[molpro-user] Problem with simple DFT calculation: Ru atom

[Peter Knowles]

Dear Anirban,

I am sorry you have had this trouble with the default grid, which exists in some form for all elements heavier than krypton in the current version of Molpro.  It turned out that a code redesign has been necessary to resolve the problem, and this was completed some time ago, and will appear in the next release of the code.  I have been considering whether the change could be put back into 2010.1, but have concluded that it would not be very safe to do so, so for the moment I hope you can get by with forcing the grid in the way that you have suggested.

Peter

On 1 Sep 2011, at 19:19, Kirk Peterson wrote:

> Dear Anirban,
> 
> great, glad this is resolved.
> 
> -Kirk
> 
> 
> On Sep 1, 2011, at 11:18 AM, anirban mandal wrote:
> 
>> Dear Kirk,
>>  
>> Indeed it was due to different grid. I switched the grid parameters, and now I get the same energy as obtained with Gaussian and Gamess. Use the following grid along with uks,b-lyp.
>>  
>> grid;
>> radial,em,2,1,5,5,5,5,;
>> angular,lebedev,0.0,0.0;
>> lmin,29,47,53,53;
>> lmax,53,53,53,53;
>>  
>> That should give the same energy as with gamess and gaussian.
>>  
>> Best regards,
>>  
>> Anirban.
>>  
>> Anirban Mandal
>> Department of Chemistry
>> Michigan State University
>> East Lansing 
>> MI48824-1322
>> From: Kirk Peterson <kipeters at wsu.edu>
>> To: Gerald Knizia <knizia at theochem.uni-stuttgart.de>
>> Cc: anirban mandal <anirbanmandal2002 at yahoo.co.in>; molpro-user molpro-user <molpro-user at molpro.net>
>> Sent: Thursday, 1 September 2011 11:27 AM
>> Subject: Re: [molpro-user] Problem with simple DFT calculation: Ru atom
>> 
>> The states look identical to me, at least in terms of occupation and orbital character.  Besides since the "bad" energy is much lower, that would imply that both Gaussian and Gamess get the wrong state, which doesn't seem too likely.  So it seems that maybe the changes in the radial grids is the answer.  Do you know if there is a way to reproduce the new grids with 2010.1 ?
>> 
>> -Kirk
>> 
>> On Sep 1, 2011, at 3:27 AM, Gerald Knizia wrote:
>> 
>> > On 01.09.2011 02:21, Kirk Peterson wrote:
>> >> Dear Anirban,
>> >> 
>> >> actually at least on the surface it looks like a bug to me in 2010.1.  I get essentially the same value as Gaussian/Gamess with the developer's version but I get your incorrect value with the production one (2010.1).  Perhaps someone involved with the DFT code can comment?
>> > 
>> > There were some changes to the radial grids in the core region, and this can have significant effects on total energies. Also, could it be possible that both versions converge to different states? There were some changes to the occupations of degenerate orbitals in the initial guess.
>> > -- 
>> > Gerald Knizia
>> 
>> 
>> 
> 
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user

--
Prof. Peter J. Knowles             
School of Chemistry, Cardiff University, Main Building, Park Place, Cardiff CF10 3AT, UK
Email KnowlesPJ at Cardiff.ac.uk 
WWW http://www.cardiff.ac.uk/chemy/contactsandpeople/academicstaff/knowles.html
Secretary: Alison Thompson, ThompsonAF at Cardiff.ac.uk Telephone +44 29 208 74805