implicit solvation model calculation was much slower than the explicit solvation model

[Jessie wang]

Dear all,

I hope to use the implicit solvation model to perform geo_opt for a metal cluster with ligand molecules.  But it takes much more time than the explicit solvation model for any SCF calculation. Is this normal or did I do something wrong?  I would be grateful if you could give me a suggestion. 

Here I have attached my input files for both calculations. For the explicit model, I added about 690 water molecules to the box for the explicit calculation. For the explicit calculation I needed 32 seconds per scf, for the implicit calculation 78 seconds per scf. All other settings are the same and the same calculation resources were used.

Best,

Jessie

[Krack Matthias]

Hi Jessie

 

That can happen and Is often due to a different convergence behavior. Also note that the implicit solvent comes not for free, because it introduces another convergence cycle for the polarization charge in each SCF iteration step which can add a significant computational overhead.

 

HTH

 

Matthias

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/5f533b72-8445-4c5f-9167-11367e359e6an%40googlegroups.com.

[Jessie wang]

Thank you so much for providing your response and your generous sharing. While I understand that performing a self-consistent field (SCF) calculation for implicit solvent can be time-consuming, I am still uncertain as to why it took significantly longer than the explicit method. If the implicit solvent model requires more time than the explicit method, what is the rationale for using this model?

[Krack Matthias]

For systems with a large solvent portion and a robust SCF convergence behavior, runs with an implicit solvent should be faster. An implicit solvent can also be advantageous, if long sampling times are required for properly averaging the fluctuations within an explicit solvent.

.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/5f781632-bb60-408c-8235-13caf54398f6n%40googlegroups.com.