========================================================================== ========================================================================== CASDILOC CASDILOC CASDILOC CASDILOC CI Program for localized orbitals. Long-range interactions between occupied and virtual orbtals are neglected ========================================================================== ========================================================================== Input files: Files created by ijkloc: ~ijcllocinfo, ~ijclloc2, ~ijclloc3, ~ijclloc4 ~Monoloc, ~Infoloc (identical to ~Mono and ~Info of molcost) ~LOinfo (where ~ means prefix...) ORB_IJKLOC eventually: topo: links between orbitals Data: &locci Compulsory data: prefix= Facultative data: noac= -1 number of active orbitals if noac=-1, noac,numac are recalculated using information from preceding programs. numac= 16*0 active orbitals ms2= 0 Sz*2 nessai= 0 trial vectors 0: on the lowest value(s) of the diagonal 1: given by the user. give &locess namelist(s) 2: read from previous calculation 3: CAS prediagonalization nvec 1 number of vectors dim_heff 50 maximal size of the effective hamiltonian, i.e. which corresponds to the number of Psi and H*Psi stored on disk. Mustbe larger than twice nvec nprint 0 0,1,2,3 print option ywcas F T: writes all coefficients of CAS determinants, whatever the value of the coefficient niter 50 number of iterations prec 1.d-6 convergence precision calcul 'MD' 'MD' normal. Direct diagonalization 'M' Matrix calculation, stored on mat file diagonaliser ensuite avec casdi (calcul='D') il faut alors : - mettre ygendet=T dans &locci, pour avoir les det - renommer la matrice MAT en Project.mat gener 'CAS+SD' CAS+SD, CAS+S, CAS+DDCI, CAS+DDC2, CAS 'SEL+SD' Selected Active Space. The Active Space is given with individual space determinants. All spin distributions are authomatically generated. possible data: SEL+SD, SEL+S, SEL+DDCI, SEL+DDC2, SEL The determinants are given as follows: after the &locci data: &locci ... / &selec for each determinant, a list of occupations in I1 format (noac numbers 0, 1 or 2, corresponding to the occupation of active orbitals, according to their position in the list of orbitals. end the Data by a line FIN or END example: &locci ... / &selec / 2200 1111 end syspin '0' '0' no spin symmetry '+' gives singlets, quintets... '-' gives triplets... is0 0 symmetry of the eigenvectors does not consider symmetry n computes vectors of symmetry n liens 'KIJ' 'KIJ' links obtained from exchange integrals 'FILE' links given in toto ascii file 'ALL' all orbitals linked 'ATOM' see below "liens='ATOM'" sl1 if liens='KIJ'. threshold to consider 2 orbitals as bonded LO1 bonded to LO2 if Kij(LO1,LO2)>sl1 (LO1 is occupied, LO2 is virtual) sl2 threshold to consider 2 pairs of LOs as bonded (LO1,LO2) bonded to i(LO3,LO4) if any (Kij(LO1/LO2,LO3/LO4)>sl2) (introduce dispersive terms) sli as sl1 and sl2, for integrals (LO1 LO2 || LO3 LO4) (one threshold only: sli1=sli2) zone ' ' To define different zones of the molecule with different sl,sl2 and sli thresholds default: one zone (the whole molecule), same sl,sl2 and sli on the whole molecule. 1. The zones are defined using file .LOinfo created at the localisation step - Edit the file .LOinfo structure of .LOinfo: column 1: rank of LO in dolo or lewis column 2: rank of LO in ijkloc (and casdiloc) column 3: LO label column 4: OCC/VIRT column 5: type of LO: SIG/PI/LP/DIFF_S/DIFF/PX... column 6: number of atoms in LO column 7...: atoms of LO example of 3 zones: zone 0 contains active orbitals add LOs to zone 0: example zone(1)='+orb:PI;+N:10;+typ:DIFF_PX' adds to zone 0: . LOs for which the string "PI" is present in label (column 3) . LO of rank 10 in column 2 . LO for which the string "DIFF_PX" is present in column 5 no other possibility for the moment idem to fill zone 1. Give zone(2)= Rem: LOs already defined in zone 0 remain in zone 0! Once given all zones until zone_n, the remaining LOs are in zone_(n+1) 2. The zones are defined by the atoms A1,A2,A3. example of 3 zones: Zone 0: all atoms, unless redefined after zone 1: a part of the molecule, defined by atoms A1,A2,A3. zone 1 = all orbitals on A1, A2 or A3 zone 2: another part of the molecule, defined by atoms A4,A5 Data: zone='A1,A2,A3','A4,A5',sl=0.001,0.003,0.01 sli=0.0001,0.0003,0.001,sl2=0.0001,0.0003,0.001 zone 0: sl=0.001,sli=0.0001,sl2=0.0001 zone 1: sl=0.003,sli=0.0003,sl2=0.0003 zone 2: sl=0.01,sli=0.001,sl2=0.001 data of type "1." cannot be mixed with data "2." The program finds whether types 1. or 2. is used no further indicqtion is needed. Warning: zones with lowests rank correspond to the most important parts of the molecule itest(29)=0 : (defaut) zone de la liaison i-j = min((zone(i),zone(j)) itest(29)=1 zone de la liaison i-j = max((zone(i),zone(j)) itest(29)=2 zone de la liaison i-j = moyenne((zone(i),zone(j)), Notice: itest(29)=0 corresponds to the LARGEST space, 1 to the smallest 2 is inbetween et les seuils consideres sont la moyenne entre les seuils des zones de i et j ysuiv,isuiv,suiv: ysuiv F T: The converged eigenvectors correspond to the trial ones, in the same order suiv 0.8 in ysuiv, "correspond" means the overlap between trial and converged vectors is larger or equal to suiv isuiv nvec+3 the maximum number of vectors to be computed in the seach of vectors corresponding to the guess. ycaspair F T: when two bonds (occ,virt) are linked to the CAS, they authomatically form a bond pair sl 0.01 when liens='KIJ': threshold of the exchange integrals to consider two orbitals as linked. s 0.03 threshold for printing of the determinants and coefficients of the eigenvector(s) stop 'fin' eventually stops the program before end 'fin': full calculation 'NDET': stops after calculation of the nb of det 'TOPO'; stops after calculation of TOPO matrix 'SL': analyses thresholds and zones, then stops 'ESSAI': computes trial vectors and stops info 'Infoloc' suffix of the name of infoloc file info1 ' ' name of the infoloc file. If info1=' ', the name is prefix//info other file names: same rules mono='Monoloc', topo, ijclloc2, ijclloc3, ijclloc4, davec, **************** Size Consistent Dressing mhab 0 0: no dressing 1: CEPA_0 dressing (Correlation Energy) 2: improved CEPA_0 dressing (Correlation Energy, without redondants) 3: Class Dressing nhab 1 Reference eigenvector for dressing iterh 2 davidson iteration after which dressing begins note: if mhab/=0 nessai is put to 3 ytestgendet F T : et mprint(7)=1 : ecrit les determinants ***************if nessai=1: for each vector: &locess it= holes ip= particles c= active occupation v= 0. coefficient end of vector: &locess c='FIN' / or &locess c='END' / example (nvec=2) &locess c='20',v=1., / &locess c='fin' / &locess c='+-',v=1. / &locess c='-+',v=-1. / &locess c='fin' / ***************if nessai=2: reads trial vector on file ess1 default: ess1="prefix"ess the name ess1 may be modified in namelist &cdifil ***************if liens='ATOM': 1. Condition: each bond name (O_....) must contain the name of the atoms involved in the bond. No other information must contain the name of an atom (be careful with "sigma" if there is a "Si" atom...) 2. all couple of bonds containing the same atom are authomatically linked 3. links with the CAS: data: &loc_at CAS list of atoms end all bonds containing an atom of the list are linked to the CAS