[Smeagol-discuss] Still problems with ZGETRF error

[Ivan Rungger]

Dear Zhiping,

   the two problems that you have are related. In the output of the SVD
it says that the rank of H1 and S1 is 0, I checked your system and saw
that in your case H1=S1=0. This means that the unit cells in the leads
are not connected along z (in fact, also in the leads calculation siesta
says that the system type is slab), i.e. the distance between the slabs
is too large given the basis cutoff you use. If you increase the length
of your basis set, then the calculations will work.

Cheers,

 Ivan

zx3 at rice.edu wrote:
> Dear all:
>
> (1)    I tried to study the conductance between graphene planes, where both the
> electrodes and conductor are graphene layers (two plane, A B stacking). (It's
> not metal, so this model maybe questionable, but i think it doesn't have
> relation wit the ZGETRF error I want to solve).
>     Using 10x10x100 k-points sampling good convergence and charge density was
> obtained in the BulkTransport calculation. Later EMTransport is performed,
> where 4 unit cells (4 A-B bilayers) was used for the scattering part. K-points
> sampling was chosen to be 10x10x1. However error was reported as
> --------------------------------------------------------------------------
> " ** On entry to ZGETRF parameter number  4 had an illegal value" from lapack
> --------------------------------------------------------------------------
> provided by Ubuntu resposity or
> --------------------------------------------------------------------------
> "MKL ERROR : Parameter  4  was incorrect on entry to  ZGETRF
> MKL ERROR : Parameter  4  was incorrect on entry to  ZGECON
> MKL ERROR : Parameter  13  was incorrect on entry to  ZGESVD
> MKL ERROR : Parameter  8  was incorrect on entry to  ZGEMM
> ..."
> --------------------------------------------------------------------------
> from MKL lapack.
>     Thanks to previous discussion between Simon, Ivan and Victor, I know this
> should be a supercell mismatch problem. From the output I knew 3X3X3 and 3X3X1
> super cell was used for the Bulk and EM Transport respectively. I am quite
> confusing that my system is quite similar the Example for Au-Infinite-K
> provided by Smeagol expected for the electronic structure. In the example 5x5X3
> and 5X5X1 was used for the supercells and it runs properly.
>     I change k-points sampling in EMTransport to 10x10x2 but it doesn't work.
> (2)    Also I want to know why there is gensvd output as in my system which does
> not show up in the Au-infinite-K example
> ----------------------------------------------
> gensvd: Leads decimation
> gensvd: Dim of H1 and S1 :     52
> gensvd: Rank of H1:             0
> gensvd: Rank of (H1,S1):        0
> gensvd: Decimated states:      52
> gensvd: Decimation from the left
> ----------------------------------------------.
>
>
>   


-- 
=================================================
Ivan Rungger,

School of Physics,  
Trinity College Dublin,  
Dublin 2,  IRELAND  
Phone: +353-1-6088454  
Email: runggeri at tcd.ie

=================================================