The post-inflationary CMB bispectrum of metric $f(R)$ gravity: full second-order transfer, templates, and observational confrontation

In metric $f(R)$ gravity, the second-order (nonlinear) Sachs-Wolfe transfer from the primordial curvature perturbation to the CMB temperature anisotropy depends on the nonlinear structure of $f(R)$ through the dimensionless combinations $\beta=Rf_{,RR}/f_{,R}$, $\gamma=R^2f_{,RRR}/f_{,R}$, $\delta=R^3f_{,RRRR}/f_{,R}$, and its coupling coefficients $b_0,b_1$ can in principle be large, unlike in general relativity where the intrinsic (post-inflationary) bispectrum sits at the $f_{\rm NL}\sim\mathcal{O}(1)$ level. Gao (2011) computed only the large-scale Sachs-Wolfe limit during matter domination and left the rest open: the full second-order Boltzmann radiative transfer in $f(R)$ gravity on all angular scales, the viable $(\beta,\gamma,\delta)$ region, dedicated bispectrum templates disentangling the $f(R)$ signal from primordial non-Gaussianity, and the confrontation with CMB bispectrum data have never been worked out. The open problem is to compute the $f(R)$-induced intrinsic CMB bispectrum with the full second-order transfer, construct its template, quantify its degeneracies with primordial and GR-secondary templates, and determine the resulting observational constraint on $f(R)$ gravity or the bias it induces in measurements of primordial non-Gaussianity.

Background

The Cosmic Microwave Background (CMB) is the relic radiation from the early universe, and its temperature anisotropies are very nearly Gaussian; the leading deviation is the bispectrum, the Fourier-space three-point correlation of the anisotropy $\Delta T/T$. Non-Gaussianity has two conceptually distinct origins. Primordial non-Gaussianity is generated in the very early universe in the statistics of the curvature perturbation $\zeta$ (the standard variable describing the initial fluctuation on uniform-density slices, conserved outside the horizon). Post-inflationary (also called intrinsic or secondary) non-Gaussianity instead arises from the nonlinear transfer of $\zeta$ into anisotropies: the nonlinear relation between the metric potentials and $\zeta$, the intrinsic temperature fluctuation on the last-scattering surface, and the gravitational redshift of photons climbing out of the potential wells — the Sachs-Wolfe effect. In general relativity (GR) this transfer has been computed to second order in the perturbations: the large-scale kernel is $b(p_1,p_2)=1/25-(3/50)g(p_1,p_2)$ with a universal momentum-shape factor $g$, and the full intrinsic bispectrum, computed with second-order Boltzmann codes (the Einstein-Boltzmann equations for photons and baryons expanded to second order in perturbations, including recombination microphysics), biases local-type primordial non-Gaussianity only at the level $f_{\rm NL}^{\rm intra}\simeq 0.5$ ($f_{\rm NL}$ is the standard amplitude parameter of the bispectrum, and 'local-type' denotes its squeezed-limit shape, the configuration most sensitive to super-horizon physics), far below current sensitivity. Metric $f(R)$ gravity is the simplest modified-gravity extension of GR, in which the Einstein-Hilbert Lagrangian is replaced by a general function $f(R)$ of the Ricci scalar $R$ in the action $S=\int d^4x\sqrt{-g}\,(f(R)/2+\mathcal{L}_m)$; the field equations contain up to fourth derivatives of $f$, and the theory carries one extra scalar degree of freedom (the scalaron). Gao (2011) computed the ordinary Sachs-Wolfe contribution to $\Delta T/T$ up to second order in the metric potential in metric $f(R)$ gravity, during matter domination, for adiabatic perturbations of a perfect fluid. The nonlinear mapping $\Delta T/T = (\Delta T/T)^{(1)}+(\Delta T/T)^{(2)}+\cdots$ takes the convolution form $(\Delta T/T)^{(2)}(k)=\frac{1}{2}\int d^3p_1\,b(p_1,k-p_1)\,\zeta_{p_1}\zeta_{k-p_1}$ with $b(p_1,p_2)=b_0-b_1 g(p_1,p_2)$, where $g$ is the same universal shape factor as in GR but the coefficients $b_0,b_1$ depend on the nonlinear structure of $f$ through $\beta=Rf_{,RR}/f_{,R}$, $\gamma=R^2f_{,RRR}/f_{,R}$, $\delta=R^3f_{,RRRR}/f_{,R}$ and background slow-roll combinations, evaluated at the emission time. In the squeezed limit (one momentum much smaller than the others) $b\to b_0$, so a large $b_0$ mimics local-type primordial non-Gaussianity; in GR the coefficients are tiny, but in $f(R)$ they can in principle be large, offering both a contaminant of primordial non-Gaussianity measurements and a new, independent test of modified gravity. Linear perturbation stability requires the Compton parameter $B=-\beta\epsilon_R/\epsilon\ge 0$ (with $\epsilon=-d\ln H/d\ln a$, $\epsilon_R=d\ln R/d\ln a$), and linear CMB and large-scale-structure observations already force viable $f(R)$ models (such as the Hu-Sawicki class, a standard family of $f(R)$ functions built to mimic a cosmological constant at late times while passing local gravitational tests through the chameleon screening mechanism, which suppresses the extra scalar force in dense regions) to deviate very little from GR at early epochs. Separately, CMB bispectrum constraints on modified gravity do exist along a different route — the late-time integrated Sachs-Wolfe (ISW)-lensing and lensing bispectra, used to forecast or derive constraints on parametrized modified gravity and $f(R)$-type models — but these late-time secondaries do not touch the intrinsic second-order radiative transfer that is specific to $f(R)$. What has never been done is the $f(R)$ generalization of the GR intrinsic-bispectrum program: the full second-order Boltzmann treatment on all angular scales, its evaluation for viable $f(R)$ models, and the corresponding templates, degeneracy analysis, and observational confrontation.

Problem Statement

Determine quantitatively the intrinsic (post-inflationary) CMB bispectrum induced by metric $f(R)$ gravity and its observational consequence. The common scientific objective is to establish whether the nonlinear radiative transfer in $f(R)$ gravity produces a detectable and identifiable CMB bispectrum signal, and what it implies — a new constraint on $f(R)$ gravity or a bias on inferred primordial non-Gaussianity. Achieving it jointly requires: (i) deriving and solving the second-order Boltzmann equations for photons and baryons in metric $f(R)$ gravity, including the radiation era and recombination microphysics, so that the nonlinear transfer from the primordial curvature perturbation $\zeta$ to the CMB temperature (and polarization) anisotropy is computed on all angular scales — reducing, in the large-scale matter-dominated limit, to the Sachs-Wolfe kernel $b(p_1,p_2)=b_0-b_1 g(p_1,p_2)$ of Gao (2011), and to the GR intrinsic bispectrum when the $f(R)$ parameters $\beta=\gamma=\delta=0$; (ii) evaluating the resulting bispectrum over the region of $f(R)$ parameter space — equivalently the allowed values of $\beta,\gamma,\delta$ at decoupling, or of model parameters such as the present-day scalaron amplitude — consistent with the background expansion, linear-perturbation stability, and existing observational constraints; (iii) constructing the bispectrum template(s) of the $f(R)$ post-inflationary signal and quantifying its degeneracies with known primordial templates (local, equilateral, orthogonal) and with GR secondary bispectra (the intrinsic bispectrum, the ISW-lensing bispectrum, the lensing bispectrum); and (iv) confronting the result with current CMB bispectrum measurements to derive the implied constraint on the $f(R)$ parameter space, or equivalently the bias it induces on inferred primordial $f_{\rm NL}$, with forecasts for future data. A rigorous demonstration that the signal is negligible (or undetectable) across the entire viable parameter space is a legitimate resolution of the same objective. Generalizations beyond metric $f(R)$, adiabatic initial conditions, or perfect-fluid matter are optional extensions and not part of the core objective.

References

1. Xian Gao, Testing gravity with non-Gaussianity, Phys. Lett. B 702 (2011) 197-200, arXiv:1008.2123, DOI 10.1016/j.physletb.2011.07.022, https://arxiv.org/abs/1008.2123
2. N. Bartolo, S. Matarrese, A. Riotto, Non-Gaussianity and the Cosmic Microwave Background Anisotropies, Adv. Astron. 2010 (2010) 157079, arXiv:1001.3957, DOI 10.1155/2010/157079, https://arxiv.org/abs/1001.3957
3. Xian Gao, Nonlinear CMB temperature anisotropy from gravitational perturbations, Phys. Rev. D 82 (2010) 103004, arXiv:1005.1219, DOI 10.1103/PhysRevD.82.103004, https://arxiv.org/abs/1005.1219
4. Yong-Seon Song, Wayne Hu, Ignacy Sawicki, Large scale structure of f(R) gravity, Phys. Rev. D 75 (2007) 044004, arXiv:astro-ph/0610532, DOI 10.1103/PhysRevD.75.044004, https://arxiv.org/abs/astro-ph/0610532
5. Guido W. Pettinari, Christian Fidler, Robert Crittenden, Kazuya Koyama, David Wands, The intrinsic bispectrum of the Cosmic Microwave Background, JCAP 04 (2013) 003, arXiv:1302.0832, DOI 10.1088/1475-7516/2013/04/003, https://arxiv.org/abs/1302.0832
6. Pedro Carrilho, Karim A. Malik, The intrinsic bispectrum of the CMB from isocurvature initial conditions, JCAP 08 (2021) 046, arXiv:2104.10204, DOI 10.1088/1475-7516/2021/08/046, https://arxiv.org/abs/2104.10204
7. Eleonora Di Valentino, Alessandro Melchiorri, Valentina Salvatelli, Alessandra Silvestri, Parametrized modified gravity and the CMB bispectrum, Phys. Rev. D 86 (2012) 063517, arXiv:1204.5352, DOI 10.1103/PhysRevD.86.063517, https://arxiv.org/abs/1204.5352
8. Bin Hu, Michele Liguori, Nicola Bartolo, Sabino Matarrese, Future CMB integrated-Sachs-Wolfe-lensing bispectrum constraints on modified gravity in the parametrized post-Friedmann formalism, Phys. Rev. D 88 (2013) 024012, arXiv:1211.5032, DOI 10.1103/PhysRevD.88.024012, https://arxiv.org/abs/1211.5032
9. Toshiya Namikawa, Francois R. Bouchet, Atsushi Taruya, CMB lensing bispectrum as a probe of modified gravity theories, Phys. Rev. D 98 (2018) 043530, arXiv:1805.10567, DOI 10.1103/PhysRevD.98.043530, https://arxiv.org/abs/1805.10567
10. Anton Chudaykin, Martin Kunz, Julien Carron, Modified gravity constraints with the Planck ISW-lensing bispectrum, Phys. Rev. D 112 (2025) 083537, arXiv:2503.09893, DOI 10.1103/9zjh-8htv, https://arxiv.org/abs/2503.09893
11. Lucas Lombriser, Constraining chameleon models with cosmology, Annalen der Physik 526 (2014) 259-285, arXiv:1403.4268, DOI 10.1002/andp.201400058, https://arxiv.org/abs/1403.4268