[Smeagol-discuss] running example file

[Ivan Rungger]

Dear Asa,

  regarding the parallel computation problems:
1)The test systems are very small, so that they can only be run on a
small set of nodes. Try to run them (with the parallel version of the
code) on 1 and 2 processors, and see if it works. For more processors
there might not be enough orbitals in the system.
2)Some of the postprocessing (i.e. after convergence is reached)
subroutines of siesta-1.3 often crash in parallel. Therefore if an error
occurs after convergence is reached, you can remove one by one the
postprocessing steps, until it runs, in which case you then know which
of the options caused the code to crash.

For the serial part: I think that maybe there are not all the input
files needed by smeagol in the running directories, since all the
crashes are for smeagol calculations.

Cheers,

 Ivan


asa aravindh wrote:
> Dear Ivan,
> Thanks for your prompt reply.
> The KINDS variable sets is for the MPICH threading specific to our
> system architecture.
>
> I face problem running most of the examples in parallel and some of
> the examples in serial mode, provided with the smeagol distribution. 
> As I am a beginner, I haven't tested my own systems . First I am 
> trying to run the examples correctly, to learn the  parameter settings
> and all in detail. Unforutnately even the examples, which were already
> tested successfully by the developers, I am not able to run in my
> system. I will list out my experience in running all examples in
> serial and parallel mode.
> ----------------------------------------------------------------
> SERIAL RUNS
> ------------------------------------------------------------------
> Successful Runs:   Fe, SiH, MgO, H2O, H2O (order N) , Au leads
> calculation in the gamma directory, leads in the kpoints directory
> ----------------------------------------------------------------------------------------------------------------------------------------
> Failed Cases: Capacitor and Infinite calculations in the kpoints
> directory, Au wire calculation in the gamma directory
> The error is: program stops at the same point for all runs, at memory
> allocation, and not proceeding at all .
>
>  siesta:                 ===============================
> SMEAGOL                 Bias step =    0, V =  0.000 Ry
>                             Begin CG move =      0
>                         ===============================
>
> InitMesh: MESH =   108 x   108 x   192 =     2239488
> InitMesh: Mesh cutoff (required, used) =   200.000   203.052 Ry
>
> * Maximum dynamic memory allocated =   136 MB
> ---------------------------------------------------------------------------------------------------------------------------------------------
> ************************************************************************************************************
> PARALLEL RUN
> ************************************************************************************************************
> Successful Cases: H2O,H2O(ordr N), SiH,
> ----------------------------------------------------------------------------------------------------------------------------------------
> Failed Cases:  Fe, Auwire.fdf, Au leads ( both  in gamma directory),
> Capacitor, Infinite and Leads( in kpoints directory) , MgO
>
> error: In all calculations except for the Fe , program stops at the
> same place giving the following error.I am pasting part of one output
> here.
> -------------------------------------------------------------------------------------------
>
> siesta: System type = bulk
>
> siesta: k-grid: Number of k-points =    75
> siesta: k-grid: Cutoff             =     7.248 Ang
> siesta: k-grid: Supercell and displacements
> siesta: k-grid:    5   0   0      0.000
> siesta: k-grid:    0   5   0      0.000
> siesta: k-grid:    0   0   5      0.000
> siesta: overlap:  rmaxh   veclen  direction
> siesta: overlap:  18.5553  5.4786   1
> siesta: overlap:  rmaxh   veclen  direction
> siesta: overlap:  18.5553  5.4786   2
> siesta: overlap:  rmaxh   veclen  direction
> siesta: overlap:  18.5553  5.4786   3
>
> superc: Internal auxiliary supercell:     7 x     7 x     7  =     343
> superc: Number of atoms, orbitals, and projectors:    686  1715  4459
>
> * Maximum dynamic memory allocated =     2 MB
>
> siesta:                 ===============================
>                             Begin CG move =      0
>                         ===============================
>
> outcoor: Atomic coordinates (scaled):
>     0.00000000    0.00000000    0.00000000   1  Mg         1
>     0.50000000    0.50000000    0.50000000   2  O          2
>
> superc: Internal auxiliary supercell:     7 x     7 x     7  =     343
> superc: Number of atoms, orbitals, and projectors:    686  1715  4459
> p3_22547:  p4_error: interrupt SIGSEGV: 11
> p0_22067:  p4_error: interrupt SIGx: 13
> p0_22067: (9.328125) net_send: could not write to fd=4, errno = 32
> -------------------------------------------------------------------------------------------------------------------------------------
> For Fe, the program proceeds till  23 scf iterations, and then stops
> with following error
>
>
> siesta: E_KS(eV) =             -782.2432
>
> siesta: E_KS - E_eggbox =      -782.2432
>
> siesta: Atomic forces (eV/Ang):
> *** glibc detected ***
> /home/rocks2/smathi/test/smeagol.1.0b/Src/smeagol: free(): invalid
> next size (fast): 0x0000000016da2370 ***
> *** glibc detected ***
> /home/rocks2/smathi/test/smeagol.1.0b/Src/smeagol: free(): invalid
> next size (fast): 0x0000000008fd82a0 ***
> *** glibc detected ***
> /home/rocks2/smathi/test/smeagol.1.0b/Src/smeagol: free(): invalid
> next size (fast): 0x0000000008ecc3d0 ***
> p0_13772:  p4_error: interrupt SIGx: 13
> -----------------------------------------------------------------------------------------------------------------------------------------
>
> Kindly guide me ,
>
> Thanks and regards
> Asa
>


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

School of Physics and CRANN,  
Trinity College Dublin,  
Dublin 2,  IRELAND  
Phone: +353-1-8968454  
Email: runggeri at tcd.ie
=================================================