restarting a job

[jdgayles16]

Hi,
I have a problem where my job will stop calculating data but it looks like it is still running the iterations look like this.
   
   # sc 42 : q =  978.40985 e  dRho =  1.2337E-01
  # Mulliken Population for sc 42
  # sc 43 : q =  977.95166 e  dRho =  8.9629E-02
   # Mulliken Population for sc 43
   # sc 44 : q =  977.60230 e  dRho =  1.0240E-01
    # Mulliken Population for sc 44
    # sc 45 : q =  977.71758 e  dRho =  3.5024E-02
    # Mulliken Population for sc 45
     # sc 46 : q =  977.48554 e  dRho =  4.1458E-02
    # Mulliken Population for sc 46
    # sc 47 : q =  977.65571 e  dRho =  2.8185E-02
    # Mulliken Population for sc 47
    # sc 48 : q =  977.64904 e  dRho =  5.1071E-03
    # Mulliken Population for sc 48
     # sc 49 : q =  977.64654 e  Etot = -3412.73662 Ry  dRho =  6.9943E-03

 It stops calculating after the 49 iteration. I think it may be a memory problem with our cluster. My question is I still have the .nc file and i had trouble restarting the job and having print out the same way.

This is my script file
 
from ATK.TwoProbe import *
from ATK.MPI import processIsMaster

# Generate time stamp
if processIsMaster():
    import platform, time
    print '#',time.ctime()
    print '#',platform.node(),platform.platform()+'\n'

# Opening vnlfile
if processIsMaster(): file = VNLFile('twoprobe.vnl')








if processIsMaster(): nlPrint(two_probe_method)

# Restore self consistent calculation from check point file
oldscf = restoreSelfConsistentCalculation(
    filename = 'gc_0_4_bias.nc'
)

runtime_parameters = runtimeParameters(
    verbosity_level = 10,
    checkpoint_filename = 'gc_0_4_bias2.nc'
)

# Perform self-consistent field calculation
scf = executeSelfConsistentCalculation(
    twoprobe_configuration,
    two_probe_method,
    runtime_parameters = runtime_parameters,
    initial_calculation = oldscf,
)


Thanks Jacob

[Anders Blom]

You mean, when you restart nothing is printed to the terminal, but the program seems to run? If I recall correctly, this is because it probably needs to recalculate the electrodes "behind the scenes", and it doesn't produce any output. Thus it might start printing output once it gets back to the two-probe part. If not, you can try setting the verbosity more directly via

Code

import ATK
ATK.setVerbosityLevel(10)

This sometimes overrides the settings made via "runtime_parameters".