Q. How to determine the
thresholds for the exponential generation of the higher-order excitation
operators?
A. Higher-order
linked excitation operators up to sextuples are generated by the exponential generation scheme described in the following
papers:
H. Nakatsuji, Chem. Phys. Letters, 177, 331 (1991).
H. Nakatsuji, J. Chem. Phys. 83, 713, 5743 (1985).
H. Nakatsuji, J. Chem. Phys. 94, 6716 (1991).
See
also description about EGOP in the Part II-C of the SAC-CI guide. Single
excitation operators are included without selection. Doubles are handled as in
SD-R part of the program.
Higher-order excitation operators are generated as the products of A and B
operators, R=A*B, A*A*B. The A and B operators are single and double excitation
operators and selected with a criterion explained below.
The second-order product generates
triple and quadruple excitation operators.
R=A(S)*B(D) and A(D)*B(S), triples
A(D)*B(D)
quadruples
gSh
and gDh denote single and double excitation operators, respectively. The
A-operators used for the second-order products are single and double excitation
operators whose coefficients in the reference SD-CI wave function are larger
than given thresholds
CThreEgABAOp=(A1Op=<thrma1>,A2Op=<thrma2>).
The
<thrma1> and <thrma2> are thresholds for the single and double
excitation operators, respectively.
The third-order product generates
quintuple and sextuple excitation operators.
R=A(D)*A(D)*B(S) and A(D)*A(S)*B(D), quintuples
A(D)*B(D)*B(D)
sextuples
The
A-operators used for the third-order product are also single and double
excitation operators whose coefficients in the reference SD-CI wave function
are larger than given thresholds
CThreEgAABAOp=
(A1Op=<thmma1>,A2Op=<thmma2>).
The
<thmma1> and <thmma2> are thresholds for the single and double
excitation operators, respectively.
For
B-operators, the thresholds are given by
CThreEgAB=(B1Op=<thrmb1>,B2Op=<thrmb2>)
for the second-order product and
CThreEgAAB=(B1Op=<thmmb1>,B2Op=<thmmb2>)
for the third-order product.
The
<thrmb1> and <thmmb1> are thresholds for the single excitation
operators, and the <thrmb2> and <thmmb2> are those for the double
excitation operators.
Thus, the
accuracy of the exponential generation of the higher R operators in the general-R
method is expressed by a set of the thresholds,
(<thrma1>,
<thrma2>, <thmma1>, <thmma2>) and
(<thrmb1>,
<thrmb2>, <thmmb1>, <thmmb2>).
The
default values are
(<thrma1>,
<thrma2>, <thmma1>, <thmma2>)=(0.04, 0.04,
0.1, 0.1)
(<thrmb1>,
<thrmb2>, <thmmb1>, <thmmb2>).=(0.04,
0.04, 0.1, 0.1)
These
parameters can be defined for each spatial- (ISYM) and spin-symmetries
(KSTATE). In addition, the linked operators up to quadruple are further
filtered by the perturbation selection with a given threshold (EThreEgR or eseleg). For
quintuple and sextuple excitation operators, no perturbation selection is
performed.