Verification of Newton-X Installation with Tests

[Gabriel Phun]

Hello,

I am trying to use Newton-X with Gaussian09 for NAMD. So far, I have gone through the tutorial (http://vdv.dcf.mybluehost.me/nx/wp-content/uploads/2020/02/tutorial-2_2.pdf) for the "Initial Conditions and Spectra Simulation: TD-DFT - Gaussian 09" and "Surface Hopping Dynamics: TD-DFT - Gaussian 09" and everything seems to be running.

I want to test that Newton-X isn't doing anything unexpected though so I am trying the tests provided in test-nx. I only have access to Gaussian 09 so I can only do a subset of the tests. 

Here are the results:

• Program: GAUSSIAN
  Method: CASSCF
  Description: Dynamics, Non-adiabatic with NAC vectors
  /rhome/gphun/newtonx-cs/bin/../test-nx/MD-GAU-CASSCF-NAD-VEC 

  • ERROR termination

• Program: GAUSSIAN       
  Method: TDDFT     
  Description: Dynamics, Non-adiabatic with cioverlap

  • It seems to be only minor numerical differences. Probably installation is OK.

• Program: GAUSSIAN      
  Method: TDA       
  Description: Dynamics, Non-adiabatic with cioverlap

  •  It seems to be only minor numerical differences. Probably installation is OK.

• Program: GAUSSIAN        

Method: TD-UDFT    

Description: Dynamics, Non-adiabatic with cioverlap

•  It seems to be only minor numerical differences. Probably installation is OK.

• Program: GAUSSIAN        

Method: TDDFT      

Description: Dynamics, Non-adiabatic with cioverlap_od

•  It seems to be only minor numerical differences. Probably installation is OK.

• Program: GAUSSIAN        

Method: TDDFT      

Description: Initial conditions

• It seems to be important differences. Something may be wrong in the installation.

• Program: GAUSSIAN/PYSOC  

Method: TDDFT      

Description: Initial conditions with spin-orbit couplings

• ERROR termination

•  Program: GAUSSIAN        

Method: TDDFT      

Description: Dyson orbitals

• It seems to be important differences. Something may be wrong in the installation.

• Program: GAUSSIAN        

Method: TDDFT      

Description: Photoelectron Spectrum

• NORMAL termination

•  Program: NEWTON-X        

Method: ANALYTICAL      

Description: Dynamics, Non-adiabatic with NAC vectors using model 1

• It seems to be only minor numerical differences. Probably installation is OK.

•  Program: NEWTON-X        

Method: TULLY 1         

Description: Dynamics, Non-adiabatic with NAC vectors using Tully model 1

• It seems to be only minor numerical differences. Probably installation is OK.

•  Program: NEWTON-X        

Method: SPIN-BOSON      

Description: Dynamics, Non-adiabatic with NAC vectors using Spin-Boson Hamiltonian

• It seems to be only minor numerical differences. Probably installation is OK.

•  Program: NEWTON-X        

Method: ANALYTIC CS-FSSH

Description: Dynamics, CS-FSSH using 2RHE Hamiltonian

•  It seems to be only minor numerical differences. Probably installation is OK.

•  Program: NEWTON-X        

Method: PICK-POINTS

Description: Initial conditions from a previous dynamics

•   NORMAL termination

•  Program: NEWTON-X        

Method: IMP_SAMP   

Description: Absorption spectrum with importance sampling

• NORMAL termination

•  Program: NEWTON-X        

Method: STATISTICS 

Description: Statistical analysis of dynamics

• NORMAL termination

Since I only have access to Gaussian 09, I had to change the gau.par file from 16 to 09 (which I think could be the reason why some tests fail or are slightly off). However, even for some of the purely Newton-X test cases, the results differ from the tests. I am wondering if it is typical to get " It seems to be only minor numerical differences. Probably installation is OK." 

Also, when I did the "Surface Hopping Dynamics: TD-DFT - Gaussian 09" tutorial (3 states, S0-S2), the NACs sometimes are all 0, 2/3 are 0, or 1/3 are 0. 

Nonadiabatic coupling terms v.h (a.u.):
     0.0000000
     0.0000000
     0.0000000
--
Nonadiabatic coupling terms v.h (a.u.):
     0.0000000
     0.0000000
    -0.0007995
--
Nonadiabatic coupling terms v.h (a.u.):
     0.0000000
     0.0000000
    -0.0019037
--
Nonadiabatic coupling terms v.h (a.u.):
     0.0000000
    -0.0009528
    -0.0021091

Is this expected? I know S1-S0 should be 0 because TD-DFT cannot properly do those couplings but should I be concerned that some steps have all 0s?

Thank you for your help,

Gabriel