3Dfull/lanl2mb, makes very = little sense due to several reasons. >=20 > First, lanl2mb is a minimal basis set and as such gives unreliable = results even at Hartree-Fock (HF) and density functional level. You need = to use a bigger basis set, e.g. def2SVP. >=20 > Second, post-Hartree-Fock methods like MP2 and CCSD require really = large basis sets to reach sufficient convergence: usually one would use = at least triple- or quadruple-zeta, and extrapolate to the complete = basis set limit. At lanl2mb level you just get garbage. If you can't = handle big basis sets, use DFT instead. >=20 > Third, I'm not sure what the electronic structure of your perovskite = is, but HF and post-HF methods yield unreliable results for metallic = systems due to the small/vanishing band gap and significant spin = contamination. You can see whether this happens by running DFT with = different fractions of exact exchange, e.g. PBE vs PBE0. >=20 > As Dr Cuffe already pointed out, the periodic and cluster structures = will differ due to surface effects. If I was you, I'd start with the = methods and geometry of the CP2K calculations. If you can reproduce the = periodic calculation, this gives you a good starting point for studying = how using a finite size cluster changes the results. > --=20 > ------------------------------------------------------------------ > Mr. Susi Lehtola, PhD Junior Fellow, Adjunct Professor > susi.lehtola,,alumni.helsinki.fi University of Helsinki > http://susilehtola.github.io/ Finland > ------------------------------------------------------------------ > Susi Lehtola, dosentti, FT tutkijatohtori > susi.lehtola,,alumni.helsinki.fi Helsingin yliopisto > http://susilehtola.github.io/ > ------------------------------------------------------------------ >=20 >=20 >=20 > -=3D This is automatically added to each message by the mailing script = =3D- > To recover the email address of the author of the message, please = change>=20>=20>=20=>=20>=20Conferences: = http://server.ccl.net/chemistry/announcements/conferences/ >=20>=20>=20>=20 >=20 --Apple-Mail=_ED054F72-D563-47D9-931E-C22B6A0B0EC9 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=utf-8 Hi = Nikhil,

When I have = seen a Gaussian error message like the one you mentioned 
"Largest valence mixing into a core orbital is =  4.37D-01
 Largest core mixing into a valence = orbital is  3.40D-01
 Excessive mixing of frozen = core and valence orbitals.=E2=80=9D

It usually means basis set is too big = or too small for some of the atoms. You could try mixing the basis set = by using "GEN=E2=80=9D command. You could apply cc-pVDZ to metal atoms = and 6-311G(d,p) for non-metals.  SCF=3DQC will probably not help = you in this case. It=E2=80=99s a way of taking smaller steps to find a = stationary point. Placing the =E2=80=9Cu=E2=80=9D in front of your = functional is just if you want unrestricted. Gaussian's default is to = use (r) restricted for singlet and (u) unrestricted for higher = multiplicities, unless you specify it. So, If all you atoms are paired = and your multiplicity is 1, you don=E2=80=99t need to define it. = Gaussian=E2=80=99s default is (r).


In = this example I will use PBE; It has worked well for me. You could use = another DFT functional or another flavor of PBE. I increased my SCF = cycle to 1000.

#P OPT FREQ LC-wPBE/GEN GEO=3Dconnectivity = SCF=3DMaxcycle=3D1000

Title
0 1

Geometry

Cu 0
cc-pVDZ
****
C O H = 0
6-311G(d,p)
****

Cu 0
cc-pVDZ
****


Also, If you have = multiple metals, they need to be described separately as shown above and = separated by a space and (****)
Gaussian will give = you an error message if it is not correct. You can reference = here https://gaussian.com/gen/ 


---
Uchenna A. Anene
Ph.D. Student, Chemical Engineering =  
University of = Connecticut
Innovation Partnership = Building 
159 Discovery = Drive
Storrs, CT = 06269
uche##uconn.edu
https://alpay.ims.uconn.edu/uche-anene/#
www.linkedin.com/in/uche-anene 
https://www.researchgate.net/profile/Uchenna_Anene









On Sep 5, 2019, at 2:25 AM, Susi Lehtola susi.lehtola/ahelsinki.fi <owner-chemistry##ccl.net> wrote:


Sent to CCL by: Susi Lehtola [susi.lehtola[-]helsinki.fi]
On = 9/4/19 8:46 PM, Nikhil Maroli scinikhil!=3D!gmail.com wrote:
Hi,
There are no studies available on this = structure ( the structure also not reported yet). The available = calculations were performed using the CP2K package with mixed Gaussian = and plane-wave basis set, using the molecularly optimized MOLOPT double = f-valence polarized (mDZVP) basis set implemented in CP2K code. Im not = asking for a basis set for my study. I have used B3LYP, CCSD and MP2 = function with different basis set to obtain an optimization geometry. I = ended up with different errors and the structure is broken at the end. I = came to know the possibility of freezing the coordinates, but at = the same time, there is suggestion that the freezing is unrealistic and = may not be accurate. So in a nutshell, information from anyone who had a = similar situation which is resolved would be helpful for me.
Thank you for the help

Dear Nikhil,


I'm = afraid the method you are running, ump2=3Dfull/lanl2mb, makes very = little sense due to several reasons.

First, = lanl2mb is a minimal basis set and as such gives unreliable results even = at Hartree-Fock (HF) and density functional level. You need to use a = bigger basis set, e.g. def2SVP.

Second, = post-Hartree-Fock methods like MP2 and CCSD require really large basis = sets to reach sufficient convergence: usually one would use at least = triple- or quadruple-zeta, and extrapolate to the complete basis set = limit. At lanl2mb level you just get garbage. If you can't handle big = basis sets, use DFT instead.

Third, I'm not = sure what the electronic structure of your perovskite is, but HF and = post-HF methods yield unreliable results for metallic systems due to the = small/vanishing band gap and significant spin contamination. You can see = whether this happens by running DFT with different fractions of exact = exchange, e.g. PBE vs PBE0.

As Dr Cuffe = already pointed out, the periodic and cluster structures will differ due = to surface effects. If I was you, I'd start with the methods and = geometry of the CP2K calculations. If you can reproduce the periodic = calculation, this gives you a good starting point for studying how using = a finite size cluster changes the results.
--
---------------------------------------------------------------= ---
Mr. Susi Lehtola, PhD =             Ju= nior Fellow, Adjunct Professor
susi.lehtola,,alumni.helsinki.fi =   University of Helsinki
http://susilehtola.github.io/ =     Finland
---------------------------------------------------------------= ---
Susi Lehtola, dosentti, FT =        tutkijatohtori
susi.lehtola,,alumni.helsinki.fi   Helsingin yliopisto
http://susilehtola.github.io/
---------------------------------------------------------------= ---



-=3D This = is automatically added to each message by the mailing script =3D-
To recover the email address of the author of the message, = please change
the strange characters on the top line to = the ## sign. You can also
look up the X-Original-From: line = in the mail header.

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

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

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

Search Messages: = http://www.ccl.net/chemistry/searchccl/index.shtml

    http://www.ccl.net/spammers.txt

RTFI: = http://www.ccl.net/chemistry/aboutccl/instructions/



= --Apple-Mail=_ED054F72-D563-47D9-931E-C22B6A0B0EC9-- From owner-chemistry@ccl.net Thu Sep 5 14:23:00 2019 From: "Uche Anene uche]^[uconn.edu" To: CCL Subject: CCL:G: Gaussian running with GPU Message-Id: <-53835-190905134741-32624-VhavjSE2E8Gz237koAgU+Q++server.ccl.net> X-Original-From: Uche Anene Content-Type: multipart/alternative; boundary="Apple-Mail=_EF38BE1E-700B-4669-9B87-7637ADAC3A3B" Date: Thu, 5 Sep 2019 13:47:32 -0400 Mime-Version: 1.0 (Mac OS X Mail 12.4 \(3445.104.11\)) Sent to CCL by: Uche Anene [uche^^uconn.edu] --Apple-Mail=_EF38BE1E-700B-4669-9B87-7637ADAC3A3B Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=us-ascii If you are using 36 CPUs (and that is in your input file) then it is = written=20 %CPU=3D0-35 Happy computing, Uche --- Uchenna A. Anene Ph.D. Student, Chemical Engineering =20 University of Connecticut Innovation Partnership Building=20 159 Discovery Drive Storrs, CT 06269 uche,+,uconn.edu https://alpay.ims.uconn.edu/uche-anene/# www.linkedin.com/in/uche-anene =20= https://www.researchgate.net/profile/Uchenna_Anene = > On Sep 5, 2019, at 6:03 AM, Alex Mathew alexmathewmd*o*gmail.com = wrote: >=20 >=20 > Sent to CCL by: "Alex Mathew" [alexmathewmd%a%gmail.com] > Dear all, >=20 > I m having a workstation with 3 GTX 1070 card and 2 x 12 core = processor=20 > (48 threads). Can anyone share me the keword/input file to use GPU+CPU = in=20 > Gaussian-16. > I have tried the following > %CPU=3D0-36 !to use 36 thread for processor > %GPUCPU=3D0,1,2=3D37,38,39 ! three gpu with id 0,1,2=20 >=20 > But it returned syntax error. >=20 >=20 >=20 > -=3D This is automatically added to each message by the mailing script = =3D- > To recover the email address of the author of the message, please = change>=20>=20>=20 > Subscribe/Unsubscribe:=20>=20>=20 > Job: http://www.ccl.net/jobs=20 > Conferences: = http://server.ccl.net/chemistry/announcements/conferences/ >=20>=20>=20>=20 >=20 --Apple-Mail=_EF38BE1E-700B-4669-9B87-7637ADAC3A3B Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=us-ascii If = you are using 36 CPUs (and that is in your input file) then it is = written 

%CPU=3D0-35

Happy computing,
Uche



---
Uchenna A. Anene
Ph.D. Student, Chemical Engineering =  
University of = Connecticut
Innovation Partnership = Building 
159 Discovery = Drive
Storrs, CT = 06269
uche,+,uconn.edu
https://alpay.ims.uconn.edu/uche-anene/#
www.linkedin.com/in/uche-anene 
https://www.researchgate.net/profile/Uchenna_Anene









On Sep 5, 2019, at 6:03 AM, Alex Mathew alexmathewmd*o*gmail.com <owner-chemistry,+,ccl.net> wrote:


Sent to CCL by: "Alex  Mathew" [alexmathewmd%a%gmail.com]
Dear = all,

I m having a workstation with 3 GTX = 1070 card and 2 x 12 core processor
(48 threads). Can = anyone share me the keword/input file to use GPU+CPU in
Gaussian-16.
I have tried the following
%CPU=3D0-36 !to use 36 thread for processor
%GPUCPU=3D0,1,2=3D37,38,39 ! three gpu with id 0,1,2

But it returned syntax error.



-=3D This is automatically = added to each message by the mailing script =3D-
To = recover the email address of the author of the message, please change
the strange characters on the top line to the ,+, sign. You can = also
look up the X-Original-From: line in the mail = header.

E-mail to subscribers: CHEMISTRY,+,ccl.net or = use:
     http://www.ccl.net/cgi-bin/ccl/send_ccl_message

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

=      http://www.ccl.net/chemistry/sub_unsub.shtml

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

Job: http://www.ccl.net/jobs
Conferences: http://server.ccl.net/chemistry/announcements/conferences/<= br class=3D"">
Search Messages: http://www.ccl.net/chemistry/searchccl/index.shtml

If your mail bounces from CCL with 5.7.1 = error, check:
     http://www.ccl.net/spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutccl/instructions/



= --Apple-Mail=_EF38BE1E-700B-4669-9B87-7637ADAC3A3B-- From owner-chemistry@ccl.net Thu Sep 5 17:58:01 2019 From: "Igors Mihailovs igorsm]=[cfi.lu.lv" To: CCL Subject: CCL:G: Gaussian running with GPU Message-Id: <-53836-190905124613-12887-AMUsl3SdjogSmVFfiFNoUg/./server.ccl.net> X-Original-From: Igors Mihailovs Content-Language: en-US Content-Type: multipart/alternative; boundary="------------03D331EE1AAFC5FDF46C0FC5" Date: Thu, 5 Sep 2019 19:46:02 +0300 MIME-Version: 1.0 Sent to CCL by: Igors Mihailovs [igorsm . cfi.lu.lv] This is a multi-part message in MIME format. --------------03D331EE1AAFC5FDF46C0FC5 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 7bit Dear Alex, From the manual on Gaussian website I can detect that the controlling CPUs listed in% GPUCPU should also appear in the value for the %CPU command. E.g., %CPU=0-39 !to use 36 thread for processor %GPUCPU=0,1,2=37,38,39 ! three gpu with id 0,1,2 Hope this helps. Igors On 9/5/19 1:03 PM, Alex Mathew alexmathewmd*o*gmail.com wrote: > Sent to CCL by: "Alex Mathew" [alexmathewmd%a%gmail.com] > Dear all, > > I m having a workstation with 3 GTX 1070 card and 2 x 12 core processor > (48 threads). Can anyone share me the keword/input file to use GPU+CPU in > Gaussian-16. > I have tried the following > %CPU=0-36 !to use 36 thread for processor > %GPUCPU=0,1,2=37,38,39 ! three gpu with id 0,1,2 > > But it returned syntax error.> > -- Yours faithfully, Igors Mihailovs Research assistant Laboratory of Organic Materials Institute of Solid State Physics University of Latvia --------------03D331EE1AAFC5FDF46C0FC5 Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: 7bit Dear Alex,

From the manual on Gaussian website I can detect that the controlling CPUs listed in% GPUCPU should also appear in the value for the %CPU command. E.g.,

%CPU=0-39 !to use 36 thread for processor
%GPUCPU=0,1,2=37,38,39 ! three gpu with id 0,1,2
Hope this helps.
Igors

On 9/5/19 1:03 PM, Alex Mathew alexmathewmd*o*gmail.com wrote:
Sent to CCL by: "Alex  Mathew" [alexmathewmd%a%gmail.com]
Dear all,

 I m having a workstation with 3 GTX 1070 card and 2 x 12 core processor 
(48 threads). Can anyone share me the keword/input file to use GPU+CPU in 
Gaussian-16.
 I have tried the following
%CPU=0-36 !to use 36 thread for processor
%GPUCPU=0,1,2=37,38,39 ! three gpu with id 0,1,2 

But it returned syntax error.E-mail to subscribers: CHEMISTRY:_:ccl.net or use:
      http://www.ccl.net/cgi-bin/ccl/send_ccl_message

E-mail to administrators: CHEMISTRY-REQUEST:_:ccl.net or use
      http://www.ccl.net/cgi-bin/ccl/send_ccl_messagehttp://www.ccl.net/chemistry/sub_unsub.shtml

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

Job: http://www.ccl.net/jobs 
Conferences: http://server.ccl.net/chemistry/announcements/conferences/

Search Messages: http://www.ccl.net/chemistry/searchccl/index.shtmlhttp://www.ccl.net/spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutccl/instructions/




-- 
Yours faithfully,
Igors Mihailovs
Research assistant
Laboratory of Organic Materials
Institute of Solid State Physics
University of Latvia
--------------03D331EE1AAFC5FDF46C0FC5-- From owner-chemistry@ccl.net Thu Sep 5 18:33:01 2019 From: "Robert Molt r.molt.chemical.physics*|*gmail.com" To: CCL Subject: CCL:G: Structure breaks even after started with correct bond length Message-Id: <-53837-190905163056-8884-/Ru5lG+YaTcSchHHS50Vdw++server.ccl.net> X-Original-From: Robert Molt Content-Language: en-US Content-Type: multipart/alternative; boundary="------------D4E374972232BC09EE94C99F" Date: Thu, 5 Sep 2019 16:30:47 -0400 MIME-Version: 1.0 Sent to CCL by: Robert Molt [r.molt.chemical.physics()gmail.com] This is a multi-part message in MIME format. --------------D4E374972232BC09EE94C99F Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 8bit Dr. Lehtola: Given that the user's question is what has gone wrong with the /geometry/, effectively, isn't it somewhat inappropriate to say that post HF methods require basis sets of TZ or larger? I believe that post-HF methods get perfectly respectable geometries using DZ basis sets; I've never encountered a time that the optimized geometry was not reasonable for a single-reference molecule. I certainly agree that the energy may require TZ or more, but if the molecule breaks up, it speaks to a larger problem with the user's starting geometry or "single zeta" basis. On 9/5/19 2:25 AM, Susi Lehtola susi.lehtola/ahelsinki.fi wrote: > > Sent to CCL by: Susi Lehtola [susi.lehtola[-]helsinki.fi] > On 9/4/19 8:46 PM, Nikhil Maroli scinikhil!=!gmail.com wrote: >> Hi, >> There are no studies available on this structure ( the structure also >> not reported yet). The available calculations were performed using >> the CP2K package with mixed Gaussian and plane-wave basis set, using >> the molecularly optimized MOLOPT double f-valence polarized (mDZVP) >> basis set implemented in CP2K code. Im not asking for a basis set for >> my study. I have used B3LYP, CCSD and MP2 function with different >> basis set to obtain an optimization geometry. I ended up with >> different errors and the structure is broken at the end. I came to >> know the possibility of freezing the coordinates, but at the same >> time, there is suggestion that the freezing is unrealistic and may >> not be accurate. So in a nutshell, information from anyone who had a >> similar situation which is resolved would be helpful for me. >> Thank you for the help > > Dear Nikhil, > > > I'm afraid the method you are running, ump2=full/lanl2mb, makes very > little sense due to several reasons. > > First, lanl2mb is a minimal basis set and as such gives unreliable > results even at Hartree-Fock (HF) and density functional level. You > need to use a bigger basis set, e.g. def2SVP. > > Second, post-Hartree-Fock methods like MP2 and CCSD require really > large basis sets to reach sufficient convergence: usually one would > use at least triple- or quadruple-zeta, and extrapolate to the > complete basis set limit. At lanl2mb level you just get garbage. If > you can't handle big basis sets, use DFT instead. > > Third, I'm not sure what the electronic structure of your perovskite > is, but HF and post-HF methods yield unreliable results for metallic > systems due to the small/vanishing band gap and significant spin > contamination. You can see whether this happens by running DFT with > different fractions of exact exchange, e.g. PBE vs PBE0. > > As Dr Cuffe already pointed out, the periodic and cluster structures > will differ due to surface effects. If I was you, I'd start with the > methods and geometry of the CP2K calculations. If you can reproduce > the periodic calculation, this gives you a good starting point for > studying how using a finite size cluster changes the results. -- Dr. Robert Molt Jr. r.molt.chemical.physics..gmail.com --------------D4E374972232BC09EE94C99F Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: 8bit

Dr. Lehtola:

Given that the user's question is what has gone wrong with the geometry, effectively, isn't it somewhat inappropriate to say that post HF methods require basis sets of TZ or larger? I believe that post-HF methods get perfectly respectable geometries using DZ basis sets; I've never encountered a time that the optimized geometry was not reasonable for a single-reference molecule. I certainly agree that the energy may require TZ or more, but if the molecule breaks up, it speaks to a larger problem with the user's starting geometry or "single zeta" basis.

On 9/5/19 2:25 AM, Susi Lehtola susi.lehtola/ahelsinki.fi wrote:

Sent to CCL by: Susi Lehtola [susi.lehtola[-]helsinki.fi]
On 9/4/19 8:46 PM, Nikhil Maroli scinikhil!=!gmail.com wrote:
Hi,
There are no studies available on this structure ( the structure also not reported yet). The available calculations were performed using the CP2K package with mixed Gaussian and plane-wave basis set, using the molecularly optimized MOLOPT double f-valence polarized (mDZVP) basis set implemented in CP2K code. Im not asking for a basis set for my study. I have used B3LYP, CCSD and MP2 function with different basis set to obtain an optimization geometry. I ended up with different errors and the structure is broken at the end. I came to know the possibility of freezing the coordinates, but at the same time, there is suggestion that the freezing is unrealistic and may not be accurate. So in a nutshell, information from anyone who had a similar situation which is resolved would be helpful for me.
Thank you for the help

Dear Nikhil,


I'm afraid the method you are running, ump2=full/lanl2mb, makes very little sense due to several reasons.

First, lanl2mb is a minimal basis set and as such gives unreliable results even at Hartree-Fock (HF) and density functional level. You need to use a bigger basis set, e.g. def2SVP.

Second, post-Hartree-Fock methods like MP2 and CCSD require really large basis sets to reach sufficient convergence: usually one would use at least triple- or quadruple-zeta, and extrapolate to the complete basis set limit. At lanl2mb level you just get garbage. If you can't handle big basis sets, use DFT instead.

Third, I'm not sure what the electronic structure of your perovskite is, but HF and post-HF methods yield unreliable results for metallic systems due to the small/vanishing band gap and significant spin contamination. You can see whether this happens by running DFT with different fractions of exact exchange, e.g. PBE vs PBE0.

As Dr Cuffe already pointed out, the periodic and cluster structures will differ due to surface effects. If I was you, I'd start with the methods and geometry of the CP2K calculations. If you can reproduce the periodic calculation, this gives you a good starting point for studying how using a finite size cluster changes the results. 
-- 
Dr. Robert Molt Jr.
r.molt.chemical.physics..gmail.com
--------------D4E374972232BC09EE94C99F--