MPB-errors

Hello, all

I usually have no problem with simulation using mpb-mpi (version 1.4.2) on
my PC but I tryed to run simulation on another PC (mpb-mpi version 1.4.2) and
I got some error massages that  I'v never had before. Below you may see the
text from .out file.

I set 19 k-points from (0 0 1.7) to (0 0 2.6) but in .out file we could see
the list of 19 zero k-points. Anyway then it starts calculation for proper
k-point (0 0 1.7).
After it finish calculation for first 11 bands calculation will be aborted.

My other observations are:
1) If I set lattice size with integer value (for example (9 7 no-size)
instead of (9.6 7.8 no-size)) it will calculate without aborting for proper
k-points but still in .out file there will be list of 19 zero k-points.

2) If I set calculation of bands for only one k-point  (0 0 1.7) mpb will
give me in .out file in the list of k-points "1 k-point (0, 9.08701e+223,
5.60633e+199)" but then will start calculation for proper
k-point "solve_kpoint (0,0,1.7):"

TEXT from .out file

init-params: initializing eigensolver data
Computing 500 bands with 1.000000e-05 tolerance.
Working in 2 dimensions.
Grid size is 618 x 504 x 1.
Solving for 11 bands at a time.
Creating Maxwell data...
Allocating fields...
Mesh size is 3.
Lattice vectors:
     (9.6, 0, 0)
     (0, 7.8, 0)
     (0, 0, 1)
Cell volume = 75.9689
Reciprocal lattice vectors (/ 2 pi):
     (0.10364, -0, 0)
     (-0, 0.12701, -0)
     (0, -0, 1)
Geometric objects:
Geometric object tree has depth 1 and 0 object nodes (vs. 0 actual objects)
Initializing dielectric function...
Using background dielectric from file "SEM.h5"...
    ...read 375x306x1 dielectric function
19 k-points:
     (0,0,0)
     (0,0,0)
     (0,0,0)
     (0,0,0)
     (0,0,0)
     (0,0,0)
     (0,0,0)
     (0,0,0)
     (0,0,0)
     (0,0,0)
     (0,0,0)
     (0,0,0)
     (0,0,0)
     (0,0,0)
     (0,0,0)
     (0,0,0)
     (0,0,0)
     (0,0,0)
     (0,0,0)
Solving for band polarization: .
Initializing fields to random numbers...
elapsed time for initialization: 11 seconds.
epsilon: 1-2.0851, mean 1.13485, harm. mean 1.07681, 19.736% > 1, 12.4272%"fill"
Outputting epsilon...
solve_kpoint (0,0,1.7):
freqs:, k index, k1, k2, k3, kmag/2pi, band 1, ..., band 500
Solving for bands 1 to 11...
1: trace = 123.21663832506 (193.098% change)

....
38: trace = 25.9645213056759 (0.000952806% change)


Finished solving for bands 1 to 11 after 38 iterations.
Solving for bands 12 to 22...
CHECK failure on line 243 of blasglue.c: non positive-definite matrix in
potrf
[31] MPI Abort by user Aborting program !
[31] Aborting program!



what means  "CHECK failure on line 243 of blasglue.c: non positive-definite
matrix in potrf
"

As I understood I should recompile mpb-mpi. And what recommendations you can
give me to except these errors?

_______________________________________________
mpb-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/mpb-discuss