Ground-state geometric derivatives

Our example is H2O:

$COMPASS
Title
 h2o
Basis
 6-31G
Geometry
O .0000000000 -.2249058930 .0000000000
H 1.4523499293 .8996235720 .0000000000
H -1.4523499293 .8996235720 .0000000000
End geometry
units
bohr
skeleton
$END

$xuanyuan
direct
schwarz
$end

$scf
RKS
DFT
BHHLYP
charge
0 
spin
1
THRESHCONV
1.d-12 1.d-12
guess
hcore
grid
medium
$end

Ground-state gradients

Input:

$resp
GEOM
norder
1
method
1
$end

Output:

Grid in resp - Medium (75,302)

  Gradient contribution from Tot-egrad
     1        0.000000000000001        0.000000000000014        0.020661555871328
     2        0.000222118661238        0.000000000000001       -0.010328989065672
     3       -0.000222118661237       -0.000000000000001       -0.010328989065672
  Sum of gradient contribution from Tot-egrad
              0.000000000000002        0.000000000000015        0.000003577739983

Grid in resp - Fine (88,590)

  Gradient contribution from Tot-egrad
     1        0.000000000000001       -0.000000000000000        0.020661280611129
     2        0.000222098322782        0.000000000000001       -0.010330740060356
     3       -0.000222098322781       -0.000000000000001       -0.010330740060356
  Sum of gradient contribution from Tot-egrad
              0.000000000000001        0.000000000000000       -0.000000199509583

Grid in resp - Ultra Fine (100,1202)

  Gradient contribution from Tot-egrad
     1        0.000000000000000        0.000000000000133        0.020661446561909
     2        0.000222140335598        0.000000000000001       -0.010330700335443
     3       -0.000222140335597       -0.000000000000001       -0.010330700335443
  Sum of gradient contribution from Tot-egrad
              0.000000000000001        0.000000000000133        0.000000045891023

It is seen that by using larger number of grids, the translation error (sum of gradients for all atoms) is reduced (3*10^-6, 2*10^-7, 5*10^-8).

Ground-state hessians

Not implemented yet.