[CP2K:4613] ALMO

Rustam rusta... at gmail.com
Thu Oct 24 22:15:45 UTC 2013


Yes, you can treat solute and solvent molecules using the ALMO methods. 
However, in the current implementation of the ALMO methods, every molecule 
(+its nearest neighbors) are treated by one core. Therefore, if your solute 
is too large to fit into the memory available to a core, the calculation 
will fail.

To get rid of the almo_max_av_neighbors increase the value of this variable 
in almo_scf_types.F and recompile the code. Alternatively, you can wait for 
a couple of days - we will provide a fix for this problem.


On Wednesday, October 23, 2013 7:27:12 PM UTC-7, Joe Greenstone wrote:
>
>
>
> > It will not work if boundaries between fragments cross covalent bonds.
>
> Can one leave the protein as a single fragment so that the scaling benefit 
> of ALMO is had for the solvent? When I do this, I get error messages 
> advising increase of almo_max_av_neighbors but I don't see a parameter with 
> this name in the current manual.
>
>
>
>  -------------------------------- ALMO SETTINGS 
> --------------------------------
>  eps_filter:                                                           
> 0.100E-08
>  blocked_eps_iter:                                                     
> 0.500E-03
>  blocked_max_iter:                                                           
> 100
>  blocked_n_diis:                                                               
> 7
>  delocalization:                                                             
> X_R
>  algorithm:                                              Blocked 
> diagonalization
>  delocalize_r_cutoff_factor:                                             
> 1.60000
>
>  -------------------------------------------------------------------------------
>
>      Lanczos converged:  F threshold:   0.100E-03
>      Est. extremal eigenvalues:    0.908E+01   0.491E-02
>      Est. condition number :    0.185E+04
>      NS sqrt iter    1 0.04928491   0.246E+01       2.725     8956.637
>      NS sqrt iter    2 0.04928491   0.117E+01       2.755     8862.305
>      NS sqrt iter    3 0.04928491   0.621E+00       3.223     7574.836
>      NS sqrt iter    4 0.04928491   0.360E+00       2.702     9034.791
>      NS sqrt iter    5 0.04928491   0.178E+00       3.004     8124.991
>      NS sqrt iter    6 0.04928491   0.935E-01       2.710     9008.010
>      NS sqrt iter    7 0.04928491   0.561E-01       2.722     8968.812
>      NS sqrt iter    8 0.04928491   0.234E-01       3.010     8109.299
>      NS sqrt iter    9 0.04928491   0.313E-02       2.694     9058.231
>      NS sqrt iter   10 0.04928491   0.100E-04       2.681     9102.792
>      Final NS sqrt iter   10 0.04928491   0.516E-12
>
>
>  ****************************************************************************
>  *** 12:29:46 ERRORL2 in almo_scf_special:almo_level2_spec1_0 :: increase 
> ***
>  *** almo_max_av_neighbors                                                
> ***
>
>  ****************************************************************************
>
>
>  *****************************************************************************
>  *** 12:29:46 ERRORL2 in almo_scf_special:almo_level2_spec1_0 processor 0  
> ***
>  *** :: err=-300 condition FAILED at line 814                              
> ***
>
>  *****************************************************************************
>
>  ===== Routine Calling Stack ===== 
>             5 almo_level2_spec1_0
>             4 almo_level1_exp1_0
>             3 almo_entry_scf
>             2 qs_energies_scf
>             1 CP2K
>  CP2K| condition FAILED at line 814
>  CP2K| Abnormal program termination, stopped by process number 0
>
>   
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20131024/37b33be6/attachment.htm>


More information about the CP2K-user mailing list