Automatic termination during epsilon calculation

[Bill Ash]

When I get to epsilon, the program epsilon.cplx.x terminates.Here are my input and output files：

epsilon.inp

# epsilon.inp
epsilon_cutoff           8.0
number_bands             178
cell_slab_truncation
frequency_dependence 0
degeneracy_check_override
begin qpoints
  0.001000000 0.000000000 0.000000000 1.0 1
  0.000000000 0.250000000 0.000000000 1.0 0
  0.000000000 0.500000000 0.000000000 1.0 0
  0.000000000 0.750000000 0.000000000 1.0 0
  0.250000000 0.250000000 0.000000000 1.0 0
end

epsilon.out

Running MPI version (parallel)
 Running with 48 MPI task(s)
 Using OpenMP. Number of threads per MPI task: 2

 Compilation flags:
 - Compiler: INTEL
 - Para. flags: MPI, OMP
 - Math  flags: USESCALAPACK, USEFFTW3, HDF5
 - Debug flags:

 Running with verbosity level 1 (default).

 We will start a new calculation from scratch.

 
 We`ll perform a uniform sampling of the full BZ.
 
 Computing the static inverse dielectric matrix

 We are using matrix communication scheme

 We are communicating via MPI

 We are using a truncated Coulomb interaction: cell slab
 Reading header of WFN
 Highest occupied band (unshifted grid) = 40
 Valence max    (unshifted grid) = -5.369403 eV
 Conduction min (unshifted grid) = -3.121088 eV
 Middle energy  (unshifted grid) = -4.245245 eV
 Fermi  energy  (unshifted grid) = -4.245245 eV

 Reading header of WFNq
 Highest occupied band (shifted grid) = 40

 Calculation parameters:
 - Cutoff of the dielectric matrix (Ry): 8.00
 - Total number of bands in the calculation: 178
 - Number of fully occupied valence bands: 40
 - Number of partially occ. conduction bands: 0
 - Monkhorst-Pack q-grid for epsilon(q): 4 4 1
 
 Running with 4 valence pools
 Number of conduction bands per processor: 12
 Number of valence bands per processor: 10

 Memory available: 7862.9 MB per PE
 Memory required for execution: 47.8 MB per PE
 Memory required for vcoul: .0 MB per PE

 Number of electrons per unit cell (from ifmax) = 40.000000
 Number of electrons per unit cell (from occupations) = 40.000000
 Plasma Frequency = 1.140342 Ry

 
 Started reading wavefunctions (WFN) with 890 state(s) at 16:03:58.
 [ 16:03:58 |   0% ] state   1 / 890.
 [ 16:03:58 |   0% ] state   3 / 890, remaining: 0 s.
 [ 16:03:58 |  10% ] state  89 / 890, remaining: 1 s.
 [ 16:03:58 |  20% ] state 178 / 890, remaining: 1 s.
 [ 16:03:59 |  30% ] state 267 / 890, remaining: 1 s.
 [ 16:03:59 |  40% ] state 356 / 890, remaining: 1 s.
 [ 16:03:59 |  50% ] state 445 / 890, remaining: 1 s.
 [ 16:03:59 |  60% ] state 534 / 890, remaining: 0 s.
 [ 16:03:59 |  70% ] state 623 / 890, remaining: 0 s.
 [ 16:03:59 |  80% ] state 712 / 890, remaining: 0 s.
 [ 16:03:59 |  90% ] state 801 / 890, remaining: 0 s.
 [ 16:03:59 | 100% ] state 890 / 890, remaining: 0 s.
 Finished reading wavefunctions (WFN) at 16:03:59.
 Elapsed time: 1 s.
 
 Initializing eps0mat.h5
 Initializing epsmat.h5
 - Screened Coulomb cutoff: 8.000 Ry
 - Total number of bands: 178
 - Number of q-points: 5

 Scissors parameters:
 - Valence:    es =   0.0000 eV, e0 =   0.0000 eV, edel =   0.0000
 - Conduction: es =   0.0000 eV, e0 =   0.0000 eV, edel =   0.0000

 Q0 and |Q0| =  0.001000  0.000000  0.000000  0.001158
 Q  and |Q|  =  0.000000  0.250000  0.000000  0.289478
 Q  and |Q|  =  0.000000  0.500000  0.000000  0.578955
 Q  and |Q|  =  0.000000  0.750000  0.000000  0.868433
 Q  and |Q|  =  0.250000  0.250000  0.000000  0.501390

 ==================================== WARNING ==================================
 ==
 You have enabled spin polarization or full spinor calculations.
 Current formalism and implementation adopted by BerkeleyGW implicitly assumes
 time-reversal symmetry (TRS), e.g. in the standard Adler-Wiser RPA polarizabili
 ty.
 Running magnetic/broken-TRS systems is fundamentally against these assumptions.
 One needs to be aware of and justify its direct usage. Use at your own risk.
 ===============================================================================
 ==
 GPU acceleration is : DISABLED
 
 Algorithms used:
 - mtxel         : CPU_ALGO
 - chi_summation : CPU_ALGO
 
 You have a slightly shifted q0 vector and a semiconductor.
 So, reading from WFNq.
 Reading header of WFNq
 Highest occupied band (shifted grid) = 40
 
 Started reading wavefunctions (WFNq) with 400 state(s) at 16:03:59.
 [ 16:03:59 |   0% ] state   1 / 400.
 [ 16:03:59 |   1% ] state   3 / 400, remaining: 0 s.
 [ 16:04:00 |  10% ] state  40 / 400, remaining: 1 s.
 [ 16:04:01 |  20% ] state  80 / 400, remaining: 4 s.
 [ 16:04:02 |  30% ] state 120 / 400, remaining: 6 s.
 [ 16:04:06 |  40% ] state 160 / 400, remaining: 10 s.
 [ 16:04:06 |  50% ] state 200 / 400, remaining: 7 s.
 [ 16:04:07 |  60% ] state 240 / 400, remaining: 5 s.
 [ 16:04:07 |  70% ] state 280 / 400, remaining: 3 s.
 [ 16:04:09 |  80% ] state 320 / 400, remaining: 2 s.
 [ 16:04:09 |  90% ] state 360 / 400, remaining: 1 s.
 [ 16:04:09 | 100% ] state 400 / 400, remaining: 0 s.

[Mauro Del Ben]

Hi Bill,

Are you sure there are no error messages? If you're running on a cluster it's possible the error messages are printed in a different file (often with an extension *.e)

Best

-M

--
You received this message because you are subscribed to the Google Groups "BerkeleyGW Help" group.
To unsubscribe from this group and stop receiving emails from it, send an email to help+uns...@berkeleygw.org.
To view this discussion visit https://groups.google.com/a/berkeleygw.org/d/msgid/help/26ce9c13-78d4-40cb-9966-bc7f0ebcfa35n%40berkeleygw.org.

[Bill Ash]

Hi M,

Thank you for your reply. Here are my epsilon input and output files, where the error file doesn't output anything.

Looking forward to your reply.

Best

Bill