# calculation of DBOC

6 months 6 days ago #1070 by Lorenzo
calculation of DBOC was created by Lorenzo
I'm using MRCC with Cfour 2.1 (I also downloaded all the available patches for MRCC). I noticed that the Diagonal Born-Oppenheimer Correction (aka DBOC or BODC or adiabatic correction) given by MRCC differ slightly from the ones computed by Cfour.
For example, consider these input for a streched geometry of the HF molecule.
The Cfour input (using Cfour for everything) is:
CFOUR calculation for HF
F  0.000000         0.00000000        0.00000000
H  0.000000         3.00000000        0.00000000

*CFOUR(CALC_LEVEL=CCSD
COORD=CARTESIAN
UNITS=1
REFERENCE=RHF
BASIS=PVDZ
CHARGE=0
MULTIPLICITY=1
DBOC=1
FROZEN_CORE=1
MEM_UNIT=GB
MEMORY_SIZE=4)

%masses
18.99840322
1.0078250321

It only takes a few seconds to run. It gives
RHF =       -99.849794298790385
CCSD=       -100.09088605228732
The total diagonal Born-Oppenheimer correction (DBOC) is:        627.971742 cm-1

The input for MRCC is the same but it contains the line
CC_PROGRAM=MRCC
. I obtain:
Total CCSD energy [au]:            -100.090886055013 (same as above)
The total diagonal Born-Oppenheimer correction (DBOC) is:        631.448228 cm-1

So the value given by MRCC is larger by about 3.5 cm-1 than the one by Cfour. Not a huge difference, but it's there. Decresing the bond length this difference MRCC-Cfour also decreases, and is 0.33 cm-1 at r=1.2 a0.

If MRCC and Cfour use slightly different ways to compute the DBOC, which one should be more trustworthy? For very accurate work these small differences may be not be negligible.

BTW, the same output as above but with AUG-PVTZ for the basis set and CCSDT method doesn't run successully on my machine ("executable xvdint finished with status 17152")

Any suggestion would be appreciated!
Lorenzo

6 months 6 days ago #1071 by kallay
Replied by kallay on topic calculation of DBOC
Dear Lorenzo,
You should try to tighten the convergence thresholds in both programs (cctol in mrcc, CC_CONV, LINEQ_CONV, SCF_CONV in Cffour). Then the correction should come closer.

Best regards,
Mihaly Kallay
The following user(s) said Thank You: Lorenzo

6 months 3 days ago #1072 by Lorenzo
Replied by Lorenzo on topic calculation of DBOC
Thanks a lot! I tightened the Cfour thresholds CC_CONV, LINEQ_CONV, SCF_CONV by a factor ten (from 1e-7 to 1e-, but the results changes insignificantly (from 627.971742 cm-1 to 627.971710 cm-1).
How can I change the "cctol" parameter for MRCC when using Cfour? I tried but failed...

6 months 3 days ago #1073 by kallay
Replied by kallay on topic calculation of DBOC
Set INPUT_MRCC=OFF in the ZMAT file. Then the mrcc input file is not written automatically, but you should write it manually. Create a MINP file including

iface=Cfour
calc=CCSD
dens=4
dboc=on
cctol=12

and run Cfour.

Best regards,
Mihaly Kallay
The following user(s) said Thank You: Lorenzo

6 months 3 days ago #1074 by Lorenzo
Replied by Lorenzo on topic calculation of DBOC
Thanks a lot for your answer! It kind-of worked, but not completely (full output attached)

#### File Attachment:

File Name: bodcmrcc.txt
File Size:73 KB

MRCC starts and does some part of the calculation, but eventually in the "CCSD analytic second derivative calculation" section, it stops with the message
forrtl: severe (36): attempt to access non-existent record, unit 62, file /scratch/lorenzo/bodc/0019/fort.62

The file fort.62 is present, but it seems MRCC is trying to get something which is not there...

##### Attachments:

6 months 2 days ago #1075 by kallay
Replied by kallay on topic calculation of DBOC
Sorry, it doesn't work in this way. The working solution is as follows.
Run a calculation with INPUT_MRCC=ON.
Edit file fort.56, change the 16th integer in the first line, e.g., to 12 (this is cctol).
Edit ZMAT and set INPUT_MRCC=OFF
Run cfour.

Best regards,
Mihaly Kallay
The following user(s) said Thank You: Lorenzo