[Wien] Limit of RmtKmax
Peter Blaha
pblaha at theochem.tuwien.ac.at
Thu Jul 27 13:59:32 CEST 2017
The value of the fermi energy is not a "proper" quantity in a
calculation of an infinite solid. It depends on the zero of energy, and
this is set to the mean interstitial potential. These changes are
unusual, but a priory no reason to worry.
You can see this shift of the energy-zero by looking at some core
eigenvalues:
grepline ':1s 001' '*scf' 1 will show you the same variations as EF.
PS: Linear dependency would either lead to a crash or at least a
significant change of total energy.
PPS: It is very valuable that you test for convergence. But you must
test these quantities which you want to publish. EF is not the right one.
------------------------
However, I would worry about the accuracy of your calculations in
general. You are using RMT values of 2.8 bohr (much larger than the
setrmt limit of 2.5 - which is already a compromise).
Large spheres are of course possible, but then you have to make sure
that all expansions are converged:
grep :FIT case.scf shows a value of 0.1E-01, indicating that the
nonspherical potential at large R is very large.
This requires a couple of changes in the default inputs:
LNSmax in case.in1 must be increased from 4 to 6 or even 8 (most severe
!! changes E-tot by several mRy)
The LM list (case.in2) should be enlarged. Adding the L=8 terms (see UG)
reduces :FIT by an order of magnitude.
Y-d states may have a linearization error. Add a HDLO for them.
Also: When varying RKMAX in such a large range, GMAX (in2) and the
IFFT-parameters (in0) may also have to be increased.
On 07/26/2017 06:47 PM, Arena Konta wrote:
> Dear Wien2k Users,
>
> How to check if my calculations are beyond "approximate numerical linear dependency"? The below calculations were converged with the criterion -ec 0.000001 and -cc 0.000001 on a fine mesh k-point to determine a good Fermi level energy accuracy (simple mettalic 3D compound cointaining Y and Al elements). No warning in scf file is appeared. From RKmax>=10 onwards the change in energy differences is smaller than 0.0002 Ry, hovewer I have problems with convergation the value of the Fermi Energy. What should I do to obtain better accuracy of EF?
>
> limit due to "approximate numerical linear dependency"
>
> 7
> :ENE : ********** TOTAL ENERGY IN Ry = -49933.78396516
> :DIS : CHARGE DISTANCE ( 0.000000 for atom 2 spin 2) 0.000001
> :FER : F E R M I - ENERGY(TETRAH.M.)= 0.4453031967
>
>
> 7.5
> :ENE : ********** TOTAL ENERGY IN Ry = -49933.79802481
> :DIS : CHARGE DISTANCE ( 0.000000 for atom 2 spin 1) 0.000001
> :FER : F E R M I - ENERGY(TETRAH.M.)= 0.4533274885
>
> 8.0
> :ENE : ********** TOTAL ENERGY IN Ry = -49933.80529502
> :FER : F E R M I - ENERGY(TETRAH.M.)= 0.4605712711
> :DIS : CHARGE DISTANCE ( 0.000000 for atom 2 spin 2) 0.000001
>
> 8.5
> :ENE : ********** TOTAL ENERGY IN Ry = -49933.80900587
> :DIS : CHARGE DISTANCE ( 0.000000 for atom 2 spin 2) 0.000001
> :FER : F E R M I - ENERGY(TETRAH.M.)= 0.4674101982
>
>
> 9.0
> :ENE : ********** TOTAL ENERGY IN Ry = -49933.81083697
> :DIS : CHARGE DISTANCE ( 0.000000 for atom 2 spin 2) 0.000001
> :FER : F E R M I - ENERGY(TETRAH.M.)= 0.4741770929
>
> 9.5
> :ENE : ********** TOTAL ENERGY IN Ry = -49933.81174098
> :DIS : CHARGE DISTANCE ( 0.000000 for atom 2 spin 2) 0.000002
> :FER : F E R M I - ENERGY(TETRAH.M.)= 0.4805577469
>
> 10
> :ENE : ********** TOTAL ENERGY IN Ry = -49933.81217067
> :DIS : CHARGE DISTANCE ( 0.000000 for atom 1 spin 1) 0.000001
> :FER : F E R M I - ENERGY(TETRAH.M.)= 0.4863678502
>
> 10.5
> :ENE : ********** TOTAL ENERGY IN Ry = -49933.81237951
> :DIS : CHARGE DISTANCE ( 0.000000 for atom 1 spin 1) 0.000001
> :FER : F E R M I - ENERGY(TETRAH.M.)= 0.4918538304
>
> 11
> :ENE : ********** TOTAL ENERGY IN Ry = -49933.81247632
> :DIS : CHARGE DISTANCE ( 0.000000 for atom 1 spin 2) 0.000001
> :FER : F E R M I - ENERGY(TETRAH.M.)= 0.4970660550
>
> 11.5
> :ENE : ********** TOTAL ENERGY IN Ry = -49933.81252520
> :DIS : CHARGE DISTANCE ( 0.000000 for atom 1 spin 2) 0.000001
> :FER : F E R M I - ENERGY(TETRAH.M.)= 0.5020619935
>
> 12
> :ENE : ********** TOTAL ENERGY IN Ry = -49933.81255122
> :DIS : CHARGE DISTANCE ( 0.000000 for atom 1 spin 1) 0.000001
> :FER : F E R M I - ENERGY(TETRAH.M.)= 0.5069889516
>
> 12.5
> :ENE : ********** TOTAL ENERGY IN Ry = -49933.81257047
> :DIS : CHARGE DISTANCE ( 0.000000 for atom 1 spin 1) 0.000001
> :FER : F E R M I - ENERGY(TETRAH.M.)= 0.5113533412
>
> 13.00
> :ENE : ********** TOTAL ENERGY IN Ry = -49933.81257716
> :DIS : CHARGE DISTANCE ( 0.000000 for atom 1 spin 2) 0.000001
> :FER : F E R M I - ENERGY(TETRAH.M.)= 0.5141763509
>
> 13.5
> :ENE : ********** TOTAL ENERGY IN Ry = -49933.81258348
> :DIS : CHARGE DISTANCE ( 0.000000 for atom 1 spin 2) 0.000001
> :FER : F E R M I - ENERGY(TETRAH.M.)= 0.5142562275
>
> 14
> :ENE : ********** TOTAL ENERGY IN Ry = -49933.81258360
> :DIS : CHARGE DISTANCE ( 0.000000 for atom 1 spin 2) 0.000001
> :FER : F E R M I - ENERGY(TETRAH.M.)= 0.5122039226
>
> 14.5
> :ENE : ********** TOTAL ENERGY IN Ry = -49933.81258849
> :DIS : CHARGE DISTANCE ( 0.000000 for atom 1 spin 2) 0.000001
> :FER : F E R M I - ENERGY(TETRAH.M.)= 0.5094474844
>
> 15
> :ENE : ********** TOTAL ENERGY IN Ry = -49933.81258989
> :DIS : CHARGE DISTANCE ( 0.000000 for atom 1 spin 2) 0.000001
> :FER : F E R M I - ENERGY(TETRAH.M.)= 0.5065405179
>
>
> Thank you
>
> Arena
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
>
--
P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-165300 FAX: +43-1-58801-165982
Email: blaha at theochem.tuwien.ac.at WIEN2k: http://www.wien2k.at
WWW: http://www.imc.tuwien.ac.at/TC_Blaha
--------------------------------------------------------------------------
More information about the Wien
mailing list