Energetic vs Inference-Based Invisibility: Fisher-Information Analysis of Two-Layer Acoustic Near-Cloaks
J. Sumaya-Martinez, J. Mulia-Rodriguez · 2026
Read Energetic vs Inference-Based Invisibility: Fisher-Information Analysis of Two-Layer Acoustic Near-Cloaks by J. Sumaya-Martinez, J. Mulia-Rodriguez (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-06091.txt.
Introduction
Making an object quiet to sound is not the same as making it unknowable. J. Sumaya-Martinez and 1 co-author develop that distinction in a 2026 theory paper on acoustic near-cloaks, coatings that nearly hide an object from sound waves in a narrow frequency band.
Their two-dimensional model wraps a cylinder in two fluid-like layers designed to cancel the strongest scattering modes. Total scattered energy dropped sharply at the design frequency. Yet Fisher information, a measure of how much noisy data reveal about the object's size and materials, fell only slightly. Estimation bounds stayed informative. The authors give an analytic argument for this split and map trade-offs between suppressing scattering and hiding parameters.
Catalog record
| CREATOR | J. Sumaya-Martinez, J. Mulia-Rodriguez |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 18 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-06091
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Energetic vs Inference-Based Invisibility: Fisher-Information Analysis of Two-Layer Acoustic Near-Cloaks J. Sumaya-Martinez, J. Mulia-Rodriguez Abstract Near-cloaks based on passive coatings can strongly suppress the scattered-field energy in a narrow frequency band, yet an observer’s ability to infer object parameters from noisy measurements need not decrease proportionally. We develop a fully theoretical, two-dimensional (2D) framework for a coated acoustic cylinder in an air background. Using an exact cylindrical-harmonic solution of the Helmholtz equation, we compute the modal scattering coefficients a m ( ω ) a_{m}(\omega) for a core of radius a a surrounded by two concentric, effective-fluid layers. We design the coating to cancel the dominant low-order multipoles (monopole m = 0 m=0 and dipole m = ± 1 m=\pm 1 ) at a target frequency, yielding a narrowband near-cloak. Beyond the conventional energetic metric (total scattering width), we quantify information-based detectability through the Fisher information matrix (FIM) and the associated Cramér–Rao lower bounds (CRLBs) for joint estimation of size and material parameters, 𝒙 = [ a , ρ 1 , c 1 ] T \bm{x}=[a,\rho_{1},c_{1}]^{T} , from noisy far-field data. A representative air-background case study exhibits a peak ∼ 25 \sim 25 dB reduction in total scattering width near the design frequency, while tr ( FIM ) \mathrm{tr}(\mathrm{FIM}) decreases by only a few dB, illustrating that energy-based and inference-based invisibility are distinct objectives. We discuss bandwidth limitations consistent with passivity/causality and delay–bandwidth constraints, and show how the FIM/CRLB viewpoint provides a task-aware and parameter-resolved characterization of near-cloaking. I Introduction Acoustic cloaking aims to mitigate an object’s scattering signature, ideally rendering it indistinguishable from the surrounding medium. Two broad paradigms dominate the literature: transformation acoustics, which prescribes spatially varying anisotropic parameters to steer waves around a hidden region 1 , and scattering cancellation (multipole cancellation), which suppresses dominant low-order scattering harmonics using coatings or metasurfaces 2 . Transformation-based cloaks are elegant but often require extreme or anisotropic parameters that are difficult to realize—particularly in air—whereas scattering-cancellation designs are analytically tractable for canonical geometries and naturally lead to bandwidth-limited near -cloaks. Fundamental constraints further motivate near-cloaks: passivity and causality impose stringent trade-offs between achievable scattering reduction and bandwidth 5 , and delay–bandwidth considerations restrict cloaking of electrically/acoustically large objects 4 . Consequently, a physically meaningful target is frequently near-cloaking—strong suppression in a narrow band—rather than ideal broadband invisibility. Most cloaking studies quantify performance using energetic metrics, such as total scattering cross section/width. In many detection and classification settings, however, an observer performs inference : estimating object parameters from noisy measurements. This motivates an information-theoretic viewpoint. We quantify parameter detectability using the Fisher information matrix (FIM) and the associated Cramér–Rao lower bounds (CRLBs), which provide fundamental limits on any unbiased estimator 3 ; 6 . Contributions. (i) We present a compact exact modal formulation for a two-layer coated 2D cylinder in an air background. (ii) We design the coating to suppress the dominant monopole and dipole scattering coefficients at a target frequency, yielding a near-cloak. (iii) We analyze joint size–material inference for \bm{x}=[a,\rho_{1},c_{1}]^{T} and demonstrate that energy-based invisibility and FIM/CRLB-based detectability can differ substantially. II Model and exact scattering formulation II.1 Geometry and governing equations We consider time-harmonic acoustics p(r,\theta)e^{-\mathrm{i}\omega t} in an unbounded air background with density \rho_{0} and sound speed c_{0} , so that k_{0}=\omega/c_{0} . A circular cylinder (core) of radius a is surrounded by two concentric layers of thicknesses t_{1} and t_{2} , yielding radii b=a+t_{1} and c=a+t_{1}+t_{2} (Fig. 1 ). Each region j\in\{1,2,3\} is modeled as an isotropic effective fluid with density \rho_{j} and sound speed c_{j} ( k_{j}=\omega/c_{j} ). Within each region the pressure satisfies \nabla^{2}p+k_{j}^{2}p=0. (1) We will use dimensionless groups \kappa\equiv k_{0}a,\qquad\beta\equiv\frac{b}{a},\qquad\gamma\equiv\frac{c}{a},\qquad Z_{j}\equiv\rho_{j}c_{j},\qquad K_{j}\equiv\rho_{j}c_{j}^{2}, (2) and note that low-order multipoles dominate as k_{0}c=\gamma\kappa\to 0 . II.2 Cylindrical-harmonic expansion For an incident plane wave p_{\mathrm{inc}}(r,\theta)=\exp(\mathrm{i}k_{0}r\cos\theta) , p_{\mathrm{inc}}=\sum_{m=-\infty}^{\infty}\mathrm{i}^{m}J_{m}(k_{0}r)\,e^{\mathrm{i}m\theta}. (3) The scattered field is p_{\mathrm{scat}}=\sum_{m=-\infty}^{\infty}a_{m}(\omega)\,H_{m}^{(1)}(k_{0}r)\,e^{\mathrm{i}m\theta}, (4) with complex coefficients a_{m}(\omega) . In the core we enforce regularity at r=0 , and in coating layers allow linear combinations of J_{m} and Y_{m} : \displaystyle p_{1}^{(m)}(r) \displaystyle=A_{1m}J_{m}(k_{1}r), (5) \displaystyle p_{2}^{(m)}(r) \displaystyle=A_{2m}J_{m}(k_{2}r)+B_{2m}Y_{m}(k_{2}r), (6) \displaystyle p_{3}^{(m)}(r) \displaystyle=A_{3m}J_{m}(k_{3}r)+B_{3m}Y_{m}(k_{3}r), (7) \displaystyle p_{0}^{(m)}(r) \displaystyle=\mathrm{i}^{m}J_{m}(k_{0}r)+a_{m}H_{m}^{(1)}(k_{0}r). (8) II.3 Boundary conditions and per-mode linear system At r=a,b,c we impose continuity of pressure and radial particle velocity u_{r}=-(1/(\mathrm{i}\omega\rho))\partial_{r}p , equivalently continuity of (1/\rho)\partial_{r}p . For each angular order m this yields a 6\times 6 linear system \mathbf{M}_{m}(\omega)\bm{u}_{m}=\bm{b}_{m}(\omega) for \bm{u}_{m}=[A_{1m},A_{2m},B_{2m},A_{3m},B_{3m},a_{m}]^{T} . An explicit construction is provided in App. A . II.4 Energetic metric: total scattering width We use the 2D total scattering width \sigma_{\mathrm{scat}}(\omega)=\frac{4}{k_{0}}\sum_{m=-\infty}^{\infty}\left|a_{m}(\omega)\right|^{2}, (9) computed with symmetric truncation \left|m\right|\leq M_{\max} chosen for convergence. III Near-cloak design and Fisher-information detectability III.1 Multipole cancellation objective and design procedure In the low-frequency regime ( k_{0}c\ll 1 ), scattering is dominated by the monopole ( m=0 ) and dipole ( m=\pm 1 ) terms. We therefore define the design objective \mathcal{J}(\omega_{0})\equiv\left|a_{0}(\omega_{0})\right|^{2}+2\left|a_{1}(\omega_{0})\right|^{2}, (10) and seek effective-layer parameters that minimize \mathcal{J} at a target frequency \omega_{0} . In the numerical illustration, we used a constrained random search over (\rho_{2},c_{2},\rho_{3},c_{3}) with positivity constraints \rho_{j}>0 , c_{j}>0 and with bounds selected to avoid trivial singular limits; the reported design is representative rather than claimed globally optimal. III.2 Fisher information and CRLB for joint size–material inference Assume K far-field samples at angles \theta_{k} with complex noisy measurements y_{k}=\mu(\theta_{k};\bm{x})+n_{k}, (11) where \bm{x}=[a,\rho_{1},c_{1}]^{T} and n_{k} is circular complex Gaussian noise with variance \sigma^{2} . We model the scattered far-field angular dependence as \mu(\theta;\bm{x})\propto\sum_{m=-M}^{M}a_{m}(\bm{x})e^{\mathrm{i}m\theta}. (12) For complex Gaussian noise, a convenient expression for the FIM is I_{ij}(\bm{x})=\frac{2}{\sigma^{2}}\,\Re\sum_{k=1}^{K}\left(\frac{\partial\mu(\theta_{k};\bm{x})}{\partial x_{i}}\right)^{\!*}\left(\frac{\partial\mu(\theta_{k};\bm{x})}{\partial x_{j}}\right), (13) which we evaluate using numerical central differences for \partial\mu/\partial x_{i} . The CRLB satisfies \mathrm{Cov}(\hat{\bm{x}})\succeq\mathbf{I}(\bm{x})^{-1} ; when \mathbf{I} is ill-conditioned we use the Moore–Penrose pseudoinverse. To report normalized results without ambiguity in units, we plot ratios such as \sigma_{\mathrm{bare}}/\sigma_{\mathrm{cloak}} and \mathrm{tr}(I_{\mathrm{bare}})/\mathrm{tr}(I_{\mathrm{cloak}}) in decibels. III.3 Why energetic and information-based invisibility can decouple (low-order analytic view) To make the distinction between energetic and inference-based invisibility explicit, consider the low-order truncation |m|\leq 1 of the far-field model in Eq. ( 12 ). Using symmetry for a plane-wave excitation of a circularly symmetric structure, the scattered pattern can be written as \mu(\theta;\bm{x})\approx A_{0}(\bm{x})+A_{1}(\bm{x})\cos\theta+B_{1}(\bm{x})\sin\theta, (14) where (A_{0},A_{1},B_{1}) are linear combinations of the complex coefficients a_{0},a_{\pm 1} . 1 1 1 For example, one convenient real trigonometric form is \mu(\theta)\propto a_{0}+a_{1}e^{\mathrm{i}\theta}+a_{-1}e^{-\mathrm{i}\theta} . An energetic near-cloak designed by multipole cancellation aims to make A_{0},A_{1},B_{1} small at \omega_{0} . However, the Fisher information depends on sensitivities through Eq. ( 13 ). Inserting Eq. ( 14 ) into Eq. ( 13 ) yields I_{ij}\propto\Re\sum_{k}\left(\partial_{x_{i}}A_{0}+\partial_{x_{i}}A_{1}\cos\theta_{k}+\partial_{x_{i}}B_{1}\sin\theta_{k}\right)^{\!*}\left(\partial_{x_{j}}A_{0}+\partial_{x_{j}}A_{1}\cos\theta_{k}+\partial_{x_{j}}B_{1}\sin\theta_{k}\right), (15) so that I_{ij} can remain large even when A_{0},A_{1},B_{1} themselves are suppressed, provided the derivatives \partial_{x}A_{0},\partial_{x}A_{1},\partial_{x}B_{1} do not vanish. This explains, at a model level, why a coating can yield large reductions in \sigma_{\mathrm{scat}} while reducing \mathrm{tr}(\mathrm{FIM}) only modestly: multipole cancellation constrains field amplitudes at \omega_{0} , whereas information suppression requires simultaneous reduction of parameter sensitivities . The latter is a stronger (and generally multi-objective) requirement. IV Results: air-background numerical illustration We consider an air background with \rho_{0}=$1.21\text{\,}\mathrm{k}\mathrm{g}\,\mathrm{m}^{-3}$ and c_{0}=$343\text{\,}\mathrm{m}\,\mathrm{s}^{-1}$ . The core is modeled as an effective fluid approximating a high-impedance object with \rho_{1}=$1200\text{\,}\mathrm{k}\mathrm{g}\,\mathrm{m}^{-3}$ and c_{1}=$2500\text{\,}\mathrm{m}\,\mathrm{s}^{-1}$ . Geometry: a=$3\text{\,}\mathrm{c}\mathrm{m}$ , t_{1}=$1\text{\,}\mathrm{c}\mathrm{m}$ , t_{2}=$1\text{\,}\mathrm{c}\mathrm{m}$ (so b=$4\text{\,}\mathrm{c}\mathrm{m}$ and c=$5\text{\,}\mathrm{c}\mathrm{m}$ ). A representative two-layer design uses effective parameters (layer 1) \rho_{2}=$8.60\text{\,}\mathrm{k}\mathrm{g}\,\mathrm{m}^{-3}$ , c_{2}=$387\text{\,}\mathrm{m}\,\mathrm{s}^{-1}$ and (layer