Marcos, 

thank you for your advice.

I've tried with Diag.DivideAndConquer F  and using zeros for the rc's and it 
doesn´t work. I will try with the energy shift.

Regards,

Silvia

El 06/06/2011, a las 17:09, Marcos Veríssimo Alves escribió:

> Silvia,
> 
> It's a rather weird error. However, there's two things you can try:
> 
> 1) Use the basis set downloaded from the siesta website, but use 
> 
> Diag.DivideAndConquer        .false.
> 
> If I'm not mistaken, most of my Cholesky bugs disappear when I disable the 
> divide and conquer routines. If this doesn't work, delet it from the fdf file 
> and
> 
> 2) try using the semicore states, but with zeros for the rc's, no ionic 
> charge and no soft confinement, for simplicity:
> 
> %block PAO.Basis
> Mn   5    
>  n=3   0   1   
>      0.00000
>      1.00000
>  n=4   0   2   
>      0.00000     0.00000
>      1.00000     1.00000
>  n=3   1   1   
>      0.00000
>      1.00000
>  n=4   1   1   
>      0.00000
>      1.00000
>  n=3   2   2   
>      0.00000     0.00000
>      1.00000     1.00000
> %endblock PAO.Basis
> 
> and if this doesn't work, as well,
> 
> 3) try playing with the energy shift, increasing it a bit while using the 
> fixed Mn basis set from the siesta website.
> 
> Best of luck,
> 
> Marcos
> 
> On Mon, Jun 6, 2011 at 4:49 PM, silvia <[email protected]> wrote:
> Sorry,
> 
> this is the correct fdf file
> 
> 
> 
> El 06/06/2011, a las 16:17, Marcos Veríssimo Alves escribió:
> 
>> Silvia,
>> 
>> There is too little information on your problem for anyone to say anything 
>> meaningful. But, no, in principle there are no such problems - I am doing 
>> calculations using Siesta LDAU using semicore states and U and I get no such 
>> error.
>> 
>> Marcos
>> 
>> On Mon, Jun 6, 2011 at 4:08 PM, silvia <[email protected]> wrote:
>> Dear siesta users,
>> 
>> I'm trying to do a spin polarized calculation of a manganese dinuclear 
>> complex with SIESTA LDAU version.  When I use semicore states and the U in 
>> the manganese atom I get this error:
>> 
>> Initializing Density Matrix...
>>  DM norm at end of initdm:    221.999999150235
>>  DM norm after initdm:    221.999999150235
>>  DM norm before normalization:    221.999999150235
>>  Qtot:    221.999999999999
>> LDAU: recalculating local occupations    1
>> LDAU: recalculating Hamiltonian
>> LDAU: Adding Hamiltonian contribution
>> 
>> InitMesh: MESH =   180 x   144 x   144 =     3732480
>> 
>> Error in Cholesky factorisation in rdiag
>> Error in Cholesky factorisation in rdiag
>> 
>> Do you know if there are any compatibility problems in the use of semicore 
>> states and the U parameter for the same atom or in the definition of the 
>> basis in this SIESTA version?
>> 
>> Best regards,
>> 
>> Silvia
>> 
>> 
>> 
> 
> 
> 

Responder a