Most accurate basis set

[kimdn]

Can I know which basis set is the most accurate one in nwchem_semiemp?

It appears nwchem_semiemp uses CNDO, INDO/1, MNDO, AM1, and PM3.
Since I can’t find “PM5” in semiemp code, can I assume that the most accurate one is PM3?

However,
when I specified

echo

start NCAA
title NCAA

memory total 128 gb
basis "ao basis" spherical print
  *  library PM5
end

semiemp
  mult 1
  apx PM5
  scftype rhf
end

driver
  maxiter 50
  xyz opt
end

charge 0
geometry noautoz units angstrom nocenter autosym
  load 1uao_Chignolin_MODEL_1_sc_H_to_I.pdb
end

task semiemp optimize

output.out shows

APPROXIMATE HAMILTONIAN=PM5

So PM5 is the most accurate one?

[SoumenChem]

Hi Kim,

You do not need to specify a basis set for semiempirical calculations using the SEMIEMP code. Please see the "manual.pdf" file for a sample input. I have an implementation of PM3 and AM1 in this code but it is not well-tested and these methods are not mentioned in the manual. I would not recommend using any method that is not mentioned in the manual without going through the code carefully. I developed this code to perform spectroscopic calculations on large systems using real-time INDO/S method. Please see this two papers where we have reported the code and relevant applications-

Soumen Ghosh, Jason C. Asher, Laura Gagliardi, Christopher J. Cramer, Niranjan Govind. A Semiempirical Effective Hamiltonian Based Approach for Analyzing Excited State Wave Functions and Computing Excited State Absorption Spectra Using Real-Time Dynamics. J. Chem. Phys. 2019, 150, 104103 .

Soumen Ghosh, Amity Andersen, Laura Gagliardi, Christopher J. Cramer and Niranjan Govind. Modeling Optical Spectra of Large Organic Systems using Real-Time Propagation of Semiempirical Effective Hamiltonians. J. Chem. Theory Comput. 2017, 13, 4410−4420.

Please let me know if you have any other questions.

Best,

Soumen Ghosh