From owner-chemistry@ccl.net Tue Dec 25 15:39:00 2012 From: "J D Whitfield jdwhitfield%a%gmail.com" To: CCL Subject: CCL:G: Unrestricted CASSCF Message-Id: <-48028-121225081456-6388-DyQ+qUXx9iTi2ZRjplx9hw%x%server.ccl.net> X-Original-From: J D Whitfield Content-Type: multipart/alternative; boundary=f46d043c7eb0df0b4604d1ad1891 Date: Tue, 25 Dec 2012 14:14:43 +0100 MIME-Version: 1.0 Sent to CCL by: J D Whitfield [jdwhitfield(!)gmail.com] --f46d043c7eb0df0b4604d1ad1891 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Hi Yavuz et al, I appreciate (and throughly understand) that the wave function is an eigenstate of the S^2 and S_z operators (by construction). I was interested in the natural orbitals which do indeed have different occupations for the alpha and beta components. I suppose that you and Nuno are correct in that there may be no need to explicitly use different orbitals for the alpha and beta components since one does the summation over all configurations of the appropriate symmetry and the one-particle basis won't matter so much. Anyway, I figured out a path to my original goal (the natural occupation numbers for all 10 spin orbitals). Since the output of gaussian provided the configurations and their coefficients in the wave function, I can form the wave function, the corresponding 1-RDM, and the occupation numbers (e.g. the eigenvalues of the 1-RDM). I asked the CCL list as I was hoping there was an option in g09 to compute this for me but nothing seemed to work as Gaussian only printed the natural occupation numbers of the spatial orbitals. The (matlab/octave) scripts I used (for future generations of quantum chemists) were the following (released under GNU GPL): "z4.m" %Script was auto-generate by converter.py from file "Z4.log" from % a CASSCF g09 calculation (tantamount to FCI for this model system= ) % % In the Z4.log file: the determinants were identified by % "Configuration 1 Symmetry 1 1aa00" % which was expanded as 11010100 which becames [1,2,4,6] below % % Elsewhere in the log file the coefficients are printed % "( 1) 0.9996351" % which were parsed to construct the vector $D$ below. % The function "ICbo" provides a linear indexing of the determinants % and function "formY4" returns the full wave function from which % the eigenvalues of the 1-RMD are computed. % % VCQ, 1-RDM project % JDWhitfield 2012 M=3D10; N=3D4; D=3Dzeros(nchoosek(10,4),1); D(ICbo([1, 7, 8, 9],10))=3D-9.9e-06; D(ICbo([5, 7, 8, 9],10))=3D2.13e-05; D(ICbo([1, 5, 9, 10],10))=3D-0.0008719; D(ICbo([3, 7, 9, 10],10))=3D7e-07; D(ICbo([3, 5, 6, 7],10))=3D0.0009427; D(ICbo([1, 4, 7, 9],10))=3D-6.1e-06; D(ICbo([1, 5, 6, 9],10))=3D3.31e-05; D(ICbo([1, 5, 7, 8],10))=3D-0.0075263; D(ICbo([1, 3, 7, 8],10))=3D-0.000956; D(ICbo([1, 3, 5, 10],10))=3D0.0006215; D(ICbo([2, 3, 7, 9],10))=3D-0.0001591; D(ICbo([3, 4, 5, 9],10))=3D-0.0010242; D(ICbo([3, 5, 9, 10],10))=3D-0.0015401; D(ICbo([1, 7, 9, 10],10))=3D-1.9e-06; D(ICbo([5, 6, 7, 9],10))=3D-1.8e-06; D(ICbo([3, 7, 8, 9],10))=3D5e-06; D(ICbo([1, 3, 8, 9],10))=3D-1.98e-05; D(ICbo([1, 3, 7, 10],10))=3D-0.0001729; D(ICbo([3, 5, 8, 9],10))=3D0.0005948; D(ICbo([3, 5, 7, 10],10))=3D0.0005788; D(ICbo([1, 2, 3, 5],10))=3D0.9996351; D(ICbo([1, 3, 4, 5],10))=3D-0.0002429; D(ICbo([1, 2, 3, 7],10))=3D2.67e-05; D(ICbo([1, 2, 3, 9],10))=3D4e-07; D(ICbo([1, 2, 5, 7],10))=3D0.0002732; D(ICbo([1, 3, 5, 6],10))=3D3.04e-05; D(ICbo([1, 3, 4, 7],10))=3D-0.0006364; D(ICbo([1, 2, 5, 9],10))=3D4.8e-06; D(ICbo([2, 3, 5, 7],10))=3D-0.0023682; D(ICbo([3, 6, 7, 9],10))=3D2e-07; D(ICbo([4, 5, 7, 9],10))=3D-1.58e-05; D(ICbo([1, 3, 4, 9],10))=3D-0.000183; D(ICbo([1, 4, 5, 7],10))=3D-0.0046893; D(ICbo([1, 3, 5, 8],10))=3D0.002368; D(ICbo([1, 3, 6, 7],10))=3D4.69e-05; D(ICbo([3, 4, 5, 7],10))=3D-0.0095676; D(ICbo([2, 3, 5, 9],10))=3D-0.0006211; D(ICbo([1, 4, 5, 9],10))=3D-0.0012404; D(ICbo([1, 2, 7, 9],10))=3D-0.0002232; D(ICbo([1, 3, 6, 9],10))=3D2.2e-06; D(ICbo([1, 5, 6, 7],10))=3D0.0003697; D(ICbo([3, 4, 7, 9],10))=3D-1.9e-06; D(ICbo([2, 5, 7, 9],10))=3D-0.0010183; D(ICbo([1, 3, 9, 10],10))=3D-0.0001246; D(ICbo([1, 5, 7, 10],10))=3D-0.0011062; D(ICbo([1, 5, 8, 9],10))=3D-8.74e-05; D(ICbo([1, 6, 7, 9],10))=3D5e-07; D(ICbo([3, 5, 7, 8],10))=3D-0.023154; D(ICbo([3, 5, 6, 9],10))=3D6.17e-05; D(ICbo([5, 7, 9, 10],10))=3D2.5e-06; PsiFCI=3DformY4(D,10); T=3Dreshape(PsiFCI,[M M^(N-1)]); [u,s,v]=3Dsvd(T); format long %these are the 1-RDM eigenvalue normalized to N=3D4 electrons disp(diag(u'*(T*T')*u)*4) "formY4.m" function Psi=3DformY4(D,M) %FormY4 computes the wave function for wave function of four coordinates %The inputs are the determinant coefficients as a vector, D, and the number %of orbitals involved % %VCQ, 1-RDM project %JDWhitfield 2012 Psi=3Dzeros(M^4,1); dimvec=3Dones(1,4)*M; Psi=3Dreshape(Psi,dimvec); %normalization nn=3Dsqrt(factorial(4)); for i1=3D1:M for i2=3D1:M if(i1=3D=3Di2) continue end for i3=3D1:M if(i1=3D=3Di3 || i2=3D=3Di3) continue end for i4=3D1:M if(i1=3D=3Di4 || i2=3D=3Di4 || i3=3D=3Di4) continue end %trick for sign of permutation %from Cleve Molar posted %on matlab central (Apr/16/1998) [~,p]=3Dsort([i1,i2,i3,i4]); t =3D speye(length(p)); sgn=3Ddet(t(:,p)); Psi(i1,i2,i3,i4)=3Dsgn*D(ICbo([i1,i2,i3,i4],M))/nn; end end end end end "ICbo.m" function k=3DICbo(combo,M) %ICBO returns the combo index %the inputs are the combo and the number of basis funcs % %based on stackoverflow.com/questions/13730772 by Origin % %VCQ, 1-RDM project %JDWhitfield 2012 %sort combination combo=3Dsort(combo); %combo=3Dcombo(end:-1:1); N=3Dlength(combo); if(length(unique(combo))~=3DN) k=3D-1; return end l=3D0; %k is the index k=3D1; for i=3D1:N while(true) l=3Dl+1; C=3Dnchoosek(M-l,N-i); if(combo(i)=3D=3Dl) break; else k=3Dk+C; end end end -- Dr. J. D. Whitfield VCQ Postdoctoral Fellow Vienna Center for Quantum Science and Technology email: james.whitfield+/-univie.ac.at web: homepage.univie.ac.at/james.whitfield On Tue, Dec 25, 2012 at 12:09 AM, Yavuz Dede wrote: Hi, Quoting: "..there is no intrinsic reason why one cannot optimize alpha and beta spin orbitals of the active space separately..." As Nuno pointed out CASSCF WF i.e. the multi determinantal (or configurational) expansion as a whole is spin adapted. So zero spin contamination. This makes the WF a proper eigenfunction of both Sz and S2 operators. If you select different spatial orbitals for alpha and beta you will have a non-zero spin contamination, hence contrary to the spirit of the method. Of course you can start form any orbital set, in that case natural orbitals > from an unrestricted calculation can be used, however they are still a single set of spatial orbitals for alpha and beta pairs. In short (U) in CASSCF does not exist, unless it is someone's misnomer for some mathematical trick in a certain code, however I've never heard of something like that. Suggested reading should involve Mark Gordon's, Ruedenberg's and late Roos'= . Best, Yavuz --=20 ||||||||||||||||||||||| Yavuz Dede G=DC-Fen Fak=FCltesi Kimya B=F6l=FCm=FC L-128 06500 Ankara Tel: 312-202-1386 Faks: 312-212-2279 |||||||||||||||||||| Yavuz Dede, Ph.D. Theoretical/Computational Chemistry Gazi University Faculty of Science Department of Chemistry 06500 Ankara TURKEY http://w3.gazi.edu.tr/~dede/ydd.htm |||||||||||||||||||||||||||||||||||| On 21.12.2012 11:41, J D Whitfield jdwhitfield-$-gmail.com wrote: First thanks for the feedback so far. More specific information about my study: I am considering a small CASSCF(4,5) calculation for the triplet state of Be in the TZ Dunning basis. This amounts to FCI for this toy model. I am interested in the natural orbital occupation numbers, but, in this case, the restricted orbitals are not sufficient as the alpha and beta spin orbitals should each have different energy. To Sam, My input script was the following: %Chk=3DCAS1 #P UCASSCF(4,5)/Gen POP=3DNOAB Density=3DAll Z=3D4, TZ, FCI 0 3 4 0.0 0.0 0.0 ! TZ (Dunning) EMSL Basis Set Exchange Library ... The output was: QPErr --- A syntax error was detected in the input line. #P UCASSCF(4,5)/Gen POP=3DNOA Density=3DAll ' Last state=3D"GCL" TCursr=3D 1046 LCursr=3D 3 To Ljiljana, would using the UHF orbitals be sufficient to obtain a solution where the alpha and beta orbitals are optimized separately. To Nuno, thank for the suggestion but please refer to a specific reference that would discuss the implementation details mundane to CASSCF. I can believe that CASSCF is not implemented in Gaussian using an unrestricted formalism but there is no intrinsic reason why one cannot optimize alpha and beta spin orbitals of the active space separately. Thanks, JDW -- Dr. J. D. Whitfield VCQ Postdoctoral Fellow Vienna Center for Quantum Science and Technology email: james.whitfield _ univie.ac.at web: homepage.univie.ac.at/james.whitfield On Thu, Dec 20, 2012 at 10:52 PM, Nuno A. G. Bandeira nuno.bandeira[] ist.utl.pt wrote: Sent to CCL by: "Nuno A. G. Bandeira" [nuno.bandeira**ist.utl.pt] On 20/12/2012 14:57, J D Whitfield jdwhitfield=3Dgmail.com wrote: Hello CCL'ers I'm trying to run an unrestricted casscf calculation in gaussian 09 but it doesn't seem to accept the UCASSCF keyword. Any suggestions? Yes. Read a book on quantum chemistry. There is no such thing as a UCASSCF. CASSCF is a strictly spin restricted formalism. Best wishes, Nuno --=20 Nuno A. G. Bandeira, AMRSC C8 - Centro de Qu=EDmica e Bioqu=EDmica FCUL, Campo Grande Lisbon 1749-016 PORTUGAL http://www.researcherid.com/rid/B-6399-2012 http://pt.linkedin.com/pub/nuno-a-g-bandeira/47/55a/2aa -- -=3D This is automatically added to each message by the mailing script =3D- E-mail to subscribers: CHEMISTRY _ ccl.net or use:E-mail to administrators: CHEMISTRY-REQUEST _ ccl.net or usehttp://www.ccl.net/chemistry/sub_unsub.shtmlhttp://www.ccl.net/spammers.txtBu elektronik posta ve ekleri sadece adreste belirtilen kisi veya kurulusun kullanimi i=E7in g=F6nderilmektedir. Bu ileti tarafiniza yanlislikla ulasir= sa, l=FCtfen g=F6nderen kisiyi bilgilendiriniz ve iletiyi sisteminizden siliniz= . iletide ve eklerinde yer alan bilgilerin her ne sekilde olursa olsun =FC=E7= =FCnc=FC kisiler ile paylasilmasi hukuki ve cezai sorumluluk dogurabilir. Gazi =DCniversitesi'nin, bu ileti ve eklerinin i=E7erigi ve yayimi ile ilgili hi= =E7bir sorumlulugu bulunmamaktadir. This email and the attachments are sent to the individual or entity defined in the address field only. If you are not the intended recipient or have received the message in error, please notify the sender and remove the message from your system immediately. Sharing the information in the message or the attachments with the 3rd parties may cause legal rules and penalties to apply. Gazi University has no responsibility on the submission of this message and the attachments. --f46d043c7eb0df0b4604d1ad1891 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable
Hi Y= avuz et al,

I appreciate (and throughly understand) that the wave function is an = eigenstate of the S^2 and S_z operators (by construction). =A0I was interes= ted in the natural orbitals which do indeed have different occupations for = the alpha and beta components. =A0I suppose that you and Nuno are correct i= n that there may be no need to explicitly use different orbitals for the al= pha and beta components since one does the summation over all configuration= s of the=A0appropriate symmetry and the one-particle basis won't matter= so much.=A0

Anyway, I f= igured out a path to my original goal (the natural occupation numbers for a= ll 10 spin orbitals). =A0Since the output of gaussian provided the configur= ations and their coefficients in the wave function, I can form the wave fun= ction, the corresponding 1-RDM, and the occupation numbers (e.g. the eigenv= alues of the 1-RDM). =A0 I asked the CCL list as I was hoping there was an = option in g09 to compute this for me but nothing seemed to work as Gaussian= only printed the natural occupation numbers of the spatial orbitals.=A0

The (matlab= /octave) scripts I used (for future generations of quantum chemists) were t= he following (released under GNU GPL):

=
"z4.m"
%Script w= as auto-generate by converter.py from file "Z4.log" from=A0
=
=A0 =A0 =A0 =A0 % a CASSCF g09 calculation (tantamount=A0to FCI for this mo= del system)
%
% In the Z4.log f= ile: the determinants were identified by
% =A0 =A0"Configuration =A0 =A0 =A0 =A0 1 Symmetry 1 = 1aa00"
% which was expanded as 11010100 which be= cames [1,2,4,6] below
%=A0
% Elsewhere in the log file the= coefficients are printed
% =A0 =A0"( =A0 =A01) 0.9996351"
= % which were parsed to construct the vector $D$ below.
% The function "ICbo" provides a linear indexing= of the determinants
% and function "formY4"= ; returns the full wave function from which=A0
% the eigenvalues of the 1-RMD are computed.=A0
%=A0
% =A0VCQ, 1-RDM project
= =A0 =A0 =A0 =A0 % JDWhitfield 2012
M=3D10;
N=3D4;
D=3Dzeros(nchoosek(10,4),1);<= /div>
D(ICbo([1,= 7, 8, 9],10))=3D-9.9e-06;
D(ICbo([5, 7, 8, 9],10))= =3D2.13e-05;
D(ICbo([1, 5, 9, 10],10))=3D-0.0008719;
D(ICbo([3, 7, 9, 10],10))=3D7e-07;
D(ICbo([3, 5, 6, 7],10))=3D0.0009427;
D(ICbo([1, 4, 7, 9],10))=3D-6.1e-06;
D(ICbo([1, 5, 6, 9],10))=3D3.31e-05;
D(ICbo([1, 5, 7, 8],10))=3D-0.0075263;
D(ICbo([1, 3, 7, 8],10))=3D-0.000956;
D(ICbo([1, 3, 5, 10],10))=3D0.0006215;
D(ICbo([2, 3, 7, 9],10))=3D-0.0001591;
D(ICbo([3, 4, 5, 9],10))=3D-0.0010242;
D(ICbo([3, 5, 9, 10],10))=3D-0.0015401;
D(ICbo([1, 7, 9, 10],10))=3D-1.9e-06;
D(ICbo([5, 6, 7, 9],10))=3D-1.8e-06;
D(ICbo([3, 7, 8, 9],10))=3D5e-06;
D(ICbo([1, 3, 8, 9],10))=3D-1.98e-05;
D(ICbo([1, 3, 7, 10],10))=3D-0.0001729;
D(ICbo([3, 5, 8, 9],10))=3D0.0005948;
D(ICbo([3, 5, 7, 10],10))=3D0.0005788;
D(ICbo([1, 2, 3, 5],10))=3D0.9996351;
D(ICbo([1, 3, 4, 5],10))=3D-0.0002429;
D(ICbo([1, 2, 3, 7],10))=3D2.67e-05;
D(ICbo([1, 2, 3, 9],10))=3D4e-07;
D(ICbo([1, 2, 5, 7],10))=3D0.0002732;
D(ICbo([1, 3, 5, 6],10))=3D3.04e-05;
D(ICbo([1, 3, 4, 7],10))=3D-0.0006364;
D(ICbo([1, 2, 5, 9],10))=3D4.8e-06;
D(ICbo([2, 3, 5, 7],10))=3D-0.0023682;
D(ICbo([3, 6, 7, 9],10))=3D2e-07;
D(ICbo([4, 5, 7, 9],10))=3D-1.58e-05;
D(ICbo([1, 3, 4, 9],10))=3D-0.000183;
D(ICbo([1, 4, 5, 7],10))=3D-0.0046893;
D(ICbo([1, 3, 5, 8],10))=3D0.002368;
D(ICbo([1, 3, 6, 7],10))=3D4.69e-05;
D(ICbo([3, 4, 5, 7],10))=3D-0.0095676;
D(ICbo([2, 3, 5, 9],10))=3D-0.0006211;
D(ICbo([1, 4, 5, 9],10))=3D-0.0012404;
D(ICbo([1, 2, 7, 9],10))=3D-0.0002232;
D(ICbo([1, 3, 6, 9],10))=3D2.2e-06;
D(ICbo([1, 5, 6, 7],10))=3D0.0003697;
D(ICbo([3, 4, 7, 9],10))=3D-1.9e-06;
D(ICbo([2, 5, 7, 9],10))=3D-0.0010183;
D(ICbo([1, 3, 9, 10],10))=3D-0.0001246;
D(ICbo([1, 5, 7, 10],10))=3D-0.0011062;
D(ICbo([1, 5, 8, 9],10))=3D-8.74e-05;
D(ICbo([1, 6, 7, 9],10))=3D5e-07;
D(= ICbo([3, 5, 7, 8],10))=3D-0.023154;
D(ICbo([3, 5, 6, 9],10))=3D6.17e-05;
D(ICbo([5, 7, 9, 10],10))=3D2.5e-06;
PsiFCI=3DformY4(D,10);
T=3Dreshape(P= siFCI,[M M^(N-1)]);
[u,s,v]=3Dsvd(T);
format long
<= div class=3D"gmail_extra"> %these are= the 1-RDM eigenvalue normalized to N=3D4 electrons
d= isp(diag(u'*(T*T')*u)*4)

"formY= 4.m"
function Psi=3DformY4(D,M)
%FormY4 co= mputes the wave function for wave function of four coordinates
%The inputs are the determinant coefficients as a vector, D, and = the number
%of orbitals involved
%
%VCQ, 1-RD= M project
%JDWhitfield 2012
Psi= =3Dzeros(M^4,1);
dimvec=3Dones(1,4)*M;
Psi=3Dreshape(= Psi,dimvec);
%normalization
nn=3Dsqrt(factorial(4= ));
for i1=3D1= :M
=A0 =A0for i2=3D1:M
=A0 = =A0 =A0 =A0if(i1=3D=3Di2)
=A0 =A0 =A0 =A0 =A0 continue
=A0 = =A0 =A0 =A0end
=A0 =A0 = =A0 =A0for i3=3D1:M =A0 =A0=A0
=A0 =A0 =A0 =A0 =A0 = =A0if(i1=3D=3Di3 || i2=3D=3Di3)
=A0 =A0 = =A0 =A0 =A0 =A0 =A0 continue
=A0 =A0 =A0 =A0 =A0 =A0= end
=A0 =A0 = =A0 =A0 =A0 =A0for i4=3D1:M =A0 =A0=A0
=A0 =A0 =A0 = =A0 =A0 =A0 =A0 =A0if(i1=3D=3Di4 || i2=3D=3Di4 || i3=3D=3Di4)
=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0continue
=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0end
=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0%trick for sign of permuta= tion
=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0%from Cleve Mo= lar posted=A0
=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0%on matlab central (Apr/16= /1998)
=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0[~,p]=3Dsort([= i1,i2,i3,i4]);
=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0t =3D speye(length(p));
=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0sgn=3Ddet(t(:,p));
=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0Psi(i1,i2,i3,i4)=3Dsgn*D(I= Cbo([i1,i2,i3,i4],M))/nn;
=A0 =A0 =A0 =A0 =A0 =A0end=
=A0 =A0 =A0 =A0end
=A0 =A0end
end
<= div class=3D"gmail_extra"> end=A0

"ICbo.m"
function k=3DICbo(combo,M)
%ICBO returns the combo index
%the inputs are the combo and the number of basis funcs
%
%based on stackoverflow.com/questions/13730772 by Origin
%
%VCQ, 1-RDM project
%JDWhitfield 2012
%sort combination
combo=3Dso= rt(combo);
%combo=3Dcombo(end:-1:1);
N=3Dlength= (combo);
if(length(unique(combo))~=3DN)
=A0 =A0k= =3D-1;
=A0 =A0return
end
l=3D0;
%k is the index
k=3D1;
for i=3D1:N =A0 =A0
=A0 =A0while(= true)
=A0 =A0 =A0 =A0l=3Dl+1;
<= span class=3D"Apple-tab-span" style=3D"white-space:pre"> =A0 =A0 = =A0 =A0C=3Dnchoosek(M-l,N-i);
=A0 =A0 =A0 =A0if(combo(i)=3D=3Dl) =A0 =A0 =A0 =A0 =A0 = =A0
=A0 =A0 =A0 =A0 =A0 =A0break;
=A0 =A0 =A0 =A0else
=A0 =A0 =A0 = =A0 =A0 =A0k=3Dk+C; =A0 =A0 =A0 =A0
=A0 =A0 =A0 =A0end =A0 =A0 =A0 =A0 =A0 =A0
= =A0 =A0end
end
<= div class=3D"gmail_extra">
--

Dr. J. D. Whitf= ield
VCQ Postdoctoral Fellow
Vienna Center for Quantum Science and Technology

email: james.whitfield+/-univie.ac.at
web: homepage.univie.ac.at/james.whitfield




O= n Tue, Dec 25, 2012 at 12:09 AM, Yavuz Dede <dede+/-gazi.edu.tr> wrote:
Hi= ,

Quoting: &q= uot;..there is no intrinsic reason why one cannot optimize alpha and beta s= pin orbitals of the active space separately..."

As Nuno pointed out CASSCF WF i.e. the= multi determinantal (or configurational) expansion as a whole is spin adap= ted. So zero spin contamination. This makes the WF a proper eigenfunction o= f both Sz and S2 operators.
If you select different spatial orbitals for alp= ha and beta you will have a non-zero spin contamination, hence contrary to = the spirit of the method.
Of course you can= start form any orbital set, in that case natural orbitals from an unrestri= cted calculation can be used, however they are still a single set of spatia= l orbitals for alpha and beta pairs.
In short (U) in CASSCF does not exist, unless = =A0it is someone's misnomer for some mathematical trick in a certain co= de, however I've never heard of something like that.

Suggested reading should involve Mark = Gordon's, Ruedenberg's and late Roos'.

Best,
Yavuz

=
--=A0
= |||||||||||||||||||||||
Yavuz Dede
G=DC-Fen Fak=FCltesi
Kimya B=F6l=FCm=FC
L-128 06500 Ankara
Tel: =A0312-202-1386
Faks: 312-212-227= 9
||||||||||||||||||||
Yavuz Dede, Ph.D.
Theoretical/Computational Chemistry
Gazi University
Facult= y of Science
Department of Chemistry
06500 Ankara TURKEY
http://w3.gazi.edu.tr/~dede/ydd.htm
||||||||||||||||||||||||||||||||||||



On 21.12.2012 11:41, J D Whitfield jdwhitfiel= d-$-gmail.com wrote:
First thanks for the feedback so far. =A0More sp= ecific information about my study: I am considering a small CASSCF(4,5) cal= culation for the triplet state of Be in the TZ Dunning basis. =A0This amoun= ts to FCI for this toy model. =A0I am interested in the natural orbital occ= upation numbers, but, in this case, the restricted orbitals are not suffici= ent as the alpha and beta spin orbitals should each have different energy. = =A0

To Sam, My = input script was the following:

<= div class=3D"gmail_extra">%Chk=3DCAS1
#P UC= ASSCF(4,5)/Gen POP=3DNOAB Density=3DAll
Z=3D4, TZ, FCI=A0
=A0
0 3=A0
=A0
4 0.0 0.0 0.0=A0
=A0
=A0! TZ (Dunning) =A0EMSL =A0Basis Set Exchange Library
...
=A0
=A0
The output was:

=A0QPErr --- A syntax error was detect= ed in the input line.
=A0#P UCASSCF(4,5)/Ge= n POP=3DNOA Density=3DAll
=A0 =A0 '
=A0Last state=3D"GCL"
=A0TCursr= =3D 1046 LCursr=3D =A0 =A03

To Ljiljana, would using the
=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 UHF orbitals be sufficient to o= btain a solution where
=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 the alpha an= d beta orbitals are optimized separately.
<= br>
To Nuno, thank for the suggestion but p= lease refer to a specific reference that would discuss the implementation d= etails mundane to CASSCF. =A0I can believe that CASSCF is not implemented i= n Gaussian using an unrestricted formalism but there is no intrinsic reason= why one cannot optimize alpha and beta spin orbitals of the active space s= eparately.=A0

Thanks,
JDW

=


<= div class=3D"gmail_extra">
=A0
--

<= /div>
Dr. J. D. Whitfield
VCQ Postdoctoral Fellow
Vienna Center for Quantum Science and Technology=

email= : james.whitfield _ univie.ac.at
<= div class=3D"gmail_extra"> web: homepage.univ= ie.ac.at/james.whitfield




On Thu, Dec 20, 2012 at 10:52 PM, Nuno= A. G. Bandeira nuno.bandeira[]ist.utl.pt= <owner-chemistry _ ccl.net> wrote:

Sent to CCL= by: "Nuno A. G. Bandeira" [nuno.bandeira**ist.utl.pt]

On 20/12/2012 14:57, J D Whitfield jdwhitfield=3Dgmail.com wrote:
Hello CCL'ers

I'm= trying to run an unrestricted casscf calculation in gaussian 09 but it doe= sn't seem to accept the UCASSCF keyword. Any suggestions?


<= div class=3D"gmail_extra">
Yes. Read a = book on quantum chemistry. There is no such thing as a UCASSCF.
CASSCF is a strictly spin restricted formalism.

Best wishes,
Nuno

--=A0
Nuno A. G. Bandeira, AMRSC
C8 - Centro de Qu=EDmica e Bioqu=EDmica
FCUL, Campo Grande
Lisbo= n 1749-016
PORTUGAL
http://www.researcherid.co= m/rid/B-6399-2012
http://pt.linkedin.com/pub/n= uno-a-g-bandeira/47/55a/2aa
--



-=3D This is automatically added to each message by = the mailing script =3D-=A0
E-mail to subscribers: CHEMISTRY _ ccl.net or use:
=A0 =A0 = =A0http://www.c= cl.net/cgi-bin/ccl/send_ccl_message

E-mail to a= dministrators: CHEMISTRY-REQUEST _ ccl.net o= r use
=A0 =A0 =A0http://www.ccl.net/cgi-bin/ccl/send_ccl_m= essage
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://s= erver.ccl.net/chemistry/announcements/conferences/

Search Mess= ages: http:/= /www.ccl.net/chemistry/searchccl/index.shtml
=A0 =A0 =A0http://www.ccl.net/s= pammers.txt

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


<= div class=3D"gmail_extra">


Bu elektronik= posta ve ekleri sadece adreste belirtilen kisi veya kurulusun kullanimi i= =E7in g=F6nderilmektedir. Bu ileti tarafiniza yanlislikla ulasirsa, l=FCtfe= n g=F6nderen kisiyi bilgilendiriniz ve iletiyi sisteminizden siliniz. ileti= de ve eklerinde yer alan bilgilerin her ne sekilde olursa olsun =FC=E7=FCnc= =FC kisiler ile paylasilmasi hukuki ve cezai sorumluluk dogurabilir. Gazi = =DCniversitesi'nin, bu ileti ve eklerinin i=E7erigi ve yayimi ile ilgil= i hi=E7bir sorumlulugu bulunmamaktadir.

This email = and the attachments are sent to the individual or entity defined in the add= ress field only. If you are not the intended recipient or have received the= message in error, please notify the sender and remove the message from you= r system immediately. Sharing the information in the message or the attachm= ents with the 3rd parties may cause legal rules and penalties to apply. Gaz= i University has no responsibility on the submission of this message and th= e attachments.


<= /div>
--f46d043c7eb0df0b4604d1ad1891--