Molcas Forum

PIjush Karak

Member

Registered: 2022-04-18

Posts: 10

Index problem related to the roots of the triplet level in NAC

Dear developers,

I am trying to calculate the NAC between the two triplet states. Therefore, I am using ALASKA module and then NAC option including the specified roots. Please clarify me whether the triplet levels starts from the root=1 or else, i.e. whether the lowest triplet state is indexed as T1 or else and then denoted for other triplet states as it is required in the NAC section. Another query is how  the Rlxroot in the NAC computation should be chosen.
I will be very thankful for your kind reply.

With regards,
Pijush

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: Index problem related to the roots of the triplet level in NAC

The index, for each multiplicity, starts from 1 being the lowest, so typically S0 will be root 1 for singlets, and T1 will be root 1 for triplets. You don't need Rlxroot for NAC (you also don't need it for gradients if you specify the root in ALASKA).

PIjush Karak

Member

Registered: 2022-04-18

Posts: 10

Re: Index problem related to the roots of the triplet level in NAC

Thank you very much Ignacio.

One question is, if I want to compute the NAC between the T2-T1 state at the geometry of T2, then what will be the way out to fix this problem. I think Rlxroot should be used in this purposes. Please clarify me about this issue.

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: Index problem related to the roots of the triplet level in NAC

You first optimize the T2 geometry (and you use Rlxroot=2 in RASSCF, or Root=2 in ALASKA). Then, at the optimized geometry you do a single-point NAC calculation for roots 1-2.

PIjush Karak

Member

Registered: 2022-04-18

Posts: 10

Re: Index problem related to the roots of the triplet level in NAC

Many thanks for clarifying this issue.

In my purposes I have also used ciroot option for state averaging calculation. My input file is attached here,(for the NAC calculation between the T2-T1 states at the T2 optimized geometry)

&GATEWAY
coord=test_t2_geom.xyz
Basis=6-311G**
Group=NOSYM

&SEWARD
DoAnalyticl

&SCF

&RASSCF
Spin=3; Nactel= 8 0 0; Inactive=30; Ras2= 8
CiRoot
  2 2 1
Rlxroot=2

&Alaska
NAC=2 1

In this case ciroot 2 2 1  has been used. Whether use of large number of roots are required to get better result or it is sufficient. In my case for the T2-T1 states, two roots with equal weights have been used.

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: Index problem related to the roots of the triplet level in NAC

There's no particular requirement regarding the number of roots for NAC (except that it should include at least the two roots concerned, of course). You'll have to examine the actual wavefunctions (or natural orbitals and occupations) to decide if the active space, number of roots, etc. is adequate for your system.

PIjush Karak

Member

Registered: 2022-04-18

Posts: 10

Re: Index problem related to the roots of the triplet level in NAC

Many many thanks for clarifying me all this problem.