Molcas Forum

tararia

Member

Registered: 2015-12-03

Posts: 13

Playing active space

Dear team,
                I am calculating magnetic properties of a 4f-metal (lanthanide) ion. I have finished the calculations considering 4f orbitals in the active space. Now,I am keen to include 4d orbitals also into the active space so that to have 19 electrons in 12 orbitals. But,as per the manual I realized increasing active space could be done by considering RAS1 and RAS3. In this context, I am initially correcting the 4f orbital active space through RASSCF calculation. Then, with this resultant RASSCF output file and orbital analysis I am doing second RASSCF calculations to consider 4d orbitals also into active space. The input is as follow:

>export MOLCAS_MOLDEN=ON
&RASSCF
Spin
6
Inactive
216
Nactel
19 1 1
Ras1
5
Ras2
7
Ras3
5
Alter
5
1 70 217
1 71 218
1 72 219
1 73 220
1 74 221
CiRoot
21 21 1
Levshft
2.5
ITERations
400 200

The above input converges correctly...........
Now, the issue is (4d+4f) consideration (above input) leads to far different SINGLE_ANSIO result from that obtained from only{4f} consideration.

Could you please help me in this regard with your invaluable suggestions.

regards,
Tulika

Steven

Administrator

From: Lund

Registered: 2015-11-03

Posts: 95

Re: Playing active space

Are you talking about 4d or 5d??

If the d orbitals are initially occupied, it would make sense to add those 5 orbitals in RAS1, and allow a single excitation out of RAS1 to find excited states with a d^9 occupation. If the d orbitals are initially empty, then you would add them in RAS3 instead. But you are adding both, do you want to include two d shells?

If you only want 1 d shell, why don't you simply extend RAS2 with 5 orbitals, instead of using RAS1/RAS3?

It is unclear exactly what you are trying to do, which ion are you computing, and which d shell(s) you are talking about.

---

Always check the orbitals.

tararia

Member

Registered: 2015-12-03

Posts: 13

Re: Playing active space

Hi,
     thanks for response.

I am calculating on Dy ion.Yes it is 4d occupied orbital ...........hence, trying to use five orbitals...........I am trying to include five 4d orbitals into the active orbitals of seven 4f orbitals...........
Yes, I attempted increasing the RAS2 from seven to twelve (7+5). But, that was not accurately placing  these 4d orbitals into active orbitals of higher energy as these 4d orbitals of very low energy.They would remain in their original state itself.
Hence, based on the manual I thought of including RAS1/RAS3 space to increase the active space. i.e. from five occupied orbital transition will happen to empty five orbitals (Hence, used five orbitals in the RAS1 and RAS3 space).
Shall I include only RAS1=5 and RAS3=0 and same accordingly in the NACTEL case to get these things converged correctly.
please help me out with your advice.

Steven

Administrator

From: Lund

Registered: 2015-11-03

Posts: 95

Re: Playing active space

Excitations from 4d will be very high-lying, and probably undoable, you would need extremely many roots and include all possible other orbitals that are higher than 4d.

So, the solution is to use something similar to when computing core-excited states, including the 4d into RAS3 and allowing 9 electrons. But you'll also need to freeze possible orbital rotations to keep the 4d orbitals in there.

---

Always check the orbitals.

tararia

Member

Registered: 2015-12-03

Posts: 13

Re: Playing active space

Dear team,
    Thanks....

I the input give above, I mentioned the number of roots used as 21 only as used for normal CASSCF calculations......

Could keeping roots 21 and keeping 5d orbitals only in RAS1, and RAS3 as empty method be attempted?  The lower energy 4d orbitals are being rotated with the higher energy orbitals (alter keyword in above input)........but, freezing all possible rotations is it not infeasible?

Is playing with active space 4d not fruitful?shall I try with 5d or 6s empty orbitals? Can then the convergence would be easier and procured results  also much reliable?

Steven

Administrator

From: Lund

Registered: 2015-11-03

Posts: 95

Re: Playing active space

Off-topic: don't draw attention to a question by re-posting useless messages. The post was deleted.

On-topic:

You are talking about all kinds of schemes, but you failed to explain *why* you wish to add the 4d orbitals to the active space. You are asking us what orbitals to add, which makes me think you have no clue how to select an active space to compute the properties that interest you. So, my first recommendation would be to go and read relevant literature on the topic of multiconfigurational calculations and lanthanides.

---

Always check the orbitals.