CC without spin-orbit couplings

CC without spin-orbit couplings can be run by omitting SOC integrals in xuanyuan and adding lscalar in CCSDSO. CCSD(T) can be enabled by setting itriple to 1.

$COMPASS
Title
 CH2O+ Molecule test run
Basis
 6-31g
Geometry
 H 0. 0.0 0.0
 F 1. 0.0 0.0
END geometry
$END

$XUANYUAN
scalar
heff
0 # non-relativistic
$END

$SCF
RHF
charge
0
spin
1
THRESHCONV
1.d-14 1.d-10
$END

$TRAINT
orbi
hforb
$END

$CCSDSO
itriple
1
ifdebug
0
maxTcyc
100
threshT
12
lscalar
$END

Results:

      T1 Diagnostic =   8.9274007366236547E-003
        E(SCF)            =     -99.977636678624
        E(CCSD)           =    -100.113388640000
 SO-CCSD calculation finish at: Mon Dec  7 13:26:53 2015
 Total cpu time for CCSD calculation   :      0.0727 seconds
                0.07 seconds walltime passed

 SCF  ENERGY:  -99.977636678624
 CCSD ENERGY:   -0.135751961376 -100.113388640000

 CCSD(T) calculation starts!
        E4T (T(CCSD))     =      -0.000922841190
        E5ST              =       0.000159516908
        E(T(CCSD))        =      -0.000763324282
        E(CCSD + T(CCSD)) =    -100.114151964282
 CCSD(T) calculation finish at: Mon Dec  7 13:26:53 2015
 Total cpu time for triple calculation :      0.0197 seconds
                0.02 seconds walltime passed


 [ccsdso_calculation]
  Escf    =      -99.9776366786
  Eccsd   =     -100.1133886400   Ecorr=       -0.1357519614
  Eccsd(t)=     -100.1141519643   Ecorr=       -0.1365152857

Including scalar relativistic effects via sf-X2C:

$XUANYUAN
scalar
heff
3 # sf-x2c
$END

Results:

 [ccsdso_calculation] 
  Escf    =     -100.0586781599
  Eccsd   =     -100.1944404677   Ecorr=       -0.1357623078
  Eccsd(t)=     -100.1952071862   Ecorr=       -0.1365290263

CC with SOC

In fact, there are several options for scalar and spin-orbit parts. They are defined by the integer values of the keywords heff (for scalar part) and hsoc (for SOC part). In the input one could use them in the xuanyuan section for integrals, e.g.,

$XUANYUAN
scalar
heff
3
soint
hsoc
2
$END

The followings are some useful options for heff: 1. heff=0: NR

2. heff=3: sf-X2C (recommended)

3. heff=5 or 6: sf-X2C + (some spin-free parts originated from high-order SOC: 5 for Wso*Wso like terms, 6 for third order).

For the keyword hsoc, there are:

1. hsoc=0: SO1e (only 1e part)

2. hsoc=1: SOMF (mean-field)

3. hsoc=2: SOMF-1c (one center approximation, recommended)

The most practical one is heff=3, hsoc=2, which is used in computing the fine-structure splittings in the paper. One might want to try the 1e-part and SOMF (hsoc=0,1) with heff=0, which gives Breit-Pauli form of SOC, to see whether the similar results can be obtained in other program.

Example:

$COMPASS    
Title    
 CH2O+ Molecule test run    
Basis    
 6-31g
Geometry    
 H 0. 0.0 0.0 
 F 1. 0.0 0.0 
END geometry    
$END    
    
$XUANYUAN
scalar
heff
3
soint
hsoc
2
$END

$SCF
RHF
charge
0
spin
1
THRESHCONV
1.d-14 1.d-10
$END

$TRAINT
orbi
hforb
$END

$CCSDSO
itriple
1
ifdebug
0
maxTcyc
100
threshT
12
memmega
20
$END

Results:

 [ccsdso_calculation] 
  Escf    =     -100.0586781599
  Eccsd   =     -100.1944443737   Ecorr=       -0.1357662138
  Eccsd(t)=     -100.1952110890   Ecorr=       -0.1365329291