Molcas Forum

carlo

Member

Registered: 2020-05-21

Posts: 2

GUGA AND MRCI issues with openmolcas

Dear (Open)Molcas users,

I am trying to run an MRCI calculation for the benzene molecule in openmolcas.

First of all, can I use RICD?
I did two calculations (with and without RICD), and I get two different energy values (several hartees different).
I don't know if trusting the values obtained with RICD, since the MOTRA requires the two electrons integral file.

Also, when I freeze all the virtual orbitals, I get the following error:

 ###    Location: alloc_mrci.f:                                     ###
 ###                                                                         ###
 ###                                                                         ###
 ###    no virtual orbitals in the basis                          ###

Isn't it possible to do an MRCI calculation with CSFs singles and doubles generated only from the active orbital reference space (without any external orbital)?

And lastly, I was trying to do an MRCI calculation for anthracene, with 10 reference configurations (14 active orbitals).
This is the input:

&gateway
coord = final.xyz
Basis set =  H.ANO-RCC-VDZP, C.ANO-RCC-VDZP
Group = nosym
RICD
End
&SEWARD
>> COPY $CurrDir/hf_anthracene_vdzp.ScfOrb     INPORB
&MOTRA
LUMORB
frozen =40
deleted = 192
&GUGA
electrons= 14
spin =1
inactive = 0
active = 14
reference
  10  14
21212221000001
21212220100001
21212220010001
12212221000100
21122220001001
21122220000101
12122220000101
11222220010100
11222220000101
11222220000011

&MRCI
nrroots = 2
REFCI

And I got an error from GUGA:

      OCCUPATION OF REFERENCE STATES

      REF.STATE  ORB: 1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16
          1           2   1   2   1   2   2   2   1   0   0   0   0   0   1
          2           2   1   2   1   2   2   2   0   1   0   0   0   0   1
          3           2   1   2   1   2   2   2   0   0   1   0   0   0   1
          4           1   2   2   1   2   2   2   1   0   0   0   1   0   0
          5           2   1   1   2   2   2   2   0   0   0   1   0   0   1
          6           2   1   1   2   2   2   2   0   0   0   0   1   0   1
          7           1   2   1   2   2   2   2   0   0   0   0   1   0   1
          8           1   1   2   2   2   2   2   0   0   1   0   1   0   0
          9           1   1   2   2   2   2   2   0   0   0   0   1   0   1
         10           1   1   2   2   2   2   2   0   0   0   0   0   1   1
      WAVE-FUNCTION SYMMETRY LABEL:  1
 Config: LMN.GT.JRX
 LMN,JRX=                 9001                 9000
 This error is almost certainly that the problem
  at hand requires larger arrays than GUGA is
  compiled for. Please check your input against
  the manual. If you are certain that this
  calculation should be possible, please report
  this as a bug to the Molcas developers. Use
  the link '[url]http://www.teokem.lu.se/molcas[/url]'
  and then 'Entry for Users' and
  'Report a new bug'.

I'd be glad if you could help me in understanding these calculations.
Thank you very much,
Best regards,

Carlo

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: GUGA AND MRCI issues with openmolcas

First of all, can I use RICD?
I did two calculations (with and without RICD), and I get two different energy values (several hartees different).
I don't know if trusting the values obtained with RICD, since the MOTRA requires the two electrons integral file.

Absolute energy values will always be different with and without RICD, but several hartrees looks like too large a difference. I believe MOTRA should support RICD, but it's not sure everything works. GUGA and MRCI and virtually unmaintained. You may have better luck with GUGADRT/GUGACI (but I wouldn't hold my breath).

Isn't it possible to do an MRCI calculation with CSFs singles and doubles generated only from the active orbital reference space (without any external orbital)?

Probably not, according to the message. You could try deleting the orbitals in MRCI instead of MOTRA.

 Config: LMN.GT.JRX
 LMN,JRX=                 9001                 9000
 This error is almost certainly that the problem
  at hand requires larger arrays than GUGA is
  compiled for

The message says it all: too large memory requirements. GUGA is probably using static memory, so increasing MOLCAS_MEM will not help. Some index is limited to 9000 and you calculation exceeds it. From the code it looks like that's the number of reference configurations

carlo

Member

Registered: 2020-05-21

Posts: 2

Re: GUGA AND MRCI issues with openmolcas

Thanks Ignacio.

Best,
Carlo