Setting the basis set in NAMD(3rd) calculation

[Kamal Dhungana]

Dear Pyxaid users,

In our third step calculation (NA-MD computations), we should choose Nmin, Nmax variables along with HOMO and LUMO to define the basis set. When I choose small Nmin and big Nmax, the simulation demands very large memory. My system has 400 electrons, and the HOMO level is 200. For my calculation, I choose Nmin=190, Nmax=220, I=200 and J=201 (excitation from Homo to Lumo). here, I am including (10 occupied orbitals). Can I change these numbers Nmin from 190 to 198 and Nmax from 220 to 210? Could someone shed light on this?

Best Regards,

Kamal 

[Alexey Akimov]

Dear Kamal,

Technically, you can. But make sure the results are not much sensitive to the size of the active space. So, I'd recommend doing a convergence studies - on how the # of intermediate states may affect the dynamics.

On the state definitions, I don't recommend using the "lazy" module - better define your states manually. It gives you a control of picking desired states and reducing the number of unnecessary states. 

If you have a few dozens of states in the step3, it should not be a memory problem, unless your trajectories are very long (you can still reduce the "namdtime" variable).

Also, it seems like your step2 generated you a bunch of the 220 x 220 (you max dimension) matrices. Did you know, you could reduce the size of these matrices during step2. That would save space and CPU time. Only, in the step3, you'd need to use different indices. 

Best,

Alexey