Geometry Optimization fails without proper messages

[Rad]

Thanks Teo, I was building the supercell using Mercury and ran into
all these errors. I set up a simple script to build them and I am ok
now.

On a different note, when I met Vallerie at CECAM he mentioned that
linear scaling O(n) algorithms are already available in CP at K download.
So when I say "USE_LINEAR_FIT CG" does CP2K invoke those efficient
algorithms for energy calculations or it is diiferent way of setting
up the input.

My understanding is that O(n) algorithms are still not available for
dynamics but useable for single point energy calculations that include
geometry optimization. Please clarify this for me.

Rad

On Oct 15, 3:42 pm, Teodoro Laino <teodor... at gmail.com> wrote:
> >  *************************************
> >  *** ERROR in build_neighbor_lists ***
> >  *************************************
>
> >  *** GEOMETRY wrong or EMAX_SPLINE too small! ***
>
> >  ===== Routine Calling Stack =====
>
> >             8 build_neighbor_lists
> >             7 build_fist_neighbor_lists
> >             6 topology_generate_bond_3
> >             5 topology_generate_bond
> >             4 connectivity_control
> >             3 topology_control
> >             2 quickstep_create_force_env
> >             1 CP2K
>
> > The key word EMAX_SPLINE is applicable only to the MM system as far as
> > I could see in the document.
>
> > Do you have any suggestions?
>
> the message says:
>
> > ** GEOMETRY wrong or EMAX_SPLINE too small! ***
>
> Since you're not doing an MM the geometry must be wrong.. and  
> indeed.. let's have a look together
> at particles 64 and 76:
>
> O      5.485      1.359      -2.189
> O      3.653      1.690      -3.321
>                                          O      5.110      6.158      
> -1.802 <=================
> O      3.817      7.944      -2.189
> [..]
> O      3.653      1.690      3.319
>                                         O      5.110      6.158      
> 4.838   <=================
> O      3.817      7.944      4.452
>
> You are using a cell of : 9.3027   9.3027   6.6403 angstrom..
> if you apply PBC particle 64 and 76 are simply the same particle. Are  
> you sure about the cell size???
>
> The reason why you get this message is because by default cp2k  tries  
> to generate a topology and
> to do that we use the neighbor lists of the classical module.
> In this case you've been lucky because this just showed your error in  
> setting up the coordinates..
>
> Cheers
> Teo
>
> p.s.: I posted the message on the group because in this way we keep  
> track of them and hopefully we
> won't repeat things N times ;-)
>
> On 15 Oct 2007, at 21:25, Rad wrote:
>
>
>
> > Teo,
>
> > I tried with SCALED key word for the fractional coordinates as well as
> > ANGSTROM for the orthogonal (not scaled) coordinates from Mercury.
> > The error I am getting is
> > CELL_REF| Grid size for subcell
> > generation                                2.000
> >  WARNING| Particles:      76     64 at distance [au]:     0.00056692
> > less than:      0.01889726; increase EMAX_SPLINE.
>
> >  *************************************
> >  *** ERROR in build_neighbor_lists ***
> >  *************************************
>
> >  *** GEOMETRY wrong or EMAX_SPLINE too small! ***
>
> >  ===== Routine Calling Stack =====
>
> >             8 build_neighbor_lists
> >             7 build_fist_neighbor_lists
> >             6 topology_generate_bond_3
> >             5 topology_generate_bond
> >             4 connectivity_control
> >             3 topology_control
> >             2 quickstep_create_force_env
> >             1 CP2K
> > [...]
>
> > The key word EMAX_SPLINE is applicable only to the MM system as far as
> > I could see in the document.
>
> > Do you have any suggestions?
>
> > Thanks
> > Rad- Hide quoted text -
>
> - Show quoted text -- Hide quoted text -
>
> - Show quoted text -