Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

APW missing energy parameters #126

Open
ndaelman-hu opened this issue Jun 15, 2023 · 4 comments · May be fixed by #139
Open

APW missing energy parameters #126

ndaelman-hu opened this issue Jun 15, 2023 · 4 comments · May be fixed by #139
Assignees
Labels
bug Something isn't working. It also represents a quick fix in response to a bug. feature / enhancement New feature or request

Comments

@ndaelman-hu
Copy link
Contributor

The energy_parameter in the APW codes is reported as an undefined vector. This has been verified for exciting and Wien2k, whereas fleur does not apply.

@ndaelman-hu ndaelman-hu added the bug Something isn't working. It also represents a quick fix in response to a bug. label Jun 15, 2023
@ndaelman-hu ndaelman-hu self-assigned this Jun 15, 2023
@ndaelman-hu
Copy link
Contributor Author

The problem was traced back to the NOMAD datamodel, where energy_parameter was defined as a vector quantity instead of a scalar.

@ndaelman-hu ndaelman-hu added the feature / enhancement New feature or request label Jun 15, 2023
@ndaelman-hu
Copy link
Contributor Author

Let's use this issue to

  • set up some tests for the APW parsers
  • doublecheck the Wien2k interpretation
  • decide on whether to unroll by order too

@ndaelman-hu
Copy link
Contributor Author

Regarding local orbitals, there are two types, each with similar boundary conditions 12:

  • LO: is a variation on SLAPW-3, with 2 $u \left(r, \epsilon_i \right)$ wavefunction and 1 $\dot{u}$ wavefunction.
    The $u$ wavefunctions are aligned to 2 different energy windows, one to a regular valence orbital of the same l-channel and the semicore state. This is typical of LAPW.
  • lo: this version has 1 $u$ and 1 $\dot{u}$ wavefunction. It is used in APW+lo.

So, this should be updated in the orbital and general overview.

Fleur, meanwhile, lists other local orbitals, namely SCLO (and those with orders > 0 called HDLO), as well as HELO. 3
For so far I can tell, SCLO corresponds to the listing above (it is created for semicore states), though it is unclear whether this is LO or lo. Based on the value range of the order parameter ($[0, +\infty [$), my guess is LO: there's always a zero-order in lo and LO, but in the latter the 2nd $u$ wavefunction can also be of order 0.

HELO is more so geared towards unoccupied orbitals, differing only in the boundary conditions to be met. 34

My takeaway is that the fleur noemclature is the most detailed, to the point that i may become niche to other APW users. Moreover, the automated determination of the energy parameter with HELO's is only sensible for fleur. I'm still unsure on which terminology to settle.

Footnotes

  1. D. Singh, “Ground-state properties of lanthanum: Treatment of extended-core states,” Phys. Rev. B, vol. 43, no. 8, pp. 6388–6392, Mar. 1991, doi: 10.1103/PhysRevB.43.6388.

  2. D. K. Schwarz, “An Augmented Plane Wave + Local Orbitals Program for Calculating Crystal Properties”.

  3. https://www.flapw.de/MaX-6.0/documentation/localOrbitalSetup/ 2

  4. M. Betzinger, C. Friedrich, S. Blügel, and A. Görling, “Local exact exchange potentials within the all-electron FLAPW method and a comparison with pseudopotential results,” Phys. Rev. B, vol. 83, no. 4, p. 045105, Jan. 2011, doi: 10.1103/PhysRevB.83.045105.

@ndaelman-hu
Copy link
Contributor Author

From today's discussion with another exciting expert:

  • Terminology used is FP-APW
  • Read the species file and not the species SCF file, since the latter is only produced when searchE is enabled(?). The only significant difference are the energy parameters.
  • Each LO can be fully customized using the <wf> tags, this includes the derivative order $\partial^n u / \partial\epsilon^n$ (matchingOrder) and the energy parameter (trialEnergy)
    • So, lo (default with apw+lo) and LO can be recreated, but one can go beyond.
    • Different LO's of the same l-channel live under different <lo> tags.
    • Then it's better to provide order and energy_parameter as vectors, and maybe still have an abbreviation when it matches the common terminology.
  • The searchE option is very unstable and not recommended: it updates the energy parameters at each SCF cycle (or step?). Overall, just a sensible value is good enough. Essentially, there's a confidence interval that can be applied.
  • Boundary conditions can be further tweaked, but are overall less important for comparisons. Await example here.
  • Meanwhile, the electronic configuration (core/valence) distinction does have to match as a pre-condition for comparing the basis sets

@ndaelman-hu ndaelman-hu linked a pull request Jul 21, 2023 that will close this issue
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working. It also represents a quick fix in response to a bug. feature / enhancement New feature or request
Projects
None yet
Development

Successfully merging a pull request may close this issue.

1 participant