QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: FlyFlower on November 9, 2012, 08:20

Title: Is this result credible
Post by: FlyFlower on November 9, 2012, 08:20
hello, everyone, when i perform a two-probe calculation by using ATK-2008.10, at a bias of 1.0V, the self-consistent iteration is as follows:

# sc 117 : q =  668.98978 e  dRho =  3.4729E+00
# sc 118 : q =  669.07139 e  dRho =  1.5897E+00
# sc 119 : q =  669.13534 e  dRho =  3.4033E+00
# sc 120 : q =  669.13711 e  dRho =  7.8281E-03
# sc 121 : q =  669.14372 e  Etot = -7024.61255 Ry  dRho =  1.2260E-01
# sc 122 : q =  669.14696 e  Etot = -7024.61370 Ry  dRho =  8.1020E-01  dEtot = -1.1522E-03 Ry
# sc 123 : q =  669.15218 e  Etot = -7024.59930 Ry  dRho =  4.0362E-01  dEtot =  1.4403E-02 Ry
# sc 124 : q =  669.14950 e  Etot = -7024.59930 Ry  dRho =  2.9539E-01  dEtot = -3.6075E-06 Ry
1.0      8.41e-08

this current has a 2~3 orders of magnitude difference from those of bias on both sides,

voltage(V)     current(A)
0.8      1.03E-5
1.0      8.41E-8
1.2      6.73E-6

is this result credible? thanks very much!
Title: Re: Is this result credible
Post by: kstokbro on November 9, 2012, 10:06
It is difficult to tell from your information and ATK-2008.10 is no longer supported so we can not test it further.
I recommend that you perform the calculation we our latest version and compare the results,

Title: Re: Is this result credible
Post by: Anders Blom on November 9, 2012, 10:59
From just looking at the output it seems this calculation has not converged nicely. It does hit a point with small dEtot, but dRho is large, so it's more a coincidence that dEtot happens to be small perhaps. Also if you need over 100 iterations and still have dRho so large after 120 iterations, it should be possible to improve the convergence. But, as was mentioned above, this version is so old that we cannot support it even if we wanted to.