Example MP2.X computation (is this example correct?)

Hi Everybody,
I am new to Psi4. Would somebody be so kind and check my implementation of the MP2.X method.
I implemented a test run for a H2O molecule and think I got it right but would like somebody to take a look just to be sure.
The basic idea is that I run:

  1. a MP3/6-31+g* single point calculation
    -> ec_mp2 mp2_correlation energy
    -> ec_mp3 mp3 correlation energy
  2. a MP2/CBS calculation with the cc-pv[dt]z basis set
    -> e_mp2_cbs

And then compute:
E(MP2.X/CBS) = e_mp2_cbs + C * (ce_mp3 - ce_mp2)

With C taken from http://onlinelibrary.wiley.com/doi/10.1002/cphc.201200850/abstract

The input file is below.

Thanks,
Alberto Gobbi

#! MP2.X calculation of H2O
memory 500 mb

molecule h2o {
O
H 1 1.0
H 1 1.0 2 104.5
}
set freeze_core true

# MP3/6-31+g* MP3/6-31+g* MP3/6-31+g* MP3/6-31+g* MP3/6-31+g*
set basis 6-31+g*
energy(‘mp3’)
e_scf = psi4.get_variable(‘SCF TOTAL ENERGY’)
ce_mp2 = psi4.get_variable(‘MP2 CORRELATION ENERGY’)
ce_mp3 = psi4.get_variable(‘MP3 CORRELATION ENERGY’)
mp3_basis = psi4.get_global_option(“BASIS”)
clean()

# MP2/CBS MP2/CBS MP2/CBS MP2/CBS MP2/CBS MP2/CBS MP2/CBS
cbs_basis = ‘cc-pv[dt]z’
set guess read
cbs(name=‘mp2’, scf_wfn=‘scf’, corl_wfn=‘mp2’, corl_basis=cbs_basis, corl_scheme=corl_xtpl_helgaker_2)
e_mp2_cbs = psi4.get_variable(‘CBS TOTAL ENERGY’)

# compute MP2.X
# coefficients for mp2.x from doi: 10.1002/cphc.201200850
mp2_x_c = { ‘3-21++g*’ : 1.12,
‘6-31g*’ : 0.86,
‘6-31+g*’ : 0.73,
‘6-311++g**’ : 0.65,
‘def2-tzvp’ : 0.67,
‘def2-tzvpp’ : 0.53,
‘cc-pvdz’ : 0.72,
‘aug-cc-pvdz’: 0.52,
‘aug-cc-pvtz’: 0.50 }
c = mp2_x_c[mp3_basis.lower()]
cor_mp2_x = c * (ce_mp3 - ce_mp2)
e_mp2_x = e_mp2_cbs + cor_mp2_x

psi4.print_out("\n\nMP2.X MP2.X MP2.X MP2.X MP2.X MP2.X MP2.X MP2.X \n")
psi4.print_out(“MP2.X (MP3(%s) CBS(%s)) C=%4.2f\n” %
(mp3_basis, cbs_basis, c))
psi4.print_out(" SCF TOTAL ENERGY = %12.7f\n" % e_scf)
psi4.print_out(" MP2 TOTAL ENERGY = %12.7f\n" % (e_scf + ce_mp2))
psi4.print_out(" MP3 TOTAL ENERGY = %12.7f\n" % (e_scf + ce_mp3))
psi4.print_out(" CBS TOTAL ENERGY = %12.7f\n" % e_mp2_cbs)

psi4.print_out(" MP2.X CORRECTION = %12.7f\n" % cor_mp2_x)
psi4.print_out(" MP2.X TOTAL ENERGY = %12.7f\n" % e_mp2_x)

1 Like

At a read-through, your implementation looks correct. The snippet above will need to change into something like energy(cbs, ... for most versions of psi4 from this calendar year (see the cbs test cases). If that paper has some reference values, it can be helpful to add `compare_values(ref_val, your_val, 6, ‘msg’) to maintain correctness.

Thanks for the quick reply!
Unfortunately, I did not find the access to specific geometries and results in the paper. I will take another look.

Alberto

I didn’t look very far, but found the below. Unfortunately only summary statistics, but may be a help. S66 geometries are at https://github.com/psi4/psi4/blob/master/psi4/share/psi4/databases/S66.py