Multi-Point Second Order Structure Functions

• Multi-point second order structure functions are statistical tools that isolate and quantify small-scale fluctuations by systematically subtracting smooth, large-scale polynomial trends.
• They extend conventional two-point methods using higher-order finite-difference stencils to filter out gradients in inhomogeneous, turbulent fields.
• Applications span astrophysical, geophysical, and hydrodynamic turbulence diagnostics, enabling clearer separation between coherent motions and stochastic turbulent cascades.

A multi-point second order structure function is a statistical diagnostic designed to isolate and quantify fluctuations in turbulent or otherwise spatially inhomogeneous fields while systematically filtering out contributions from large-scale gradients or smooth trends. This methodology extends the familiar two-point structure function by using higher-order finite-difference stencils, which remove polynomial trends up to degree $p-2$ for a $p$-point function, thereby enabling the analysis of genuine small-scale turbulence even in the presence of strong shear, rotation, or periodic coherent structures that can otherwise dominate conventional diagnostics. Multi-point structure functions have become fundamental in modern turbulence research, particularly in astrophysical and geophysical applications, where flows are rarely statistically homogeneous.

1. Formal Mathematical Definition

Let $v(x)$ be a scalar or vector component of a physical field—typically velocity—evaluated along a spatial coordinate $x$ (or more generally, at coordinate $\mathbf{x}$). The two-point second-order structure function is defined as

$S_{2pt}(r) = \langle |v(x + r) - v(x)|^2 \rangle$

where $\langle \cdot \rangle$ denotes averaging over all $x$. Multi-point structure functions generalize this by using symmetric finite-difference operators that annihilate polynomials of degree up to $p-2$:

$$S_{p\mathrm{-}pt}(r) = \frac{1}{C_p} \langle | \sum_{j=0}^{p-1} (-1)^j \binom{p-1}{j} v(x + (j - (p-1)/2)r) |^2 \rangle$$

where $C_p = \sum_{j=0}^{p-1} [\binom{p-1}{j}]^2$ normalizes the sum of squared coefficients. For example:

• 3-point: $$S_{3pt}(r) = (1/3) \langle |v(x-r) - 2v(x) + v(x+r)|^2 \rangle$$
• 4-point: $$S_{4pt}(r) = (1/10) \langle |v(x-r) - 3v(x) + 3v(x+r) - v(x+2r)|^2 \rangle$$
• 7-point (scalar field): weights $\{+1, -6, +15, -20, +15, -6, +1\}$ and normalization 462 (Lee et al., 8 Oct 2025)

This construction ensures that any smooth signal locally represented by a polynomial of degree up to $p-2$ is exactly subtracted, leaving the contribution from fluctuations on scales near $r$.

2. Advantages for Turbulence Diagnostics

Conventional two-point structure functions are sensitive to large-scale gradients, shear, and system-scale coherent motions. In compressible, rotating, or otherwise non-homogeneous flows, these low-wavenumber components can dominate the statistics for large $r$, masking inertial-range scaling or genuine small-scale turbulence. Multi-point structure functions of order $p \geq 3$ are immune to polynomial trends of degree up to $p-2$: for instance, the 3-point difference $v(x-r)-2v(x)+v(x+r)$ vanishes for any linear trend.

This property has particular significance in astrophysical and galactic flows, where gradients from rotation, outflow, or large-scale shear can overwhelm the inertial-range scaling in the two-point SF. The use of higher-order stencils in multi-point SFs enables extraction of true small-scale structure even in strongly inhomogeneous fields (Goldman, 21 Jan 2026, Lee et al., 8 Oct 2025).

3. Computational Procedures and Implementation

The implementation of multi-point second order structure functions entails:

• Data Preparation: Interpolating discrete or noisy measurements to obtain a continuous representation $v(x)$, as done for lens-corrected galaxy data sampled at pixel size $\Delta x$ (Goldman, 21 Jan 2026).
• Summation and Averaging: For each lag $r = m\Delta x$, the multi-point increment is calculated at all permitted $x$ so that the argument of $v$ in the stencil remains within the domain. The structure function is then obtained by averaging $|\text{increment}|^2$ over all valid $x$.
• Normalization: Each structure function is normalized by the sum of squared weights to ensure comparability across orders.
• Range Selection: The minimum lag is set by the data sampling interval; the maximum by the requirement that all stencil points fall within the domain ($r_{\mathrm{max}} \approx L/\lceil(p-1)/2\rceil$ for length $L$).
• Scaling and Plateau Analysis: The structure function is often plotted as $\log S_{p\mathrm{-}pt}$ versus $\log r$. A rising power law indicates a cascade; a plateau or turnover marks the outer scale of the corresponding turbulent range.

4. Empirical Results and Physical Interpretation

Astrophysical Turbulence

In the gravitationally lensed, star-forming galaxy CSWA13 at $z=1.87$, Goldman (2024) employed 3- through 6-point structure functions of nebular and outflowing wind velocity fields, finding:

• The large-scale (two-point SF) velocity increments followed a Burgers-like (compressible) cascade up to the global scale ($L \approx 6.4~\mathrm{kpc}$), with scaling exponent $\alpha \approx 1$.
• Multi-point SFs revealed pronounced plateaus at $l_s \approx 240~\mathrm{pc}$ (nebular gas) and $l_s \approx 290~\mathrm{pc}$ (wind), with corresponding velocity dispersions $$\sigma_s \approx 1.8-1.9~\mathrm{km}\,\mathrm{s}^{-1}$$.

The plateau onset identifies the largest scale of small-scale turbulence: below $l_s$ the structure function ceases to rise, indicating energy injection at these smaller scales, plausibly by stellar sub-clumps or compact massive star clusters. This is interpreted as direct evidence for superposition of global, merger-driven turbulence and localized, feedback-driven turbulent cascades (Goldman, 21 Jan 2026).

Interstellar Medium Studies

Seven-point structure functions applied to H I emission in the Small Magellanic Cloud provide clean separation between large-scale and small-scale turbulent components (Lee et al., 8 Oct 2025). The seven-point SF isolated break features at 34–84 pc (median ∼50 pc), corresponding to local feedback from supernova shells and star formation. Correlations with indicators of stellar feedback (e.g., Hα intensity, young stellar object count, H I shell density) are strong for the seven-point SF slope but negligible for the two-point SF. This suggests that higher-order SFs are required to diagnose small-scale turbulent driving mechanisms where large-scale organization is dominant.

5. Structure Function Tensors, Budgets, and Analytical Extensions

In tensor-valued, inhomogeneous, or anisotropic turbulence, the structure function is generalized to the second-order structure-function tensor:

$$S_{ij}(\mathbf{X}, \mathbf{r}, t) = \langle [u_i(\mathbf{X}+\mathbf{r}/2, t) - u_i(\mathbf{X}-\mathbf{r}/2, t)][u_j(\mathbf{X}+\mathbf{r}/2, t) - u_j(\mathbf{X}-\mathbf{r}/2, t)] \rangle$$

The Anisotropic Generalized Kolmogorov Equations (AGKE) provide exact dynamical budgets for all components of $S_{ij}$ (including off-diagonal terms), fully accounting for production, spatial and scale-space transport, inter-component energy redistribution (via pressure-strain), and viscous dissipation (Gatti et al., 2020, Gattere et al., 2023). Further extensions—such as triple decomposition—enable the AGKE formalism ($\varphi$AGKE) to resolve coherent and stochastic contributions and analyze their interplay in phase-resolved flows or flows with periodic coherent structures.

Unlike spectral or single-point Reynolds-stress approaches, tensor structure functions and their budgets give simultaneous space- and scale-resolved information, including in strongly inhomogeneous or anisotropic domains. This is essential for dissecting the full multi-scale structure of complex turbulence, as in the wall-bounded, separated, or periodic flows detailed in (Gattere et al., 2023, Gatti et al., 2020).

6. Key Applications and Impact

Multi-point second order structure functions have found application in:

• Astrophysics: Unraveling the coexistence of galactic-scale and local, feedback-driven turbulence in galaxies and the ISM (Goldman, 21 Jan 2026, Lee et al., 8 Oct 2025).
• Hydrodynamics: Quantifying wall-normal and spanwise scale interactions in channel flows, or resolving phase-locked energy redistribution in periodically forced turbulence (Gattere et al., 2023).
• Flow Diagnostics: Providing robust, scale- and position-resolved turbulence diagnostics in the presence of strong large-scale gradients or coherent modes.

The ability of high-order structure functions to cleanly subtract smooth, large-scale organization and reveal the genuine turbulent cascade at small scales is pivotal in both empirical and simulation-based turbulence research.

7. Limitations, Assumptions, and Future Directions

Multi-point structure functions rely on several key assumptions and face specific limitations:

• Polynomial Removal: Only polynomial trends up to degree $p-2$ are removed; non-polynomial or highly non-smooth large-scale features may still contaminate the results (Lee et al., 8 Oct 2025).
• Sampling: The maximum usable lag $r$ is limited by the available data domain; high-order SFs require more contiguous data points, reducing the maximum scale analyzable (Goldman, 21 Jan 2026).
• Interpretive Limits: Identification of turnover or plateau scales is often visual or heuristic; more automated or quantitatively rigorous diagnostics remain active research topics.
• Formal Validity: AGKE and their triple-decomposed generalizations are formally exact but impose high-dimensional data and ensemble-averaging requirements that can limit their practical implementation (Gattere et al., 2023).
• Physical Interpretation: Extraction of driving mechanisms from the observed behavior of multi-point SFs frequently depends on circumstantial evidence (e.g., spatial coincidence with star-forming clumps or supernova shells) and is rarely definitive without supporting multi-wavelength or dynamical evidence.

A plausible implication is that, as simulation and observation datasets grow in size and fidelity, multi-point structure function methodologies will further increase their reach, especially in regimes lacking statistical homogeneity or with mixed coherent/stochastic dynamics.

---

References:

• (Goldman, 21 Jan 2026)
• (Lee et al., 8 Oct 2025)
• (Gattere et al., 2023)
• (Gatti et al., 2020)