Re: [Phonopy-users] phono3py and large systems

[Antonio C. <cam...@fe...>]

Dear all,

I have a small update. I repeated the test using the smallest mesh 
possible (1x1x1) and I also tried the --br option instead of the --lw 
one but, in both cases, the calculation hangs at the same point. I also 
recompiled phono3py after enabling the macro

define_macros += [('MEASURE_R2N', None)]

to monitor the performances, as suggested by Prof. Togo, but it didn't 
produce any extra output. If this is related to the size of the system, 
I wonder if it is possible to modify the phono3py source or to specify 
different compilation flags in order to better handle big matrices.

Any help is greatly appreciated.

All the best

Antonio


On 18.8.2017 11:55, cammaant wrote:
> Dear all,
>
> I am trying to calculate the linewidht of only one band at 50K of a 
> 384-atoms system at a specific k-point using a 6x6x6 mesh. I do the 
> calculation on a 2x16-cores computation node with Intel Xeon E5-2682 
> v4 and 128GB RAM. I used the --cutoff-pair option with 8A cutoff, 
> obtaining 5159 displacements. phono3py has been compiled with 
> multithreding support and it works properly on a 96-atoms system with 
> P1 symmetry and no cutoff-pair set. The last lines of the standard 
> output are the following
>
> --------------------------------- Settings 
> ---------------------------------
> Mesh sampling: [ 6 6 6 ]
> Band indices: [ [1152] ]
> BZ integration: Tetrahedron-method
> Temperature:  50.0
> Grid point to be calculated: 0 108 122 14 111 3
> Cutoff frequency: 0.01
>
> after printing these, phono3py keeps running without producing any 
> output then it stops due to wallclock limit (4 days). I'm writing to 
> kindly ask:
>
> 1) is there any limitation on the number of atoms and/or the number of 
> displacements beyond which the computation hangs?
> 2) is it possible to compile phono3py with some options which allow to 
> print extra debugging messages and produce more output on the partial 
> progress of the computation?
>
> Thanks a lot in advance for your kind reply.
>
> My best regards
>
> Antonio Cammarata
>

-- 
_______________________________________________
Antonio Cammarata, PhD in Physics
Assistant Professor in Applied Physics
Advanced Materials Group
Department of Control Engineering - KN:E-s136
Faculty of Electrical Engineering
Czech Technical University in Prague
Karlovo Náměstí, 13
121 35, Prague 2, Czech Republic
Phone: +420 224 35 7442
Fax:   +420 224 91 8646