[molpro-user] Number of electrons in MRCC

Fri Nov 14 06:46:10 GMT 2014

Dear Hyungjun,

since your RHF calculation proceeds to do 3 iterations rather than 1, this means your version of Molpro is not up-to-date. This was fixed some time ago. You would need to update to the latest 2012.1 in order for this work correctly.

best regards,

-Kirk

On Nov 14, 2014, at 6:09 AM, Hyungjun Kim <hyungjun96 at kaist.ac.kr<mailto:hyungjun96 at kaist.ac.kr>> wrote:

 Dear Peterson,

I have added rhf before the mrcc calculation and have performed the test calculation for ionization potential of Hf atom.

The MRCC requires ignore_error option since the RHF gives not converged result.

The IP calculated by using CCSDT/aVTZ-PP is slightly different from the reported value in JCP, 130, 164108, 2009.

The CCSDT/aVTZ-PP gives the ionization potential of 150.42 kcal/mol, therefore the delta CCSDT is obtained as 0.33 kcal/mol, which is different from the reported value of 0.10 kcal/mol.

The output shows that the RHF energy is slightly changed even the "maxit,1" is used. (SCF stops after 3 cycles)

I think this is the possible problem. What other options are required to stop SCF just after the first cycle?

I have attached the input and output files of CCSDT ionization potential calculation according to your comments.

** When I manually modified the MRCC input file, fort.56, I can exactly reproduce the same results in your paper of 0.10 kcal/mol.

But Kallay gives me the comment that there are some problems of integral list which is passed from Molpro to MRCC when the multi concifugration is used as the reference.

I also attached 'Hf-CCSDT-IP1.zip' file which is manually modified MRCC input and corresponding results.

Could you please check them?

Yours sincerely,

---------------------------------------------------
Hyungjun Kim, Ph. D. Student,
Quantum Chemistry Laboratory, Department of Chemistry,

Korea Advanced Institute of Science and Technology [KAIST],
Daejeon 305-701, Korea
e-mail: hyungjun96 at kaist.ac.kr<mailto:hyungjun96 at kaist.ac.kr> or hyungjun96 at gmail.com<mailto:hyungjun96 at gmail.com>

phone : +82-42-350-2861

mobile phone : +82-10-8537-5051

web : http://qclab.kaist.ac.kr

-----------------------원본메일-----------------------
보낸사람 : "Peterson, Kirk" <kipeters at wsu.edu<mailto:kipeters at wsu.edu>>
받는사람: Hyungjun Kim <hyungjun96 at kaist.ac.kr<mailto:hyungjun96 at kaist.ac.kr>>
참 조: molpro-user molpro-user <molpro-user at molpro.net<mailto:molpro-user at molpro.net>>
보낸날짜: 2014-11-13 18:27:37 GMT +0900 (Asia/Seoul)
제목: Re: [molpro-user] Number of electrons in MRCC

Dear Hyungjun,

since the last program run before mrcc was multi, where the occ and closed cards do not correspond to a single determinant, mrcc got a little confused.  Before the call to mrcc, simply insert:

{rhf;
   occ,4,1,1,0,1,0,0;
closed,2,1,1,0,1,0,0;
    wf,12,1,2;
    start,2140.2;maxit,1}

Note that I've use the multi orbitals as initial guess and set the maximum number of iterations to 1. This will result in orbitals unchanged from your symmetry equivalenced ones from multi. You can then leave out all the options to mrcc except method=ccsdt.

regards,

-Kirk

On Nov 13, 2014, at 1:28 AM, Hyungjun Kim <hyungjun96 at kaist.ac.kr<mailto:hyungjun96 at kaist.ac.kr>> wrote:

Dear Peterson,

These are input and output files.

Actually the input file for Zr atom is made by the simple modification on the input file you sent to me for Hf atom.

Yours sincerely,

-Hyungjun

---------------------------------------------------
Hyungjun Kim, Ph. D. Student,
Quantum Chemistry Laboratory, Department of Chemistry,

Korea Advanced Institute of Science and Technology [KAIST],
Daejeon 305-701, Korea
e-mail: hyungjun96 at kaist.ac.kr<mailto:hyungjun96 at kaist.ac.kr> or hyungjun96 at gmail.com<mailto:hyungjun96 at gmail.com>

phone : +82-42-350-2861

mobile phone : +82-10-8537-5051

web : http://qclab.kaist.ac.kr<http://qclab.kaist.ac.kr/>

-----------------------원본메일-----------------------
보낸사람 : "Peterson, Kirk" <kipeters at wsu.edu<mailto:kipeters at wsu.edu>>
받는사람: Hyungjun Kim <hyungjun96 at kaist.ac.kr<mailto:hyungjun96 at kaist.ac.kr>>
참 조: "molpro-user at molpro.net<mailto:molpro-user at molpro.net>" <molpro-user at molpro.net<mailto:molpro-user at molpro.net>>
보낸날짜: 2014-11-12 21:43:09 GMT +0900 (Asia/Seoul)
제목: Re: [molpro-user] Number of electrons in MRCC

Dear Hyungjiun,

Can you post your actual input? There are some tricks to using orbitals from multi in mrcc.

Regards,

Kirk

Sent from my iPhone

On Nov 12, 2014, at 6:54 AM, Hyungjun Kim <hyungjun96 at kaist.ac.kr<mailto:hyungjun96 at kaist.ac.kr>> wrote:

Dear Molpro developers,

I have reported the same problem to the MRCC forum and the developer of MRCC, Mihaly Kallay, suggests me to report this error to Molpro developer.

This is my post on the MRCC forum.

***

I have tried to calculate the electronic structure of Zr atom using CCSDT in MRCC program interfaced to Molpro. Stuttgart ECP is employed during this calculation. (Therefore only 12 electrons are left.)

I have symmetrized HF orbitals and have done CCSDT calculation using these orbitals.

While 4s and 4p orbitals are left as core electrons, 2 electrons in 5s, and 2 electrons in 4d orbitals are distributed, respectively.

When MRCC reads CASSCF MO, there are 4 electrons.
Frozen orbitals: 4 ( 1 1 1 0 1 0 0 0)
Active orbitals: 6 ( 3 0 0 1 0 1 1 0)
Active electrons: 4
Spin quantum number: 1.0

However, the output created by MRCC gives there are only 2 electrons (1alpha and 1beta).
Number of alpha electrons: 1
Number of beta electrons: 1

I don't know why the number of electrons is reduced from 4 to 2. Could you give me a guide to include '4d' electrons in MRCC calculation?

Any comments are welcome.

Yours sincerely,

***

The reply of Kallay to this question is as follows:

***

Dear Hyungjun,
The occupation vector and the specification of active orbitals passed over to mrcc is not correct. You should contact the molpro developers about this issue. The quick solution is to specify the directory where mrcc is executed, that is, the corresponding line in the molpro input should be extended, e.g., as
{MRCC,method=CCSDT,spatial=1,HF=0,nacto=0,dir='/scr/mrcc';
Then the scratch files required by mrcc will be saved to that directory. To correct the error edit file fort.56, and replace the last two line by the following ones:
2 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 1 0 0 0
Then execute dmrcc in that directory. Note that in this case an MRCCSDT calculation will be run. If you want to perform a simple CCSDT calculation, the last line of fort.56 must be deleted.

***

I have fixed the problem referring to Kallay's comment.

Could you please check this problem?

Yours sincerely,

---------------------------------------------------
Hyungjun Kim, Ph. D. Student,
Quantum Chemistry Laboratory, Department of Chemistry,

Korea Advanced Institute of Science and Technology [KAIST],
Daejeon 305-701, Korea
e-mail: hyungjun96 at kaist.ac.kr<mailto:hyungjun96 at kaist.ac.kr> or hyungjun96 at gmail.com<mailto:hyungjun96 at gmail.com>

phone : +82-42-350-2861

mobile phone : +82-10-8537-5051

web : http://qclab.kaist.ac.kr<http://qclab.kaist.ac.kr/>

_______________________________________________
Molpro-user mailing list
Molpro-user at molpro.net<mailto:Molpro-user at molpro.net>
http://www.molpro.net/mailman/listinfo/molpro-user

<zr-atom.inp><zr-atom.out>

대용량파일
파일 이름   파일 용량   다운로드 가능기간
Hf-CCSDT-IP1.zip<https://mail.kaist.ac.kr/Mail?act=BIGFILE_DOWNLOAD&ukey=546591193fd858f6f4a644f6&mhost=kaist.ac.kr&language=ko>
        19.6 MB 2014-12-04
<hf-ccsdt-ip1.inp><hf-ccsdt-ip1.out>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20141114/cc9314fa/attachment-0001.html>