[molpro-user] on Davidson correction

Kirk Peterson kipeters at wsu.edu
Wed Sep 16 20:19:50 BST 2015


Dear Somnath,

actually a reference function coefficient of 0.94 is ok and a contribution of about 35 mEh from the Davidson correction is not unusual.  Remember that much of this will cancel in any energy difference, so it’s not like it will make a 1 eV change on any energetics of interest.

best regards,

-Kirk


> On Sep 16, 2015, at 9:30 AM, Somnath BHOWMICK <somnath.bhowmick at etu.univ-amu.fr> wrote:
> 
> Dear Molpro users,
> 
> I am calculating excited states of lithium and fluorine clusters in MRCI theory. The active space is large enough (around 85000 CSFs) to describe the ground and few low-lying excited states. There is a large gain in correlation energy (about 7-8 eV) from the reference energy at the MRCI level. Moreover, the coefficient of the reference function in the MRCI wave function, i.e. c_n (as mentioned in https://www.molpro.net/info/2012.1/doc/manual/node310.html) is only around 0.94. This makes a huge gain in correlation energy (1 eV) in the Davidson correction term (for both fixed and rotated references) from the conventional MRCI energy. I was wondering if this is normal for most MRCI calculations or there is a limit of this energy difference or there exist some predefined conditions on the breakdown of Davidson correction.
> 
> Thank you in advance.
> 
> With very best of regards
> Somnath Bhowmick
> 
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> https://urldefense.proofpoint.com/v1/url?u=http://www.molpro.net/mailman/listinfo/molpro-user&k=EWEYHnIvm0nsSxnW5y9VIw%3D%3D%0A&r=ndj2Vta%2FisUMMOFOMobgvg%3D%3D%0A&m=hfPdY34JsTAvYYn2KnUB%2B5jrvjRJnnIhV6Fl4yXalWM%3D%0A&s=057c77cdcbb964380b39fc88d95bd2919e85fb7915945a37c235056fd0492f24

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20150916/ce703035/attachment.html>


More information about the Molpro-user mailing list