Pith. sign in

REVIEW 4 major objections 6 minor 2 references

Prediction magnetocrystalline anisotripy Fe-Rh thin films via machine leaning

T0 review · 4 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A LASSO linear model using orbital-moment anisotropies and averages predicts the magnetocrystalline anisotropy of Fe-Rh thin films on MgO(001).

desk verdict The paper's own results section refutes its headline claim of a linear Bruno-type relation for Fe-Rh/MgO; only the systematic DFT dataset has value. read the letter →

arxiv 1908.04762 v1 pith:NXNFBKLQ submitted 2019-08-06 cond-mat.mtrl-sci

Eun Sung Jekal , Hyunwoo Park This is my paper · ORCID
classification cond-mat.mtrl-sci
keywords magnetocrystallineanisotropyLASSOorbitalmomentFe-RhthinfilmsdensityfunctionaltheorymachinelearningMgOsubstrateBrunorelation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper applies LASSO regression to first-principles density-functional data for 128 atomic-layer configurations of seven-layer Fe-Rh films on MgO(001). It claims that the magnetocrystalline anisotropy energy (EMCA) is linearly correlated with the anisotropies and averages of the Fe and Rh orbital moments. The fitted model is tested with leave-one-out cross-validation and reported to agree with the DFT-computed EMCA values. The paper also includes a passage noting that the same descriptors may not be sufficient for systems with strong spin-orbit coupling, and that the discrepancy between calculated and predicted EMCAs suggests a breakdown of the Bruno relation.

What carries the argument

The machinery is the linear regression model of Eq. (1), in which EMCA is expanded in the orbital-moment anisotropies and averages of Fe and Rh at each atomic layer, together with LASSO's L1 regularization for coefficient selection and leave-one-out cross-validation. The supporting DFT calculations use the FLAPW method with the force theorem to obtain EMCA and orbital moments for the 128 layer configurations. The 'Bruno relation' — the proportionality between EMCA and the anisotropy of the orbital moment — is the physical hypothesis the model tests.

What would settle it

Apply the identical LASSO descriptor set (orbital-moment anisotropies and averages) to a different strong-SOC binary thin film, such as Co-Pt or Fe-Pd on MgO, and compare the predicted EMCA against first-principles calculations; if the linear relation fails there, the descriptor set is not sufficient.

Watch

Extended reading notes

Core claim

The paper's central claim is that LASSO regression on orbital-moment descriptors yields a linear relationship between EMCA and the anisotropy of orbital moments for Fe-Rh thin films, so that EMCA can be predicted from orbital moments alone. The authors assert this confirms the applicability of the Bruno relation to these binary films, where the strong spin-orbit coupling of Rh enhances the perpendicular anisotropy. The paper also contains an explicit caveat: the discrepancy between the first-principles and predicted EMCAs indicates that orbital-moment anisotropies and averages alone may not be sufficient descriptors, and that spin-flip terms can break the Bruno relation in strong-SOC systems.

Load-bearing premise

The model assumes that magnetocrystalline anisotropy energy is fully determined by a linear combination of the orbital-moment anisotropies and averages of Fe and Rh at each atomic layer, with no other physical ingredients.

Editorial extensions

If this is right

  • If the linear relation is correct, magnetocrystalline anisotropy energies for Fe-Rh/MgO films could be estimated directly from orbital-moment calculations, avoiding the more expensive force-theorem computations.
  • The fitted LASSO coefficients identify which atomic layers and which species contribute most to MCA, offering a possible design rule for tuning perpendicular anisotropy in Fe-Rh films.
  • The result supports the Bruno relation as a valid approximation for Fe-Rh binary films, extending its applicability to systems where 5d elements provide strong spin-orbit coupling.
  • The same regression procedure could be reused to screen other binary magnetic film configurations for high perpendicular MCA before performing full DFT calculations.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A natural next step is to test the same descriptor set on other strong-SOC film systems, such as Co-Pt or Fe-Pd; if the linear relation fails there, the boundary of the Bruno relation's validity would be mapped.
  • The paper's own admission that the descriptors may be insufficient suggests that augmenting the model with spin-flip terms or interface-specific orbital-moment components could recover predictive power.
  • Because the dataset exhaustively covers all 128 layer orderings, it could be reused to benchmark richer descriptor sets and settle whether the observed linear trend is an artifact of collinearity in the orbital-moment features.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The manuscript applies LASSO regression to DFT-computed magnetocrystalline anisotropy (MCA) energies and orbital moments of Fe-Rh thin films on MgO(001), claiming that a linear relationship exists between the MCA energy and the anisotropy of orbital moments. The authors model 128 seven-layer binary Fe-Rh configurations, fit four coefficient sets (A, B, C, D) in Eq. (1) using leave-one-out cross-validation, and compare predicted with calculated EMCA values. The abstract and conclusions assert success, while the Results section states that the predicted values are 'not satisfying' and that the descriptors 'might not be sufficient', explicitly acknowledging a breakdown of the Bruno relation.

Significance. If the claimed linear relation were established with validated predictive accuracy, the work could provide a computationally cheap route to estimate MCA in Fe-Rh/MgO films from orbital moments alone. However, the manuscript reports no quantitative goodness-of-fit metrics, no fitted coefficients, no data, and no code, and its own Results section undermines the central claim. As presented, the paper does not constitute a verifiable scientific contribution beyond the underlying DFT data.

major comments (4)
  1. [Section 3 (Results)] The central claim in the Abstract and Conclusions that 'we have successfully found a linear behavior' is directly contradicted by the same Section's statements: the predicted EMCAs 'are not satisfying', the orbital-moment descriptors 'might not be sufficient', and the analysis results in 'a breakdown of the Bruno relation'. The paper thus self-refutes its main assertion, and no evidence is offered to reconcile these contradictions.
  2. [Section 3, Fig. 3c] No quantitative error metric (RMSE, MAE, R²) is reported for the LASSO predictions, and the same paragraph describes the comparison as both 'agree well' and 'not satisfying'. Without error bars or a comparison against a null model, leave-one-out cross-validation on 64 samples with roughly 28 descriptors cannot establish predictive power.
  3. [Eq. (1) and following text] The descriptor set consisting only of anisotropies and averages of orbital moments is explicitly admitted to be insufficient for systems with strong spin-orbit coupling. Because the linear relation is the paper's central assertion, this admission invalidates the load-bearing assumption of Eq. (1). A concrete test would be to report the fitted coefficients and evaluate predictions on a held-out set of configurations that were not used in fitting; neither is provided.
  4. [Reproducibility] The manuscript does not disclose the LASSO coefficients, the regularization parameter, the data, or the code. The claim that a large coefficient indicates physical importance cannot be checked, and the regression is not reproducible from the information given.
minor comments (6)
  1. [Title and Abstract] The title contains typos ('anisotripy', 'leaning' for 'anisotropy' and 'learning'), and the abstract has 'expectially' instead of 'especially'.
  2. [Figures] Figure numbering is duplicated and inconsistent: both Section 2 and Section 3 contain 'Figure 1' and the captions are reused, which makes the narrative difficult to follow.
  3. [Eq. (1)] Equation (1) is referenced but never displayed in the text, so the exact form of the regression model is unclear.
  4. [Notation] The text refers to 'AAu,1' in the coefficient explanation, but the system contains only Fe and Rh atoms, not Au.
  5. [Section 2] The number of configurations is written as '27 = 128', which should be '2^7 = 128'.
  6. [References] The reference list contains formatting errors, including missing closing brackets in [13] and [14], and a misspelled author name ('Tibshirani, Rober' should be 'Robert').

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the LASSO fit is a standard (if weakly validated) regression, and the paper's internal contradictions affect correctness, not circularity.

full rationale

The claimed derivation chain is: first-principles DFT computations produce EMCA and orbital moments for 128 Fe-Rh/MgO layer configurations; a LASSO linear model with orbital-moment anisotropies and averages as descriptors is fitted to a training subset; leave-one-out cross-validation is used to evaluate the fitted coefficients on held-out configurations. This is supervised regression, not a definitional reduction: the held-out EMCA values are generated by coefficients fitted on other samples, so the 'prediction' is not identical to the fitted input by construction. The 'linear behavior' reported in the abstract is the fitted regression itself, which is the normal content of such a study rather than a circular step. The paper's Section 3 explicitly states that the LASSO-predicted EMCAs are 'not satisfying', that the chosen descriptors 'might not be sufficient', and that strong SOC leads to 'a breakdown of the Bruno relation'; this is an internal refutation of the abstract's success claim and a serious correctness/reproducibility problem (no coefficients, data, code, or error metrics are reported), but it is not an equivalence-by-construction. The absence of an external benchmark is a weakness in predictive validity, not circularity. The self-citations in the reference list are not used for the load-bearing LASSO-Bruno argument, and no uniqueness theorem is imported from prior work by the authors. Therefore no circular step meeting the required evidentiary standard is found.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central claim rests on DFT data, the force theorem, and the Bruno-relation ansatz. The only fitted parameters are the LASSO coefficients, which are not disclosed, and the regularization strength.

free parameters (2)
  • LASSO regression coefficients A_i, B_i, C_i, D_i = not reported
    Fitted to DFT data via LASSO; values are not listed, making reproduction impossible.
  • LASSO regularization parameter (lambda) = not reported
    The paper does not specify how the penalty strength was chosen, a free parameter.
assumptions (4)
  • domain assumption DFT within GGA accurately describes the electronic structure and MCA of Fe-Rh/MgO films
    All training data is computed from DFT; no experimental validation is provided.
  • domain assumption The force theorem applies for MCA energy calculation
    Used in Section 2 to compute E_MCA from eigenvalue differences.
  • domain assumption The Bruno relation between MCA and orbital moment anisotropy holds for these systems
    The paper assumes this linear relationship in Eq. 1, but later argues strong SOC may break it.
  • domain assumption The 128 atomic-layer configurations are a representative sample of Fe-Rh/MgO films
    All possible binary configurations of 7 layers are used, but only one substrate thickness and one vacuum size.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Prediction magnetocrystalline anisotripy Fe-Rh thin films via machine leaning." pith.science (2026). https://pith.science/paper/NXNFBKLQ

@misc{pith2026190804762,
  author       = {Pith},
  title        = {Pith review of: Prediction magnetocrystalline anisotripy Fe-Rh thin films via machine leaning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NXNFBKLQ}},
  note         = {Machine review of arXiv:1908.04762}
}
read the original abstract

Least absolute shrinkage and selection operator (Lasso) was originally formulated for least squares models and this simple case reveals a substantial amount about the behavior of the estimator. It also shows that the coefficient estimates need not be unique if covariates are collinear. Using this Lasso technique, we analyze a magnetocrystalline anisotropy energy which is a long-standing issue in transition-metal thin films, expectially for Fe-Rh thin film systems on a MgO substrate. Our LASSO regression took advantage of the data obtained from first principles calculations for single slabs with seven atomic-layers of binary Fe-Rh films on MgO(001). In the case of Fe-Rh thin films, we have successfully found a linear behavior between the MCA energy and the anisotropy of orbital moments.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

2 extracted references · 2 canonical work pages

  1. [4]

    Finally we show that the EMCA of the Fe-Rh thin films is strongly correlated to the anisotropy of orbital moments

    Conclusions We found calculation results which are combined first-principles calculations (density functional theory) and LASSO which is machine learning analysis. Finally we show that the EMCA of the Fe-Rh thin films is strongly correlated to the anisotropy of orbital moments. Furthermore, the estimated EMCA shows a reasonable agreement with the calculat...

    work page 2002

  2. [5]

    Pushing the limits of magnetic anisotropy in the Sm-Co system

    Li, Yingfeng, and B. Kish, Laszlo, Fluctuation and Noise Letters, 6, L127 (2006).[6] Pey, Kin Leong, et al, 2019 Electron Devices Technology and Manufacturing Conference (EDTM). IEEE, (2019).[7] Cubukcu, Murat, et al., IEEE Transactions on Magnetics, 54, 1 (2018).[8] Thomas, L., et al., 2017 IEEE International Magnetics Conference (INTERMAG). IEEE, (2017)...

    work page Pith review arXiv 2006

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.