GW0 only completes 1
Posted: Wed Jul 22, 2015 7:57 am
Hi,
I'm hoping that someone can provide a little insight into why my GW0 (NELM=4) calculation fails/freezes after the first step. I'm able to complete the G0W0 calculation quickly (within 20 mins) so I would think doing a GW0 calculation is possible. I'm just trying to do a gamma-point calculation. Here are my input variables:
Here is the end of my output files:
As you can see, the first step completes successfully (G0W0); however, the calculation freezes when starting the 2nd step. I have tried increasing (2x) the number of nodes but I still have the same problem. In addition, there seems to be substantial memory available at this point.
I have the same problem when trying to do a scGW0 calculation.
FYI, I'nm using VASP-5.4.1
Thanks in advance for any suggestions
I'm hoping that someone can provide a little insight into why my GW0 (NELM=4) calculation fails/freezes after the first step. I'm able to complete the G0W0 calculation quickly (within 20 mins) so I would think doing a GW0 calculation is possible. I'm just trying to do a gamma-point calculation. Here are my input variables:
Code: Select all
ISTART = 1
NBANDS = 192
LPLANE = .TRUE.
NCORE = 16
PREC = Normal
LREAL = .FALSE.
ENMAX = 400.00
NELM = 4
ISYM = 0
ISMEAR = 0
SIGMA = 0.01
ALGO = GW0
NOMEGA = 100
Code: Select all
NQ= 1 0.0000 0.0000 0.0000,
|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........|.........
performed 1439000 updates of chi_q(r,r)
total number of BLAS operations 73916.93 Gflops
resolving degeneracies of screened two electron integrals
calculate QP shifts <psi_nk| G(iteration)W_0 |psi_nk>: iteration 1
shmem allocating 15 responsefunctions rank= 2544
response function shared by NCSHMEM nodes 1
Doing 1 frequencies on each core in blocks of 15
NQ= 1 0.0000 0.0000 0.0000,
Code: Select all
QP shifts <psi_nk| G(iteration)W_0 |psi_nk>: iteration 1
for sc-GW calculations column KS-energies equals QP-energies in previous step
and V_xc(KS)= KS-energies - (<T + V_ion + V_H > + <T+V_H+V_ion>^1 + <V_x>^1)
k-point 1 : 0.0000 0.0000 0.0000
band No. KS-energies QP-energies sigma(KS) V_xc(KS) V^pw_x(r,r') Z occupation
1 -22.4286 -33.3835 -26.6849 -16.5882 -26.9624 1.0850 2.0000 0.2084
2 -22.3997 -33.0948 -26.7707 -16.6552 -27.0174 1.0573 2.0000 0.1826
3 -22.3995 -33.1753 -26.7044
.
.
.
190 2.9146 5.8485 -2.0153 -4.9741 -1.9847 0.9916 0.0000 -0.0013
191 2.9162 5.8542 -2.0194 -4.9711 -1.9842 0.9954 0.0000 0.0002
192 3.0134 4.2175 -0.7690 -1.9960 -0.7392 0.9812 0.0000 0.0022
--------------------------------------------------------------------------------------------------------
QP_SHIFT: cpu time 0.4429: real time 0.4528
GWLOOP: cpu time 913.9981: real time 916.8295
shmem allocating 15 responsefunctions rank= 2544
response function shared by NCSHMEM nodes 1
all allocation done, memory is now:
total amount of memory used by VASP on root node 1719366. kBytes
========================================================================
base : 30000. kBytes
nonl-proj : 9312. kBytes
fftplans : 3203. kBytes
grid : 3103. kBytes
one-center: 150. kBytes
HF : 750. kBytes
nonlr-proj: 1065. kBytes
wavefun : 304. kBytes
2eintegral: 12. kBytes
response : 1671467. kBytes
--------------------------------------------------------------------------------------------------------
NQ= 1 0.0000 0.0000 0.0000,
KRAMKRO: cpu time 39.9149: real time 39.9221
I have the same problem when trying to do a scGW0 calculation.
FYI, I'nm using VASP-5.4.1
Thanks in advance for any suggestions