Sparsity for isotropic spherical random fields
Giacomo Greco, Domenico Marinucci · 2026
Read Sparsity for isotropic spherical random fields by Giacomo Greco, Domenico Marinucci (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-2601-21535.txt.
Introduction
Giacomo Greco and Domenico Marinucci study random fields spread over a sphere and how to make them sparse. Such fields model data on globes, from cosmic background radiation to climate maps. A simple new representation lets the authors compare several meanings of sparsity under rotational symmetry.
They also build sparse fields that mimic the spectra of popular non-Gaussian fields. The mimicry holds for both the basic power pattern and higher-order patterns. At the same time, the sparse form allows fast simulation. Fast, faithful simulation is gold for cosmology, where thousands of fake skies are needed to judge real ones.
Catalog record
| CREATOR | Giacomo Greco, Domenico Marinucci |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 65 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-21535
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Sparsity for isotropic spherical random fields Thanks: Both authors are associated to INdAM (Istituto Nazionale di Alta Matematica “Francesco Severi”) and the group GNAMPA; they are supported by the PRIN project GRAFIA (CUP: E53D23005530006) and MUR Departement of Excellence Programme 2023-2027 MatMod@Tov (CUP: E83C23000330006). Giacomo Greco, Domenico Marinucci Abstract . We introduce a simple representation for isotropic spherical random fields and we discuss how it allows to discuss different notions of sparsity under isotropy. We also show how a suitable construction of sparse fields can mimic well the angular power spectrum and the polyspectra of some popular non-Gaussian fields, at the same time allowing for computationally efficient simulation algorithms. Using related ideas we also show how it is possible to obtain sparse approximations of spherical random fields which preserve isotropy, thus addressing an issue which has been raised in the Cosmological literature. Keywords and Phrases: Sparsity, Isotropy, Gaussianity, Spherical Random Fields MSc Classification: 60G60, 33C55, 43A90 1. Background and objectives Over the last two decades, isotropic spherical random fields have attracted a lot of interest and have been considered for an enormous number of applications in very diverse areas, including Cosmology, Climate Science, Geophysics, Quantum Mechanics, Machine Learning and Neural Networks, among others. The overwhelming majority of this literature has focused on Gaussian circumstances; this is also due to the mathematical difficulty to introduce principled non-Gaussian models which do not violate isotropy. In particular, the main tool for the investigation of finite variance, isotropic spherical random fields is the spectral representation theorem, which allows expressing T(x), x\in\mathbb{S}^{2} as (1.1) T(x)=\sum_{\ell=0}^{\infty}\sum_{m=-\ell}^{\ell}a_{\ell m}Y_{\ell m}(x)\text{ ,} where the equality holds both in the \mathrm{L}^{2}(\Omega) and in the \mathrm{L}^{2}(\Omega\times\mathbb{S}^{2}) sense; the spherical harmonics Y_{\ell m}(.) are an orthonormal system of eigenfunctions of the spherical Laplacian operator, and the coefficients \left\{a_{\ell m}\right\} are zero mean and uncorrelated, with variance \mathbb{E}[|a_{\ell m}|^{2}]=C_{\ell} , the latter defining the angular power spectrum of the random field. The randomness of the whole field is hence entirely encoded in the vector of harmonic coefficients (\mathbf{a}_{\ell})_{\ell\in\mathbb{N}}=(a_{\ell m})_{\ell\in\mathbb{N},m=-\ell,\dots,+\ell} , (see [ Yad83 ] , [ Leo99 ] , [ MP11 ] ) . A question that can appear natural is whether it is possible to achieve an alternative representation that, while preserving isotropy, may turn out to be "more sparse", i.e. to require a smaller number of random coefficients under some circumstances. The question arises in particular if one keeps in mind that many of the most popular modern statistical or machine learning procedures (such as the Lasso, see [ BvdG11 ] ) are based on various forms of penalization/sparsity-enforcing algorithms. Naive applications trying to enforce sparsity on the coefficients a_{\ell,m} in spectral representation eq. 1.1 have been considered in the literature, but they necessarily lead to anisotropic features, as discussed in [ CM15 ] , [ FMM+14 ] , [ LGSWW20 ] and others. Indeed, it can be shown that a spherical random field is isotropic if and only if the triangular array of coefficients \left\{a_{\ell m}\right\} satisfies the following invariance conditions; we must have that \mathbf{a}_{\ell}\overset{Law}{=}D_{\ell}(g)\mathbf{a}_{\ell}\text{ ,} where we write \mathbf{a}_{\ell}=(a_{\ell,-\ell},\,\dots,\,a_{\ell\ell})^{T} for the (2\ell+1)\times 1 column vector of random spherical harmonic coefficients. Here we have introduced the set of “Wigner” (2\ell+1)\times(2\ell+1) matrix-valued functions D_{\ell}(g), g\in SO(3) , which provide a family of irreducible representations for the special group of rotations SO(3) (see [ VMK88 ] , Chapter 4 and [ MP11 ] , Chapter 3.3). In particular, this implies that the spectral representation cannot be sparse for isotropic fields, i.e., it cannot be imposed that some of these coefficients are zero-valued, because this would clearly violate the invariance condition: the matrix functions D_{\ell}(.) yield an irreducible representations of the group SO(3) and hence no non-trivial subspace of \mathbb{R}^{2\ell+1} can be invariant to their action; hence if some components of the vector of harmonic coefficients \mathbf{a}_{\ell} are null, they must necessarily be non-zero for some other rotation g\in SO(3) . In order to preserve isotropy and enforce sparsity the only existing approach that we are aware of is the one proposed in [ LGSWW20 ] , where sparsity is enforced by introducing a regularization term on the sample angular power spectrum at different multipoles. As discussed in more detail in section 3.1 , the approach is rigorous and computationally feasible; however, from a certain point of view it leads more to smoothness in a Sobolev sense than to sparsity in the sense we advocate in this paper, that is, a drastically lower number of random coefficients. Moreover, the approach preserves isotropy and Gaussianity, which is definitely a useful feature for many applications; however, it cannot enforce sparsity for given values of the angular power spectrum sequence, which can be a limitation for some applications. The purpose of this paper is to introduce a novel notion of sparsity for isotropic random fields, alongside with a general construction method that generates isotropic (although generally non-Gaussian) spherical random fields. We shall exploit this construction in different directions: • We shall show how our construction is flexible with respect to higher-order moments/cumulants of the spherical harmonic coefficients (the so-called polyspectra). In particular, this construction allows us to generate isotropic, non-Gaussian random fields with explicit prescribed power spectra and polyspectra, generating set of random coefficients whose cardinality is (much) smaller than constructions with similar angular resolutions, based instead on the spectral representation eq. 1.1 . • We will show, in particular, that these models can easily mimic the most important features for Cosmological applications, i.e., those of “local" non-Gaussianity, while also covering a plethora of cases which are not covered by local models. We refer to the classical survey [ BKMR04 ] for a full overview on other forms of nonGaussianity which can be induced on Cosmic Microwave Background radiation by the so-called inflationary models, the dominant paradigm for the dynamics of the primordial Universe. • We shall show how, given the realization of a spherical random field, it is possible to approximate the latter to an arbitrary degree of accuracy in terms of this new representation. More precisely, we will introduce a sparse reconstruction algorithm which is invariant to the choice of coordinates and is computationally feasible, as illustrated in our Monte Carlo simulations. The main idea behind our strategy is to consider a general decomposition of isotropic random fields built as weighted superpositions of deterministic waves centered in random directions on the sphere, i.e., (1.2) T(x)=\sum_{\ell\geq 0}\sum_{k=1}^{K_{\ell}}\eta_{\ell k}\frac{2\ell+1}{4\pi}P_{\ell}(\left\langle\xi_{k},x\right\rangle)\text{ , where }\xi_{k}\sim\mathrm{Unif}(\mathbb{S}^{2})\,, where \left\{\eta_{\ell k}\right\} is a finite-variance random sequence, whose properties will be discussed below. Remark 1.1 . It is immediate to see that the random fields defined in ( 1.2 ) are strongly isotropic; indeed for any g\in SO(3) we have that \displaystyle T^{g}(x):= \displaystyle T(gx)=\sum_{\ell\geq 0}\sum_{k=1}^{K_{\ell}}\eta_{\ell k}\frac{2\ell+1}{4\pi}P_{\ell}(\left\langle\xi_{k},gx\right\rangle) \displaystyle= \displaystyle\sum_{\ell\geq 0}\sum_{k=1}^{K_{\ell}}\eta_{\ell k}\frac{2\ell+1}{4\pi}P_{\ell}(\left\langle g^{-1}\xi_{k},x\right\rangle)\overset{Law}{=}T(x)\,. Remark 1.2 . Neglecting the weighting factors [\eta_{\cdot,\cdot}] and considering fields supported on a single frequency/multipole \ell , we obtain as special cases the Spherical Poisson Waves recently introduced by [ BDMT24 ] , see also [ CD25 ] . We shall show (see Theorem 3.1 ) that all isotropic random fields admit the representation ( 1.2 ); moreover, with suitable assumptions on the array of random variables \{\eta_{\ell k}\} we will be able to construct fields which mimic the