1% 2% $Id$ 3% 4\label{sec:mcscf} 5 6The NWChem multiconfiguration SCF (MCSCF) module can currently perform 7complete active space SCF (CASSCF) calculations with at most 20 active 8orbitals and about 500 basis functions. It is planned to extend it to 9handle 1000+ basis functions. 10 11\begin{verbatim} 12 MCSCF 13 STATE <string state> 14 ACTIVE <integer nactive> 15 ACTELEC <integer nactelec> 16 MULTIPLICITY <integer multiplicity> 17 [SYMMETRY <integer symmetry default 1>] 18 [VECTORS [[input] <string input_file default $file_prefix$.movecs>] 19 [swap <integer vec1 vec2> ...] \ 20 [output <string output_file default input_file>] \ 21 [lock] 22 [HESSIAN (exact||onel)] 23 [MAXITER <integer maxiter default 20>] 24 [THRESH <real thresh default 1.0e-4>] 25 [TOL2E <real tol2e default 1.0e-9>] 26 [LEVEL <real shift default 0.1d0>] 27 END 28\end{verbatim} 29Note that the \verb+ACTIVE+, \verb+ACTELEC+, and \verb+MULTIPLICITY+ 30directives are {\em required}. The symmetry and multiplicity may 31alternatively be entered using the \verb+STATE+ directive. 32 33\section{{\tt ACTIVE} --- Number of active orbitals} 34 35The number of orbitals in the CASSCF active space must be specified 36using the {\tt ACTIVE} directive. 37 38E.g., 39\begin{verbatim} 40 active 10 41\end{verbatim} 42 43The input molecular orbitals (see the vectors directive, Sections 44\ref{sec:mcscfvectors} and \ref{sec:vectors}) must be arranged in 45order 46\begin{enumerate} 47\item doubly occupied orbitals, 48\item active orbitals, and 49\item unoccupied orbitals. 50\end{enumerate} 51 52\section{{\tt ACTELEC} --- Number of active electrons} 53 54The number of electrons in the CASSCF active space must be specified 55using the the {\tt ACTELEC} directive. An error is reported if the 56number of active electrons and the multiplicity are inconsistent. 57 58The number of closed shells is determined by subtracting the number 59of active electrons from the total number of electrons (which in turn 60is derived from the sum of the nuclear charges minus the total system 61charge). 62 63\section{{\tt MULTIPLICITY}} 64 65The spin multiplicity must be specified and is enforced by projection 66of the determinant wavefunction. 67 68E.g., to obtain a triplet state 69\begin{verbatim} 70 multiplicity 3 71\end{verbatim} 72 73\section{{\tt SYMMETRY} --- Spatial symmetry of the wavefunction} 74 75This species the irreducible representation of the wavefunction as an 76integer in the range 1---8 using the same numbering of representations 77as output by the SCF program. Note that only Abelian point groups are 78supported. 79 80E.g., to specify a $B_1$ state when using the $C_{2v}$ group 81\begin{verbatim} 82 symmetry 3 83\end{verbatim} 84 85\section{{\tt STATE} --- Symmetry and multiplicity} 86 87The electronic state (spatial symmetry and multiplicity) may 88alternatively be specified using the conventional notation for an 89electronic state, such as $^3B_2$ for a triplet state of $B_2$ 90symmetry. This would be accomplished with the input 91\begin{verbatim} 92 state 3b2 93\end{verbatim} 94which is equivalent to 95\begin{verbatim} 96 symmetry 4 97 multiplicity 3 98\end{verbatim} 99 100\section{{\tt VECTORS} --- Input/output of MO vectors} 101\label{sec:mcscfvectors} 102 103Calculations are best started from RHF/ROHF molecular orbitals (see 104Section \ref{sec:scf}), and by default vectors are taken from the 105previous MCSCF or SCF calculation. To specify another input file use 106the \verb+VECTORS+ directive. Vectors are by default output to the 107input file, and may be redirected using the \verb+output+ keyword. 108The \verb+swap+ keyword of the \verb+VECTORS+ directive may be 109used to reorder orbitals to obtain the correct active space. 110See Section \ref{sec:vectors} for an example. 111 112The \verb+LOCK+ keyword allows the user to specify that the ordering 113of orbitals will be locked to that of the initial vectors, insofar as 114possible. The default is to order by ascending orbital energies within 115each orbital space. One application where locking might be desirable 116is a calculation where it is necessary to preserve the ordering of a 117previous geometry, despite flipping of the orbital energies. For such 118a case, the \verb+LOCK+ directive can be used to prevent the SCF 119calculation from changing the ordering, even if the orbital energies 120change. 121 122Output orbitals of a converged MCSCF calculation are canonicalized as 123follows: 124\begin{itemize} 125\item Doubly occupied and unoccupied orbitals diagonalize the 126 corresponding blocks of an effective Fock operator. Note that in 127 the case of degenerate orbital energies this does not fully 128 determine the orbtials. 129\item Active-space orbitals are chosen as natural orbitals by 130 diagonalization of the active space 1-particle density matrix. 131 Note that in the case of degenerate occupations that this 132 does not fully determine the orbitals. 133\end{itemize} 134 135\section{{\tt HESSIAN} --- Select preconditioner} 136\label{sec:mcscfhessian} 137 138The MCSCF will use a one-electron approximation to the orbital-orbital 139Hessian until some degree of convergence is obtained, whereupon it 140will attempt to use the exact orbital-orbital Hessian which makes the 141micro iterations more expensive but potentially reduces the total 142number of macro iterations. Either choice may be forced throughout 143the calculation by specifying the appropriate keyword on the 144\verb+HESSIAN+ directive. 145 146E.g., to specify the one-electron approximation throughout 147\begin{verbatim} 148 hessian onel 149\end{verbatim} 150 151\section{{\tt LEVEL} --- Level shift for convergence} 152 153The Hessian (Section \ref{sec:mcscfhessian}) used in the MCSCF 154optimization is by default level shifted by 0.1 until the orbital 155gradient norm falls below 0.01, at which point the level shift is 156reduced to zero. The initial value of $0.1$ may be changed using 157the \verb+LEVEL+ directive. Increasing the level shift may make 158convergence more stable in some instances. 159 160E.g., to set the initial level shift to 0.5 161\begin{verbatim} 162 level 0.5 163\end{verbatim} 164 165\section{{\tt PRINT} and {\tt NOPRINT}} 166 167Specific output items can be selectively enabled or disabled using the 168\verb+print+ control mechanism~(\ref{sec:printcontrol}) with the 169available print options listed in table(\ref{MCSCF_print_options}). 170 171\begin{table}[htb] 172 173\label{MCSCF_print_options} 174 175\center 176 177\vspace{.2in} 178\begin{tabular}{lrl} 179\hline\hline 180Option & Class & Synopsis \\ 181\hline 182\verb+ci energy+ & default & CI energy eigenvalue \\ 183\verb+fock energy+ & default & Energy derived from Fock matrices \\ 184\verb+gradient norm+ & default & Gradient norm \\ 185\verb+movecs+ & default & Converged occupied MO vectors \\ 186\verb+trace energy+ & high & Trace Energy \\ 187\verb+converge info+ & high & Convergence data and monitoring \\ 188\verb+precondition+ & high & Orbital preconditioner iterations \\ 189\verb+microci+ & high & CI iterations in line search \\ 190\verb+canonical+ & high & Canonicalization information \\ 191\verb+new movecs+ & debug & MO vectors at each macro-iteration \\ 192\verb+ci guess+ & debug & Initial guess CI vector \\ 193\verb+density matrix+ & debug & One- and Two-particle density matrices \\ 194\hline\hline 195\end{tabular} 196 197\caption{MCSCF Print Options} 198 199\end{table} 200 201 202