[DFTB-Plus-User] "Failure in diagonalisation routine dsygvd.." error

[Vinay Hegde]

Dear Jan,

Thanks!
Looking at the structure with that information, I found that a Zn atom
and an O atom were almost on top of each other.
So, I distorted the structure by manually putting them apart and
re-ran my calculation from that geometry. This continued for 15 steps.
Then they again got too close to each other (maximal force component
on the order of 10 a.u.) and again the same error occurred. How can I
prevent this (the atoms getting too close to each other) from
happening?

All I can do is manually separate them to a reasonable distance
(looking at other Zn-O bond lengths, etc.). Any idea what's causing
them to do this? And any option I have overlooked?

Thanks and regards,
Vinay

On 10 July 2013 19:55, Jan M. Knaup <Jan.Knaup at bccms.uni-bremen.de> wrote:
> Dear Vinay,
>
> your problem is most probably caused by two atoms coming too close to each
> other. The problem ist connected to the 1091st orbital in the system. You
> can find out which atom it belongs to by counting the orbitals of the atoms
> in your geometry ("s" atoms have 1 orbital, "p" atoms ave 4 and "d" atoms
> have 9).
>
> Changing miximg parameters or other parameters of the electronic calculation
> is not likely to help, because the converged electronic structure should
> always be the same. I suggest that you carefully check the last output
> geometry that is written by the program, it should give you some hint at
> what is going wrong with your calculations.
>
> Best,
> Jan
>
> Jan M. Knaup                      | Fon +49-(0)421-218-62351
> Dipl. Phys. Dr. rer. nat.         | Fax +49-(0)421-218-62770
> Universität Bremen - BCCMS        |
> Am Fallturm 1                     | Jan.Knaup at bccms.uni-bremen.de
> 28359 Bremen                      | JanKnaup at gmail.com
> Germany                           | www.bccms.uni-bremen.de
>
>
> 2013/7/9 Vinay Hegde <hegdevinayi at gmail.com>
>>
>> Dear users and developers,
>>
>> I am studying N_2 and CO_2 adsorption properties in a MOF using the
>> 'znorg-0-1' parameter set. All bulk calculations ran fine but I am
>> facing a strange error when performing constrained geometry
>> optimization for the (001) surface with ~20 Angstrom of vacuum.
>> Interestingly, in all the different configurations (I have 12-15
>> different configurations of the molecule on the MOF), this error
>> occurs exactly (this is the only reason I called it a strange error)
>> at the 31st geometry step:
>>
>>
>> ********************************************************************************
>>  ** Geometry step: 31
>>
>> ********************************************************************************
>>
>>    iSCC Total electronic   Diff electronic      SCC error
>>  ERROR!
>>  -> Failure in diagonalisation routine dsygvd, non-positive definite
>> overlap! Minor   1091 responsible.
>>
>> I have tried changing the mixing parameter, the mixer itself,
>> distorting the location of atoms a little and so on but the error
>> keeps occurring there: at the 31st geometry step.
>>
>> I have attached my dftb_in.hsd file and the standard output with this
>> mail. I am using DFTB+ 1.2 and the precompiled binary for my
>> calculations.
>>
>> Any suggestions or pointers are most welcome.
>>
>> Regards,
>> Vinay
>>
>> --
>>
>> Vinay Ishwar Hegde
>> Graduate Student
>> Theoretical Sciences Unit (TSU)
>> Jawaharlal Nehru Centre for Advanced Scientific Research (JNCASR)
>> Jakkur, Bangalore 560 064
>> Email: vinayhegde at jncasr.ac.in
>> Phone: +91-96639-92336
>>
>> _______________________________________________
>> DFTB-Plus-User mailing list
>> DFTB-Plus-User at dftb-plus.info
>> http://www.dftb-plus.info/mailman/listinfo/dftb-plus-user
>>
>
>
> _______________________________________________
> DFTB-Plus-User mailing list
> DFTB-Plus-User at dftb-plus.info
> http://www.dftb-plus.info/mailman/listinfo/dftb-plus-user
>

--

Vinay Ishwar Hegde
Graduate Student
Theoretical Sciences Unit (TSU)
Jawaharlal Nehru Centre for Advanced Scientific Research (JNCASR)
Jakkur, Bangalore 560 064
Email: vinayhegde at jncasr.ac.in
Phone: +91-96639-92336