Skip to content

Fitbauer v5.2.3

Choose a tag to compare

@github-actions github-actions released this 15 Oct 11:02
· 17 commits to main since this release

Fitbauer v5.2.3

🇪🇸 Notas en español

A release about NORMOS compatibility, based on a review of the NORMOS source
code and the WinNormos manual. It adds the in-field Le Caër distribution and
the applied field of the Ising relaxation. It also fixes how .JOB files with
a negative VMAX are opened.


Important fix: .JOB files with a negative VMAX

In NORMOS the sign of VMAX reverses the spectrum. The importer took its
absolute value, so those jobs opened mirrored, with the isomer shift of the
wrong sign. The sign is now kept on import and export, with a warning. The
benchmark of real NORMOS jobs exposed the problem: every job with a negative
VMAX folded differently from NORMOS, and with the sign they agree to noise
level. If you imported such a job, import it again.

New: in-field Le Caër distribution

A new Le Caër shape for P(ΔEQ): the random Czjzek field gradient seen by
sextets in a fixed field, for disordered magnetic phases such as metallic
glasses and nanocrystalline materials.

  • The grid is the signed coupling e²qQ/2.
  • Each subspectrum is the exact powder average of a sextet whose EFG is
    randomly oriented with respect to the field.
  • The weights are the distribution of the signed Vzz of a Czjzek tensor
    (checked by Monte Carlo); σ is the only shape parameter.

It is NORMOS-DIST's DISTRI=4 with METHOD=7, and such .JOB files are now
imported with it. In P(ΔEQ) both Czjzek and Le Caër can always be chosen:
choosing Le Caër switches on the fixed-field sextet (33 T if the field was 0),
and choosing Czjzek switches it off. In the CLI: --shape lecaer.

New: applied field in the Blume–Tjon relaxation

The Blume–Tjon component has a new parameter, External field BEXT (T),
NORMOS's BEXT. The hyperfine field reverses with the spin and the applied one
does not, so the two states see BEXT ± BHF. In the fast limit the spectrum
collapses to a sextet of field BEXT instead of a single line.

It is a datum of the experiment: it is fixed by default, «Free all» does not
release it, and with 0 nothing changes. It matches Blume's matrix equation to
1e-10.

More from NORMOS

  • Ising relaxation (IRELAX) is imported as a Blume–Tjon component, with
    ν = OME·10⁶/2, P = BHF/BSAT and the global BEXT. Before, only a warning was
    given.
  • Gaussian distributions (DISTRI=2): NORMOS broadens STG with the grid
    step, so the real width is √(STG² + step²). The import warning now gives it.
    It used to say, wrongly, that STG was not carried over.
  • METHOD=5/7 keep the DISTRI shape. NORMOS's spline file holds the
    subspectra, not the distribution. Before, they were imported as «Fixed».
  • Shape of P: the panel and the report give the skewness and the excess
    kurtosis of the distribution, as NORMOS-DIST does.
  • Hesse–Rübartsch parameter: NORMOS-DIST's smoothing criterion,
    Σ(y − y_theo)²/Σy ≈ χ²/N, is shown in the panel and the report.

Validation

The benchmark of real NORMOS jobs has been redone. It covers 407 distinct
jobs, and only those that are a valid reference are kept:

  • jobs where NORMOS fitted well (reduced χ² ≤ 2): 161 jobs where it did not
    are discarded;
  • jobs where both programs see the same spectrum.

On the 216 remaining jobs, starting from NORMOS's solution, Fitbauer reaches
the same or a lower χ² in 211. The median agreement is 0.0014 mm/s in δ and
0.016 T in BHF.


Full suite: 759 tests passing.