Intrinsic Alignment Bispectra

• Intrinsic alignment bispectra are three-point correlation functions that quantify shape and density correlations of galaxies and dark matter halos.
• They decompose into scalar (E-mode) and pseudo-scalar (B-mode) components, exhibiting distinct angular, parity, and configuration dependencies.
• They are modeled using techniques like the TATT framework and effective field theory, improving cosmological parameter constraints and mitigating systematics.

Intrinsic alignments (IA) of galaxies or dark matter halos generate shape correlations whose three-point functions—the intrinsic alignment bispectra—constitute significant systematics and astrophysical signals in cosmic shear and large-scale structure analyses. The IA bispectra involve cross-correlating the intrinsic shape field (represented as a spin-2 tensor) and the density field, leading to distinct angular, parity, and configuration dependence compared to matter or lensing-only bispectra. The following sections outline the mathematical structure, physical interpretation, and practical measurement of intrinsic alignment bispectra and their components.

1. Structure and Definition of IA Bispectra

The IA bispectra are three-point correlation functions involving the matter density field (δ) and the intrinsic shape field, often decomposed into scalar (E-mode) and pseudo-scalar (B-mode) components. For a generic trio of fields $X, Y, Z \in \{\delta, E, B\}$, the bispectrum is defined in Fourier space as

$$\langle X(\mathbf{k}_1) Y(\mathbf{k}_2) Z(\mathbf{k}_3) \rangle = (2\pi)^3 \delta_D(\mathbf{k}_1 + \mathbf{k}_2 + \mathbf{k}_3) B_{XYZ}(\mathbf{k}_1, \mathbf{k}_2, \mathbf{k}_3)$$

The spin-2 shear field $\gamma$ is typically constructed from projected shapes (triaxial axes or inertia tensors), and then decomposed into E- and B-modes: $$\tilde{\gamma}_E(\mathbf{k}) = \tilde{\gamma}_+(\mathbf{k}) \cos 2\varphi + \tilde{\gamma}_\times(\mathbf{k}) \sin 2\varphi$$

$$\tilde{\gamma}_B(\mathbf{k}) = \tilde{\gamma}_\times(\mathbf{k}) \cos 2\varphi - \tilde{\gamma}_+(\mathbf{k}) \sin 2\varphi$$

where $\varphi$ is the polar angle of the transverse component of the wavevector. The decomposition is essential, as only specific combinations of E and B survive under statistical isotropy and parity constraints, detailed in (Bakx et al., 14 Apr 2025, Vlah et al., 2019).

In practice, five bispectra are commonly encountered in the absence of parity violation:

• $B_{\delta\delta E}$: two density and one E-mode shape field (the dominant IA-lensing contamination)
• $B_{\delta EE}$: one density and two shape fields (quadratic IA effects)
• $B_{EEE}$: pure IA bispectrum (auto-correlation of shapes)
• $B_{\delta\delta B}$ and $B_{\delta EB}$: parity-odd analogues (nonzero at three-point order even in parity-conserving universes)

2. Classification: Bispectrum Components

Intrinsic alignment bispectra are classified according to their field composition, with special nomenclature in the projected (observational) context:

Configuration	Fourier Space Notation	Legacy Notation	Description
δδδ	$B_{\delta\delta\delta}$	GGG	Pure matter/lensing bispectrum (benchmark)
δδE	$B_{\delta\delta E}$	GGI (Igg)	Two lensing, one IA; leading IA contamination
δEE	$B_{\delta EE}$	GII (IIg)	One lensing, two IA; quadratic IA
EEE	$B_{EEE}$	III	Pure IA; dominates on small angular scales
Mix w/ B-mode	$B_{\delta\delta B}$, etc	—	Parity-odd contributions

Parity properties dictate which bispectra are nonzero and their symmetry under triangle reflection. Parity-even bispectra involve zero or two B-modes; parity-odd involve one or three. All combinations (DDD, DDE, DDB, DEE, DEB, DBB, EEE, EEB, EBB, BBB) are in principle allowed (Bakx et al., 14 Apr 2025).

3. Physical Models and Bias Expansion

The origin and modeling of the IA bispectrum components rely on nonlinear shape–density relations. The most common framework is the Tidal Alignment and Tidal Torquing (TATT) model (Gomes et al., 14 Jan 2026), which expands the intrinsic shape field in gravitational operators to second (or higher) order: $$\gamma^I_{ij}(\mathbf{x}) = C_1 s_{ij} + C_{1\delta}[\delta s_{ij}] + C_2 \left[s_{ik}s_{kj} - \frac{1}{3}\delta_{ij}s^2\right] + C_t t_{ij}$$ where $s_{ij}$ is the tidal shear, $t_{ij}$ is the velocity-shear operator (in extended models), and $C$ coefficients are IA amplitudes. These map onto bispectrum components via distinct physically motivated kernels (see Sec. 2 of (Gomes et al., 14 Jan 2026)).

In effective field theory treatments, the shape field is decomposed into irreducible tensor modes, and the IA bispectra are expanded via kernels $K^{(1)}$, $K^{(2)}$ associated with first and second-order bias coefficients $b_1^g, b_{2,1}^g, b_{2,2}^g, b_{2,3}^g$ (Bakx et al., 9 Jul 2025, Vlah et al., 2019). Pure quadratic terms generate vector and tensor helicity modes in the bispectrum, which are unequivocal signatures of nonlinear IA physics (Akitsu et al., 2023).

4. Triangle Configuration and Symmetry Dependence

IA bispectra components display strong dependence on triangle configuration (shape and orientation) as well as projection parameters:

• Equilateral configurations: $B_{EEE}$ dominates over the lensing $B_{GGG}$ at high multipole ($\ell \gtrsim 600$); their amplitude ratio reaches $\sim 10$ at $\ell\!\sim\!3000$ (Merkel et al., 2014).
• Squeezed or flattened triangles: The configuration dependence shifts the contamination regime, facilitating "intrinsic-safe" angular scale ranges ($\ell \lesssim 500$) where cosmic shear dominates.
• Parity and angular multipoles: Multipole expansions in spherical harmonics and associated Legendre polynomials isolate line-of-sight and transverse contributions. Parity-odd bispectra, such as $B_{\delta\delta B}$, are nonzero at the three-point level even in the absence of parity violation, due to the reflection properties of triangle configurations (Bakx et al., 14 Apr 2025, Bakx et al., 9 Jul 2025).

The configuration dependence enables the separation of IA and lensing signals and significantly aids in self-calibration and mitigation.

5. Measurement, Modeling, and Forecasts

Simulations (e.g., IllustrisTNG) and recent analytic work provide quantitative modeling and direct measurement of IA bispectrum components:

• The dominant parity-even bispectrum $B_{\delta\delta E}$ is detected at $>30\sigma$ significance in $1\,(\mathrm{Gpc}/h)^3$ simulations for $k \lesssim 0.11\,h/\mathrm{Mpc}$; $B_{\delta EE}$ and $B_{EEE}$ are also detected but are more affected by stochastic shape noise (Bakx et al., 9 Jul 2025, Pyne et al., 2022).
• Parity-odd bispectra ($B_{\delta\delta B}, B_{\delta EB}$) are measured at $>10\sigma$ and match parity-even sector predictions, confirming theoretical expectations (Bakx et al., 9 Jul 2025, Bakx et al., 14 Apr 2025).
• Bias parameters inferred from bispectrum fit are consistent with, and greatly tightened compared to, those from the power spectrum. Multipole decomposition breaks bias degeneracies, enabling factor-of-5 reduction in errors for quadratic bias terms (Bakx et al., 9 Jul 2025, Akitsu et al., 2023).
• In projected observables, the IA bispectrum contamination to lensing is typically $\sim5$–$20\%$ for $B_{\delta\delta E}$ at Stage III/IV survey precision, smaller but still significant contributions for $B_{\delta EE}$ and $B_{EEE}$ (Gomes et al., 14 Jan 2026).

Fast Fourier Transform techniques enable efficient estimation of all bispectrum multipoles, both parity-even and odd (Bakx et al., 14 Apr 2025).

6. Implications for Cosmology and Weak Lensing

IA bispectra are a primary non-Gaussian contamination in cosmic shear three-point statistics. Their unique amplitude and configuration dependence provide several crucial consequences:

• Systematics for lensing surveys: For Euclid-type experiments, intrinsic alignments dominate the observed bispectrum at small angular scales, presenting a primary limiting systematic for cosmological inference (Merkel et al., 2014).
• Self-calibration: The redshift and configuration dependence of III, GII, and GGI bispectrum components allow model-independent self-calibration, reducing IA contamination by factors of 5–10 at $\ell > 300$ (Troxel et al., 2012, Troxel et al., 2012).
• Parameter constraints: Joint analysis of two- and three-point statistics in the TATT or EFT frameworks enables breaking of degeneracies among IA and cosmological parameters and robust estimation of intrinsic alignment amplitudes (Gomes et al., 14 Jan 2026).

7. Future Directions and Methodological Advances

Precision modeling and measurement of IA bispectra open multiple avenues:

• Incorporation into cosmological analyses: The consistent EFT/TATT framework validated at both two- and three-point levels allows systematic inclusion of IA bispectra as both contaminants and probes.
• Use of configuration and parity information: The detection of vector and tensor bispectrum components, unattainable by simpler models, provides new information on nonlinear gravitational and galaxy formation physics (Akitsu et al., 2023).
• Synergy of imaging and spectroscopic surveys: Forecasts for DESI × LSST indicate strong ($\mathrm{S/N}\sim30$) detection of IA bispectrum monopoles, and probe of parity-odd sectors ($\mathrm{S/N}\sim5$) for sample sizes $N\sim10^6$ (Bakx et al., 14 Apr 2025).

The precise characterization of IA bispectra will be essential for Stage IV survey science, both to control bias in lensing and as a fundamental astrophysical observable measuring the interplay of tidal fields, galaxy formation, and large-scale structure.