Poisson solver for non cubic cells not yet implemented

[Megha Anand]

Dear All,

I am trying to reproduce the results from this manuscript http://pubs.acs.org/doi/abs/10.1021/jo3023439

I tried my best to follow the computational methods in this manuscript specifying the cell dimensions as 24 × 15 × 15 Å and suggestions by Luca in this post:

https://groups.google.com/forum/#!searchin/cp2k/ts$20search%7Csort:relevance/cp2k/58O-DVV2OAU/v_CWUYyQMwAJ

However, I am getting the error: 

 ***************************************************************************

 *** 22:23:55 ERRORL2 in ps_wavelet_types:ps_wavelet_create processor 0  ***

 *** :: err=-300 Poisson solver for non cubic cells not yet implemented  ***

 ***************************************************************************

Can someone help me with this error. I do not really know how to decide the parameters for the MT poisson solver. Is it because I am using the older version of CP2K?? Any insights will be greatly appreciated.

Thank you so much for your time and help in advance!!

Best regards,

Megha Anand

Post-doctoral Associate

Dept. of Chemical Engineering, MIT

Cambridge, MA 02140

[Matt W]

Hi,

that paper used the Martyna-Tuckerman (MT)  algorithm to solve the Poisson equation. This is selected in the CP2K_INPUT / FORCE_EVAL / DFT / POISSON / PSOLVER keyword.

Your error message shows that you selected the wavelet solver instead, which is a different algorithm for doing the same thing (but as you discovered, requires a cubic cell, currently).

HTH,

Matt

[Megha Anand]

Thanks Matt! 

Will you please answer one more question: How did the authors in the current manuscript came with the cell dimensions of 24 × 15 × 15 Å? My research has not been moving because I am not able to use CP2K properly. I greatly appreciate you help!

Thanks,

Megha

[Matt W]

He Megha,

this isn't really about CP2K, but about the MT Poisson solver method. The original paper, which gets cited when you run with MT, shows that you need a box twice the size of the electron density of the system to be reliable.

In this case it looks like they have ( the size of the molecule + a little bit of extra room for the electron density to decay ) * 2  in each direction, which is consistent with what the original paper says. 

Matt

[Megha Anand]

Thanks Matt! 

I get the electron density from the E_DENSITY_CUBE subsection in the FORCE_EVAL / DFT / PRINT section. Am I right? 

I have attached a cube file with electron density that I got from one of the computations. How do I proceed from this point to obtain the size of the electron density (and then decide the box size). My apologies if the question is too trivial.

Thanks a lot for your time and help!

Megha

--
You received this message because you are subscribed to a topic in the Google Groups "cp2k" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/cp2k/zrqMHB9F_qM/unsubscribe.
To unsubscribe from this group and all its topics, send an email to cp2k+unsubscribe@googlegroups.com.
To post to this group, send email to cp...@googlegroups.com.
Visit this group at https://groups.google.com/group/cp2k.
For more options, visit https://groups.google.com/d/optout.