[CPMD-list] VDW / kpoints, the last episode ??

[Belabbas Imad]

Dear CPMD users,

Here my final conclusions on the application of VDW correction in the case of using k-points sampling. (for more detail on this issue have a look on my previous e-mail in the list)

The problem seems to be solved by adding some modification to the k_forces.F subriutine. These modifications are identical to those suggested by Axel (thanks Axel for the useful comments !!!!!!). I am attaching the corrected subroutine.

I have performed  single point calculations for the Acetonitrile crystal and I have obtained the following values : 

*********************************************************************
Crystal case:

Without VDW + gamma         Enr = -92.13954257 A.U.
Without VDW + kgrid111        Enr = -92.13954257 A.U.

With     VDW + gamma          Enr =  -92.17799191 A.U.       
With     VDW + kgrid111        Enr =   -92.17799191 A.U.
*********************************************************************

This time the results are coherent !! the values of the energy are identical if we consider the gamma point approximation or a kgrid 111. 

In the other hand I have also tried to optimize the structure of the crystal and compared the forces experienced by the atoms in each case. Here are the results for the first iteration and the 20th one :

VDW + gamma:

   ****************************************************************
  *** TOTAL STEP NR.    27           GEOMETRY STEP NR.      1  ***
  *** GNMAX=  1.124796E-01                ETOT=    -92.177992  ***
  *** GNORM=  6.601544E-02               DETOT=     0.000E+00  ***
  *** CNSTR=  0.000000E+00                TCPU=         52.45  ***
  ****************************************************************
 
    ****************************************************************
  *** TOTAL STEP NR.   403           GEOMETRY STEP NR.     20  ***
  *** GNMAX=  2.382693E-03 [6.14E-03]     ETOT=    -92.369401  ***
  *** GNORM=  1.071287E-03               DETOT=    -8.554E-05  ***
  *** CNSTR=  0.000000E+00                TCPU=         40.64  ***
  ****************************************************************
 
  
VDW + kgrid 111

  ****************************************************************
  *** TOTAL STEP NR.    29           GEOMETRY STEP NR.      1  ***
  *** GNMAX=  1.124796E-01                ETOT=    -92.177992  ***
  *** GNORM=  6.601537E-02               DETOT=     0.000E+00  ***
  *** CNSTR=  0.000000E+00                TCPU=        133.37  ***
  ****************************************************************
 
    ****************************************************************
  *** TOTAL STEP NR.   406           GEOMETRY STEP NR.     20  ***
  *** GNMAX=  3.087134E-03 [6.38E-03]     ETOT=    -92.369356  ***
  *** GNORM=  1.178630E-03               DETOT=    -1.247E-04  ***
  *** CNSTR=  0.000000E+00                TCPU=         97.30  ***
  ****************************************************************
For the first iteration, the forces are identical in both cases (gamma, kgrid111). However, at the 20th iteration some small differences occur in the forces. I guess this due to numerical errors !!!. This not very important.

One final remark, in the present implementation VDW.F is called each SCF cycle !! it should only called one time if the geometry is fixed. This is not a serious problem as the numerical effort needed to evaluate the VDW energy, forces and stress is not important compared to the rest.

In conclusion, the VDW correction works now in the case of k-points sampling !!

Cheers,
Imad




 
  
             
---------------------------------
 Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! Mail 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://cpmd.org/pipermail/cpmd-list/attachments/20080110/0b29d987/attachment.html 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: k_forces.F.gz
Type: application/postscript
Size: 5609 bytes
Desc: 3927473760-k_forces.F.gz
Url : http://cpmd.org/pipermail/cpmd-list/attachments/20080110/0b29d987/attachment.ps 
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: Compare-forces.txt
Url: http://cpmd.org/pipermail/cpmd-list/attachments/20080110/0b29d987/attachment.txt