[Molpro-user] CCSD(T) for radicals and calculation of bond energies

[Hans-Joachim Werner]

some answers to your questions are inserted below:
Am 15.09.2008 um 21:16 schrieb Don Aue:

> Hans,
>
> I saw you had replied to a similar question on the Molpro users  
> acrchives and hoped you might be able to point me in the right  
> direction here.
>
> I'm just beginning to use Molpro at the NCSA Illinois supercomputer  
> site for its capabilities in CCSD and CCSD open-shell calculations  
> relative
> to Gaussian and have some questions below.
>
> Don Aue
>
>
> 1) I get ca. 1 kcal/mole higher HF energies on Molpro than  
> Gaussian03 for HF/6-31G(d) jobs.  I assume because Molpro uses a  
> slightly different 6-31G(d) basis set? though that seems odd.  There  
> is no energy difference using Dunning cc-pVXZ basis sets.
Probably because Gaussian uses cartesian basis functions (at least for  
Pople basis sets), Molpro by default spherical harmonics.
With the directive "cartesian" (see manual) you can use cartesians.

> 2) The main problem I have is that I get quite different energies  
> from Molpro and Gaussian for open-shell molecules as shown below for  
> methyl
> radical output data.  I understand that Molpro uses the RHF  
> reference energy and should give different final CCSD(T) energies,  
> but the Molpro data are so different that they give vastly incorrect  
> bond energy data when
> compared to Gaussian, using what I gathered are the proper energies  
> for methane and hydrogen atom calculated to be the same with Molpro  
> or Gaussian. Do you see what the problem is here?  I must be missing  
> something.
> (all calcs from the same MP2 geom and aug-cc-pVQZ basis)
Gaussian probably uses UHF orbitals in UCCSD, while Molpro uses spin- 
restricted HF (RHF).
But here the main difference is that you computed a wrong state (wrong  
symmetry):
{hf;wf,symmetry=3}
yields  ERHF=-39.57595392
If you don't specify the symmetry, the default is 1.
I dont know why Gaussian gives a slightly higher ROHF energy.

H.-J. Werner
>
>
> 			CH3 radical    H atom	    CH4     Bond Ee(kcal)
> Molpro UCCSD(T)		-39.55705752 -0.4999483 -40.4517185  247.69
> Gaussian UCCSD(T)	-39.7733227  -0.4999483 -40.4517185  111.98
>
>
>
> Note the energy differences:
>
> Molpro:
> UCCSD(T)/aug-cc-pVQZ        HF-SCF
>    -39.55705752          -39.38058519
>
> Molpro:
>        RCCSD(T)        HF-SCF
>    --39.55702945    -39.38058519
>
>
>
> Gaussian: same geometry, UCCSD(T)-FC\Aug-CC-pVQZ
>
>    CCSD(T)=-39.7733227  HF=-39.5804909
>
>                        ROHF=-39.5758929
>
> Input and Output files from Molpro on Illinois Supercomputer site:
>
> [aue at co-login ~]$ more ch3r-6mp-6mx4aug.inp
> memory,200,M
> geomtyp=xyz
> geometry
> 4
> ch3r-6mp-6mx4aug D3h CH3  //MP2(FC)/6-31+g(d,p)
> C                     0.000000    0.000000    0.000000
> H                     0.000000    1.074991    0.000000
> H                     0.930970   -0.537496    0.000000
> H                    -0.930970   -0.537496    0.000000
> end
>
> set,spin=1
> basis=aug-cc-pVQZ
> hf
> UCCSD(T)
>
> [aue at co-login ~]$ tail ch3r-6mp-6mx4aug.out
> PROGRAMS   *        TOTAL  UCCSD(T)        HF       INT
> CPU TIMES  *        30.26     16.08      6.16      8.00
> REAL TIME  *        34.20 SEC
> DISK USED  *         1.37 GB
>
> **********************************************************************************************************************************
>
>        UCCSD(T)        HF-SCF
>    -39.55705752    -39.38058519
>
> **********************************************************************************************************************************
> Variable memory released
>
>
>
> Gaussian on a cpu here at UCSB:
>
> ccl00:~/aue/ark/ch3> more ch3r-6mp-6mx4aug.com
> %mem=550mb
> #n CCSD(T,E4T)/aug-cc-pVQZ
> maxdisk=8gb
>
> ch3r-6mp-6mx4aug D3h CH3  //MP2(FC)/6-31+g(d,p)
>
> 0 2
> C                     0.000000    0.000000    0.000000
> H                     0.000000    1.074991    0.000000
> H                     0.930970   -0.537496    0.000000
> H                    -0.930970   -0.537496    0.000000
>
> ccl00:~/aue/ark/ch3> tail -40 ch3r-6mp-6mx4aug.log
> 1\1\GINC-ORIGIN\SP\UCCSD(T)-FC\Aug-CC-pVQZ\C1H3(2)\AUE\30-Nov-2007\0\\
> #N CCSD(T,E4T)/AUG-CC-PVQZ MAXDISK=8GB\\ch3r-6mp-6mx4aug D3h CH3 //MP2
> (FC)/6-31+g(d,p)\\0,2\C,0,0.,0.,0.\H,0,0.,1.074991,0.\H,0,0.93097,-0.5
> 37496,0.\H,0,-0.93097,-0.537496,0.\\Version=SGI64-G03RevC.02\State=2-A
> "\HF=-39.5804909\MP2=-39.7497864\MP3=-39.7657491\MP4D=-39.7700535\MP4D
> Q=-39.7666569\MP4SDTQ=-39.7720756\PUHF=-39.583656\PMP2-0=-39.751777\PM
> P3-0=-39.7668658\MP4SDQ=-39.7674657\CCSD=-39.7679668\CCSD(T)=-39.77332
> 27\S2=0.76166\S2-1=0.752975\S2A=0.750071\RMSD=7.434e-09\PG=C03H [O(C1)
> ,SGH(H3)]\\@
>
>
> GROWING OLD ISN'T SO BAD -- WHEN YOU CONSIDER THE ALTERNATIVE.
>
>                     -- MAURICE CHEVALIER
> Job cpu time:  0 days  6 hours 53 minutes  4.2 seconds.
> File lengths (MBytes):  RWF=   1065 Int=      0 D2E=      0 Chk=      
> 12 Scr= 1
> Normal termination of Gaussian 03 at Fri Nov 30 05:35:13 2007.
>
>
>