Different basis ordering when point charges included

More
2 weeks 4 days ago #1545 by jxzou
Different basis ordering when point charges included was created by jxzou
Dear MRCC developers,

Thanks for developing this powerful program. I have a question that, MRCC seems to use different basis ordering when point charges are included in a calculation. To illustrate this problem, a C1-symmetry (i.e. no symmetry) molecule is taken as an example. Here are two `MINP` files
Code:
mem=4GB charge=0 mult=1 calc=RHF basis=PVTZ ecp=none scfiguess=mo symm=0 geom=xyz 8 C         -0.00002000         0.54842100         0.03465800 H         -0.03352200         1.24437300        -0.77739400 H          0.00599800         1.08230800         0.96192700 N          1.21850600        -0.26726500        -0.06903300 H          2.02170600         0.32739600        -0.03371100 H          1.25072200        -0.91646300         0.69090400 O         -1.14865700        -0.30170500        -0.01845500 H         -1.94499900         0.23418600        -0.00232200

and
Code:
mem=4GB charge=0 mult=1 calc=RHF basis=PVTZ ecp=none qmmm=Amber scfiguess=mo symm=0 geom=xyz 8 C         -0.00002000         0.54842100         0.03465800 H         -0.03352200         1.24437300        -0.77739400 H          0.00599800         1.08230800         0.96192700 N          1.21850600        -0.26726500        -0.06903300 H          2.02170600         0.32739600        -0.03371100 H          1.25072200        -0.91646300         0.69090400 O         -1.14865700        -0.30170500        -0.01845500 H         -1.94499900         0.23418600        -0.00232200 pointcharges 1 50.0 50.0 50.0 0.1

The molecule will not be re-oriented in the 1st job since it is C1-symmetry. A small point charge is put in a distant position in the 2nd job. If we run two jobs sequentially and provide MOCOEF of the 1st job to the 2nd one, one would expect the SCF to be converged in 1~3 cycles in the 2nd job since the point charge has little effect on the molecule. But the fact is that it takes 10 cycles to reach convergence and the SCF initial energy is higher by 5 a.u. than the converged energy

Warning: Spoiler!


This is counterintuitive and it makes two similar MRCC jobs inconsistent with each other. After some checking, I think it is probably due to different basis orderings are used with/without point charges included. If this is the case, would you please consider to use the same basis ordering, like in the next MRCC update? That would make two MRCC jobs more consistent inherently.

By the way, MRCC 25.1.0 is used here. And you can find related files are attached.
 

File Attachment:

File Name: charge_test.tar.gz
File Size:156 KB


Thanks a lot.

Best wishes,
Jingxiang

Please Log in or Create an account to join the conversation.

  • nagypeter
  • Offline
  • Premium Member
  • Premium Member
  • MRCC developer
More
2 weeks 4 days ago #1546 by nagypeter
Replied by nagypeter on topic Different basis ordering when point charges included
Dear Jingxiang,

without the point charge, the system is transformed to standard orientation, which is not performed with the point charge.

At the moment you try these workarounds: place your molecules in the input in standard orientation or add a zero point charge.

Bests,
Peter

Please Log in or Create an account to join the conversation.

More
2 weeks 4 days ago #1547 by jxzou
Replied by jxzou on topic Different basis ordering when point charges included
Dear Prof. Peter,

The molecule shown above is in C1 symmetry, and the Cartesian coordinates are already in its standard orientation, which can be checked again from the MRCC output. But the 2nd job cannot be converged in less than 3 cycles. So my guess is that standard orientation is not the issue here, but the basis ordering is.

Best wishes,
Jingxiang

Please Log in or Create an account to join the conversation.

  • nagypeter
  • Offline
  • Premium Member
  • Premium Member
  • MRCC developer
More
2 weeks 4 days ago #1548 by nagypeter
Replied by nagypeter on topic Different basis ordering when point charges included
If you turn on verbosity=3, then you will see that your example with no charge is reoriented (cf. sign changes in the coordinates).

Please Log in or Create an account to join the conversation.

More
2 weeks 4 days ago #1549 by jxzou
Replied by jxzou on topic Different basis ordering when point charges included
Thank you very much. The signs of some Cartesian coordinates do change (although absolute values do not change). Then this issue should still be a re-orientation problem.

Please Log in or Create an account to join the conversation.

Time to create page: 0.043 seconds
Powered by Kunena Forum

MRCC Login