Can I set two different basis for optimization and frequencies within the same input?

[Hudd Ren]

Hi, 

I'm trying to first optimize geometry in gas phase with 6-31G* and then calculate vibrational frequencies in solution with 6-311+G**.

Can I do all these with one input file?  Or should i separate the input file for optimization and frequencies?

The below is the input file that I'm trying to run both at once with. As you can see, I switched the basis and added cosmo after 'task dft optimize' Would this work?

(And please kindly let me know if you see anything else wrong/strange as I just started using NWChem today!)

echo

start product

geometry units angstrom autosym

  C    -5.11803    0.15705    0.42668

  S    -4.21136   -0.16848   -1.10416

  C    -2.46901   -0.40230   -0.69500

  C    -1.83325    0.81703   -0.18808

  C    -0.41763    0.76235    0.19861

  O    -0.08715    1.37766    1.25788

  N     0.48716    0.04731   -0.60499

  C     1.86828   -0.10399   -0.37889

  C     2.54915    0.43050    0.68006

  N     2.58963   -0.83871   -1.28736

  C     3.92513    0.21874    0.80337

  C     3.91072   -1.04198   -1.16811

  C     4.61874   -0.52221   -0.12393

  H    -5.71525   -0.76721    0.62660

  H    -5.78601    1.03732    0.33530

  H    -4.38759    0.36702    1.23083

  H    -2.35475   -1.17847    0.11788

  H    -1.91666   -0.84649   -1.56206

  H    -2.34771    1.74137   -0.08673

  H     0.05768   -0.41891   -1.46788

  H     2.06180    1.01883    1.43761

  H     4.44700    0.64333    1.63820

  H     4.43246   -1.63125   -1.91519

  H     5.69666   -0.69851   -0.04494

end

basis

  * library 6-31*

end

dft

  xc b3lyp

  iteration 300  direct  disp vdw 3

end

task dft optimize

basis

  * library 6-311+G**

end

cosmo

  do_gasphase False

end

task dft freq

[Edoardo Aprà]

Yes,

You can switch basis.

However, you should re-optimize the geometry in solvated phased before computing the frequencies.

The attached input file contains a few modifications

[Hudd Ren]

Thanks a lot! This is a tremendous help for me as a beginner in NWchem.

By the way, just looking at the system size, how much time would you expect to calculate with 4 processors in general?

I came from Jaguar and this takes 30 minutes with only one processor on my Macbook but somehow, NWchem that I set up on AWS with faster CPUs takes about 10 hours with 4 processors.

I compiled it with OpenMPI and MKL and I see all processors are running concurrently so I doubt that the compilation was not done incorrectly.

I wonder if some parts in this NWchem input is requiring a significantly expensive accuracy.

Again I really appreciate your help!