Spectral Stacks: Theory & Applications

• Spectral stacks are composite structures that combine mathematical, algorithmic, or physical methods to aggregate and analyze spectral data.
• They underpin key applications in algebraic geometry, moduli theory, and observational astronomy by formalizing structures like even periodization and spectral stacking.
• Advanced methodologies such as inverse-variance weighted stacking enhance signal recovery and noise suppression in large-scale spectral and imaging datasets.

A spectral stack, in the broadest sense, refers to a composite structure—mathematical, algorithmic, or physical—for combining, analyzing, or encoding spectral data. Across disciplines as diverse as algebraic geometry, observational astronomy, and photonics, the term encompasses advanced methodologies for forming and manipulating aggregated spectral representations. This article surveys the rigorous technical foundations and applications of spectral stacks in both mathematical theory (particularly spectral algebraic geometry and moduli theory) and empirical sciences such as astronomy and imaging.

1. Spectral Stacks in Algebraic and Homotopical Geometry

In higher and derived algebraic geometry, a spectral stack is defined as a functor of the form

$$X \colon \mathrm{CAlg} \longrightarrow \mathrm{Ani}$$

where $\mathrm{CAlg}$ denotes the (not necessarily connective) $\mathbb{E}_\infty$-ring spectra and $\mathrm{Ani}$ the $\infty$-category of spaces (or anima). $X$ is required to satisfy fpqc descent: $$X(A) \xrightarrow{\sim} \operatorname{Tot}(X(B^{\otimes_A \bullet}))$$ for every faithfully flat extension $A \to B$, with the colimit taken in the category of presheaves. The class of such functors forms the $\infty$-category of spectral stacks $$\mathrm{SpStk} = \operatorname{Shv}_{\mathrm{fpqc}}(\mathrm{Aff})$$, where $\mathrm{Aff}$ indexes $\operatorname{Spec}(A)$ for $A \in \mathrm{CAlg}$, generalizing classical stacks from schemes (or Deligne–Mumford stacks) to derived or spectral settings (Gregoric, 2021, Gregoric, 23 Apr 2025).

Geometric spectral stacks are characterized by affineness of diagonals and the existence of a cover by a single affine. Their underlying "ordinary" stack is obtained by restricting the functor to discrete commutative rings and forms the bridge to classical algebraic geometry frameworks.

Spectral stacks provide the moduli formalism for derived geometry, e.g., the moduli of oriented formal groups $\mathcal{M}_{\mathrm{FG}}^{\mathrm{or}}$ (Gregoric, 2021), and parameterize families of spectral objects (such as generalized sheaves, Higgs bundles, or motivic spectra) along derived base loci.

2. Even Periodization and Structural Operations in Spectral Stacks

The even periodization functor $\operatorname{Per}^{\mathrm{even}}$ on a spectral stack $X$ is defined by Kan extension: $\operatorname{Per}^{\mathrm{even}}(X) = j_!j^* X$ where $$j \colon \operatorname{Aff}^{\mathrm{evp}} \hookrightarrow \operatorname{Aff}$$ includes the site of even-periodic $E_\infty$-rings ($\pi_{2n+1}(A)=0$ and $\pi_2(A)$ invertible over $\pi_0(A)$). On affines, this periodization yields: $$X^{\mathrm{evp}} \simeq \varinjlim_{\operatorname{Spec}(A) \to X, \, A \in \mathrm{CAlg}^{\mathrm{evp}}} \operatorname{Spec}(A)$$ with the result being a stack supported only on even-degree (periodic) data. The role of even periodization is the systematic extraction of 2-periodic structural information, for example, recovering the even filtration of Hahn–Raksit–Wilson from spectral stacks (Gregoric, 23 Apr 2025).

This functor is pivotal in the chromatic theory of spectra, underlies the geometrization of the Nygaard filtration (in $p$-adic prismatic cohomology), and connects with symmetric monoidal autoequivalences (shearing) in free-graded module categories, particularly over $\mathrm{MU}$ or $\mathrm{TMF}$.

3. Spectral Stack Moduli: Compactified Jacobians, Norm Maps, and Prym Stacks

For moduli problems associated to Higgs bundles or generalized sheaves, the notion of spectral stack acquires a more explicit algebro-geometric instantiation:

• Given a projective curve $X$ (possibly reducible/non-reduced), the compactified Jacobian stack $\overline{\mathcal{J}}(X,d)$ parameterizes torsion-free rank-1 sheaves of degree $d$ flat over any base $T$ (Carbone, 2020).
• The spectral correspondence assigns, for each $G$-Higgs pair $(E, \Phi)$ on a base curve, a point $a$ in the Hitchin base $A(G)$, defining a spectral curve $X_a$ equipped with a finite flat projection $\pi \colon X_a \to C$. The fiber of the $G$-Hitchin fibration over $a$ is then a (stacky) fiber of the norm map: $$\mathrm{Nm}_\pi : \overline{\mathcal{J}}(X_a,d') \longrightarrow \mathrm{Pic}(C)$$
• The Prym stack, defined as the kernel (equalizer stack) of $\mathrm{Nm}_\pi$, encodes the symmetry and invariant sub-data relevant for groups such as $SL(r)$, $Sp(2r)$, $PGL(r)$, and $GSp(2r)$.

Such structures are essential for describing the fibers of the Hitchin map in terms of spectral data and relating them to Lagrangian or integrable systems (Carbone, 2020).

4. Weighted Spectral Stacking in Observational Astronomy

Spectral stacking in radio astronomy and extragalactic surveys refers to methodologies for combining multiple spectral (or spectral-image) cubes to enhance detection sensitivity and suppress non-astronomical contaminants.

The weighted stacking algorithm, as implemented in "Spectral-Weighting" (Murgia et al., 2024), is designed to produce a "clean" spectral cube $I_{\mathrm{out}}(x, y, \nu)$ from $N$ input cubes $I_i(x, y, \nu)$ by leveraging per-voxel, inverse-variance weights: $$w_i(x, y, \nu) = \frac{1}{\operatorname{Var}\{ I_i(x', y', \nu) \mid (x', y') \in \mathcal{N}(x, y) \}}$$ where $\mathcal{N}(x, y)$ is a local neighborhood. The merged cube is computed as

$$I_{\mathrm{out}}(x, y, \nu) = \frac{\sum_{i=1}^N w_i(x, y, \nu) \cdot I_i(x, y, \nu)}{\sum_{i=1}^N w_i(x, y, \nu)}$$

Interfering signals (e.g., radio-frequency interference, RFI), which are intermittent across the cube ensemble, are statistically down-weighted, while persistent celestial features are preserved.

On both simulated data and real observations (e.g., multi-epoch M31 observations with the Sardinia Radio Telescope), weighted stacking demonstrated:

• SNR improvement by a factor $\simeq 1.4$ over naïve averaging.
• Residual artifact levels $< 0.6 \sigma_N$ in $95\%$ of output pixels, where $\sigma_N$ is the noise standard deviation.
• Enhanced sky recovery (additional $15\%$ area) and reduced residuals compared to classical flag-and-average pipelines, with noise suppression scaling as $\sqrt{N_{\mathrm{eff}}}$ (Murgia et al., 2024).

This approach is computationally efficient (parallelizable, memory-efficient for $2048\times2048\times4096$ cubes), robust to broad-band interference, and preserves the scientific integrity of faint source signals.

5. Methodological and Computational Aspects

Weighted spectral stacking necessitates several detailed algorithmic steps:

• Noise-floor estimation: identification of RFI-free background to calibrate local variance statistics.
• Sliding-window estimators: convolution of data and squared data with small box kernels to compute local variance.
• Handling large datasets: memory-mapped arrays, tiling along frequency blocks, and parallelization via OpenMP for scalability.

Comparison to alternative strategies:

• Naïve averaging: susceptible to outlier contamination and RFI artifacts.
• Flag-and-average: excises affected channels but loses spatial coverage unnecessarily.
• Inverse-variance (optimal) weighting: achieves highest SNR when input noise is uncorrelated and variance is known.

Figures of merit for algorithmic assessment include SNR improvement factors, bias in recovered astrophysical quantities (e.g., line flux, integrated intensity), and structure-preservation metrics (RMS difference between stack and model input).

6. Broader Contexts and Related Constructions

The concept of spectral stack, while rigorously defined in spectral algebraic geometry, also underlies:

• The architecture of advanced optical systems, such as multiplexed volume-phase holographic gratings for simultaneous acquisition of multi-order spectra (Alessio et al., 2017).
• The theoretical infrastructure of moduli problems, where the spectral stack of a group $G$ (parameterizing spectral covers and associated sheaves) is essential for describing abstract Hitchin systems (Carbone, 2020).
• The categorical and homotopical frameworks underpinning derived algebraic geometry, KK-theory, and motivic filtrations (Gregoric, 23 Apr 2025, Gregoric, 2021).

A cross-disciplinary technical understanding of spectral stacks is therefore necessary for high-fidelity data analysis in astronomy, fundamental research in algebraic topology, and the development of computational pipelines for large-scale, noise-dominated spectral datasets.

---

References:

• Weighted stacking of astronomical images: (Murgia et al., 2024)
• Nonconnective spectral stacks and periodization: (Gregoric, 23 Apr 2025, Gregoric, 2021)
• Spectral correspondence and stacks in Higgs moduli: (Carbone, 2020)
• Multiplexed VPHG stacks in optics: (Alessio et al., 2017)