This thread is locked.Only browsing is available.
Top Page > Browsing
How to specify contracted basis
Date: 2012/02/18 10:53
Name: Larry Fried   <>

Dear Openmx users,

I have started using openmx. I am confused about how to specify one of the pre-made contracted basis sets versus a basis set based on primitive psuedo-atomic orbitals.

For example, if I specify

C C5.0-s1p1

I understand that the file C5.0.pao is used. It is not clear to me, however, whether the contractions specified by <Contraction.coefficients2 are being used, or whether the primitive pseudo-atomic orbitals specified in <pseudo.atomic.orbitals are used. How do I know whether the code is using a contraction or not, and how can I control whether contractions or primitives are used ?

Thanks for your help,

Larry Fried

Page: [1]

Re: How to specify contracted basis ( No.1 )
Date: 2012/03/10 09:18
Name: Larry Fried  <>

After studying the documentation, the code, and the output here is my understanding of this issue. The contraction coefficients specified with <Contraction.coefficients1 ... Contraction coefficients1> are not actually read by openmx. The contraction coefficients appear to be present as documentation for how the psuedo-atomic orbital contraction was generated. In the orbital optimization method, the psuedo-atomic orbitals specified by <psedo.atomic.orbitals.L=0 are contractions up to the maximum multiplicity specified in the orbital optimization input. For instance, if C5.0-s4>1p4>1 is specified in orbital optimization, then the lowest energy state of L=0 and L=1 are stored as contractions in the PAO file. The other orbitals listed (higher energy states of L=0, L=1, all states of L=2, L=3) are primitive pseudo-atomic orbitals and are not contracted. Apparently, these are present to give the user the flexibility to augment the optimized contracted orbitals if desired. If the user wants to experiment with primitive pseudo-atomic orbitals, they need to be generated by the user, since not all primitive orbitals are present in the PAO file.

If anyone has any further tips, I would appreciate your input.

Best regards,

Larry Fried

Page: [1]