Negative numbers in density of states

[Roc]

Dear everyone,

Error happens in ATK caculation, as follows,


Traceback (most recent call last):
  File "AumolecularAuscript.py", line 291, in ?
    mulliken_population = calculateMullikenPopulation(self_consistent_calculation = scf)
NLTypeError: The population must contain non-negative numbers only.[/b]

that‘s to say, some negetive numbers happen in DOS, the .dat document is that,

# Density Of States
# -----------------------------------------------------------------------------
# Energy (eV)   DOS (1/eV)
        -4.00       0.0001
        -3.90       0.0002
        -3.80       0.0003
          ................
         3.60      -0.0012
         3.70      -0.0012
         3.80      -0.0016
         3.90      -0.0017
         4.00      -0.0016
how to deal with it?
Thanks!

Moderator edit: Changed subject to improve searching

[Nordland]

This one is a bit nasty - I guess it should not be able to happen, but in some way the convergence has gone haywire, but still managed to find a fix point.
You need to improve the quality of your twoprobe calculation by getting a more physical correct convergence.