initialize a job

[jdgayles16]

I had trouble initializing for a bias calculation. I get this error

Contour Integration Error  : 0.018386

from this code

Code

###############################################################
# Initial State
###############################################################
device_configuration = nlread('analysis.nc', object_id="gID0")[0]

###############################################################
# Calculator
###############################################################
#----------------------------------------
# Numerical Accuracy Settings
#----------------------------------------
left_electrode_numerical_accuracy_parameters = NumericalAccuracyParameters(
    k_point_sampling=(1, 1, 100),
    )

right_electrode_numerical_accuracy_parameters = NumericalAccuracyParameters(
    k_point_sampling=(1, 1, 100),
    )

device_numerical_accuracy_parameters = NumericalAccuracyParameters(
    grid_mesh_cutoff=100.0*Rydberg,
    )

#----------------------------------------
# Electrode Calculators
#----------------------------------------
left_electrode_calculator = LCAOCalculator(
    numerical_accuracy_parameters=left_electrode_numerical_accuracy_parameters,
    )

right_electrode_calculator = LCAOCalculator(
    numerical_accuracy_parameters=right_electrode_numerical_accuracy_parameters,
    )

#----------------------------------------
# Device Calculator
#----------------------------------------
calculator = DeviceLCAOCalculator(
    numerical_accuracy_parameters=device_numerical_accuracy_parameters,
    electrode_calculators=
        [left_electrode_calculator, right_electrode_calculator],
    electrode_voltages=( 0.1*Volt, 0.0*Volt)
    )

device_configuration.setCalculator(calculator,
    initial_state=device_configuration,
)

device_configuration.update()
nlsave('analysis1.nc', device_configuration)
nlprint(device_configuration)

[Anders Blom]

Error or warning? Does the calculation stop?

[jdgayles16]

the calculation doesn't stop I just get similar error after every density matrix calc

[Anders Blom]

That's what I thought So basically this means that the contour integration isn't as precise as it wants to be. This is just a warning message, not an error message, it's an error estimate, we should perhaps reformulate the message a bit

We often see errors of 0.01, and that's no real problem, yours is a little bit higher, but probably it's not too serious.

You can experiment with increasing the number of points on the contour, and/or on the linear part (real_axis_point_density), see http://quantumwise.com/documents/manuals/latest/ReferenceManual/XHTML/ref.doublecontourintegralparameters.html for details.

[Anders Blom]

I did some digging in the source code, and actually this message is always printed, in each iteration, under finite bias. It's really just an information message with an unfortunate formulation, it should just say "Contour integration accuracy". 0.018 is a fine accuracy, nothing to worry about. (You can still improve it, by fiddling with the real space point density, but it's not really necessary at all.)

[jdgayles16]

Nice!! thanks , I should send you the voltage drop I got with the new, let me know if your not to busy.

[Anders Blom]

Send it, I'm curious to see!