← Latest papers
🔬 materials science

Compressed Sensing of Compton Profiles for Fermi Surface Reconstruction: Concept and Implementation

This paper proposes a compressed sensing approach that leverages the hidden sparsity of momentum distributions to accurately reconstruct the Fermi surface of bcc-Li from Compton scattering data using as few as 14 scattering directions, thereby overcoming the long acquisition times that typically limit practical applications of the technique.

Original authors: J. Otsuki, K. Yoshimi, Y. Nakanishi-Ohno, M. Sekania, L. Chioncel, M. Mizumaki

First appeared 2022-11-09 · Updated 2026-11-18
📖 1 min read☕ Coffee break read

Original authors: J. Otsuki, K. Yoshimi, Y. Nakanishi-Ohno, M. Sekania, L. Chioncel, M. Mizumaki

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Technical Summary: Compressed Sensing of Compton Profiles for Fermi Surface Reconstruction

Problem Statement
Compton scattering is a established technique for probing the electronic states in solids by measuring the electron momentum density (EMD), ρ(p)\rho(\mathbf{p}), projected along specific scattering directions. While the inverse problem of reconstructing the full three-dimensional EMD from these projections is theoretically analogous to computed tomography (CT), practical application is severely limited. Traditional reconstruction methods require a dense distribution of scattering angles to accurately recover the EMD. However, experimental constraints typically limit measurements to a small number of directions (approximately 10). This scarcity of data renders the inverse problem underdetermined, making it difficult to identify Fermi surface features, which manifest as discontinuities in ρ(p)\rho(\mathbf{p}) but appear only as subtle cusps in the measured Compton profiles, Jζ(pz)J_\zeta(p_z).

Methodology
The authors propose a reconstruction framework based on compressed sensing to overcome the limitation of sparse scattering data. The methodology proceeds as follows:

  1. Mathematical Formulation: The relationship between the Compton profiles and the EMD is treated as a linear system y=Axy = Ax, where yy represents the measured profiles and xx represents the discretized EMD. Since the number of measurements is fewer than the unknown variables, the system is underdetermined.
  2. Sparsity Assumption: The method exploits a "hidden sparsity" in the momentum distribution. In regions away from the Fermi surface, energy bands are either fully occupied or empty, implying that the gradient of the EMD, ∇ρ(p)\nabla \rho(\mathbf{p}), is zero. Thus, the solution is sparse in its derivative domain.
  3. Optimization (Generalized LASSO): Instead of interpolating missing data (as in traditional Fourier methods), the authors solve for ρ(p)\rho(\mathbf{p}) by minimizing a generalized least absolute shrinkage and selection operator (LASSO) objective function:
    F({ρj})=12∑i∈measured[Bi−∑jAijρj]2+λ∑j∑ξ∣∑j′(Dξ)jj′ρj′∣F(\{\rho_j\}) = \frac{1}{2} \sum_{i \in \text{measured}} \left[ B_i - \sum_j A_{ij}\rho_j \right]^2 + \lambda \sum_j \sum_{\xi} \left| \sum_{j'} (D_\xi)_{jj'} \rho_{j'} \right|
    Here, the first term ensures fidelity to the measured data (via the Fourier transform of the profiles), and the second term penalizes non-zero gradients (∇ρ\nabla \rho), enforcing sparsity. The matrix AA is constructed using the direct Fourier transform basis, ensuring it is dense, which is a prerequisite for successful compressed sensing.
  4. Constraints: The optimization is subject to physical constraints: non-negativity of the EMD (ρ≥0\rho \geq 0) and the sum rule (total electron count).
  5. Parameter Tuning: The regularization parameter λ\lambda is determined objectively using a 5-fold cross-validation (CV) method to minimize the validation error, balancing the fit to the data against the sparsity constraint.
  6. Symmetry Reduction: To manage computational cost, crystal symmetry operations (specifically OhO_h point group for alkali metals) are applied to reduce the number of independent variables in the optimization.

Key Contributions and Results
The method was validated using first-principles calculated Compton profiles for body-centered cubic (bcc) Lithium (Li).

  • Data Efficiency: The authors successfully reconstructed the three-dimensional EMD and identified the Fermi surface using data from only 14 scattering directions, a significant reduction compared to the dense angular sampling required by conventional tomography.
  • Accuracy: The reconstructed EMD exhibited a clear discontinuity at the Fermi momentum (pF≈0.58p_F \approx 0.58 a.u.) without the artificial oscillations often seen in direct Fourier transform methods.
  • Noise Robustness: The study investigated the impact of Gaussian noise (simulating experimental errors) with standard deviations σ=10−3,10−2,\sigma = 10^{-3}, 10^{-2}, and 10−110^{-1}. While higher noise levels blurred the discontinuity in ρ(p)\rho(\mathbf{p}), the Fermi surface could still be reliably identified by tracking the ridge in the gradient magnitude ∣∇ρ(p)∣|\nabla \rho(\mathbf{p})|.
  • Visualization: The reconstructed results correctly reproduced the nearly isotropic occupied states of the free-electron gas model and the anisotropy of the bcc lattice, with the Fermi surface clearly visible in the intensity maps of the gradient.

Significance
The paper claims that this compressed-sensing approach addresses the primary bottleneck in Compton scattering experiments: the long acquisition time required to measure sufficient scattering directions. By leveraging the sparsity of the momentum distribution's gradient, the method enables the reconstruction of the Fermi surface with unprecedented accuracy using a minimal set of scattering axes. The authors suggest this technique will permit wider applications of Compton scattering to investigate complex materials, including those with three-dimensional Fermi surfaces, thereby accelerating research in fermiology. The work serves as a proof-of-concept demonstration on bcc-Li, laying the groundwork for future applications to more complex systems.

Drowning in papers in your field?

Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.

Try Digest →