Using ADC(2)/MP2 simulation

[Derreck Wanlambok Nongspung]

Derreck Wanlambok Nongspung

24 Mar 2026, 18:21:58

to Newton-X

Dear Newton-X users,

I am currently trying to perform  S1 to S0 deactivation employing  ADC(2)/MP2 methodology. As ADC(2) is a single reference method...it is recommended to stop the dynamics when this energy gap becomes too small (and possibly continue only in the ground state)." 

After the trajectories stop when the S1-S0 energy gap is < 0.2 eV, how would I continue the simulation in the ground state with MP2?

Thank you very much for your time.

Regards,
Derreck WN
Research Scholar
Department of Chemistry
IIT Guwahati, Assam
781039, Indi

[Rafael Souza]

Dear Derreck,

It should be possible, but it is not very automated. You will need to:

- From the trajectory that stopped, create a new trajectory.

- Edit the input to change the number of states

- Edit Turbomole control file to change the method

The step-by-step is the following. Since I imagine that for the new MP2 calculation, you do not intend to compute the other states anymore, the trajectory should be continued from a new folder, such as:

1 - Go to the TRAJ folder that you want to continue
2 - nx_restart -s LAST_STEP

This LAST_STEP should be the timestep in which the trajectory stopped. This will create the folder "inputs_from_step_LAST_STEP".

3 -Copy this input folder to a new place so that you will remember it is a continuation of the original trajectory, for example, if you are in TRAJ1, then:
     cp inputs_from_step_LAST_STEP ../TRAJ1_B

Although it is a continuation, NX-NS will understand this as a new trajectory, and you have to adjust the input:

4 - rename the input file: 

      mv nx-restart-config.nml user_config.nml

5 - Replace the ipnuts parameters for the number of states:

      nstat = 1

      nstatdyn  =1

6 - Remove some restart information from the input: 

      delete lines with init_geom, init_veloc, init_wf, init_phase

7 - You are starting a simulation in a single state, and the coefficient information will not be important anymore, then delete the files

      rm phase.inp wf.inp

Do not touch the geom and veloc files, since they already contain the proper information from this restart point.

8 - Modify the control file in JOB_AD: in $ricc2 replace adc(2) by mp2.

This should be it. Although it may seem complicated, it should be simple to do and is scriptable.

At a later time, you will need to analyze the results. Ulamdyn will not recognize which folders compose a single trajectory, so you will have to read them separately and use scripting or a Jupyter notebook to either analyze the data separately or merge them. This cannot happen automatically because the output will contain multiple columns for all excited states in the initial part, and only columns for the ground state in the second part. Due to this difference in format, you will need to parse them separately.

I did an example with the files from the example folder in NX-NS. I'm adding it in the attachment. This doesn't make sense in this case, since I'm stopping the simulation at the very beginning and jumping from S2 to S0, but it should show the procedure.

Best regards,

Rafael