betaFeedback

WorksPhysics and Astronomy

Probing Lorentz-violating electrodynamics with CMB polarization. II. Updated constraints and forecasts for the Simons Observatory and LiteBIRD

Abstract

We update the constraints on Lorentz-violating electrodynamics from the polarization of the cosmic microwave background (CMB). Within the minimal Standard Model Extension, we consider the renormalizable CPT-odd and CPT-even operators of the photon sector, which source cosmic birefringence and circular polarization, respectively. Combining Planck, Atacama Cosmology Telescope (ACT) and South Pole Telescope (SPT) temperature and EE-mode data with BB-mode measurements from BICEP/Keck, SPT-3G D1 and SPTpol, we constrain the CPT-odd coefficients to ∣k(V)00(3)∣<7.75×10−45|k^{(3)}_{(V)00}|<7.75\times10^{-45} GeV and ∣kAF∣<3.82×10−45|\mathbf{k_{AF}}|<3.82\times10^{-45} GeV at 95% credible level. These are the strongest bounds to date from any probe, and are driven by the SPT-3G D1 BB-mode data. For the first time in this type of analysis, we also include the Planck PR4 EBEB spectrum, which yields a mild preference for a non-zero ∣k(V)00(3)∣|k^{(3)}_{(V)00}|. For the CPT-even coefficient, we find kF,E+B<5.99×10−32k_{F,E+B}<5.99\times10^{-32}. Including the latest circular polarization measurements from the Cosmology Large Angular Scale Surveyor (CLASS) tightens this bound by about 7%. This is the first time that CMB circular polarization data contribute to constraining this coefficient. Finally, we forecast the sensitivity of the Simons Observatory (SO) and LiteBIRD. With linear polarization only, the two experiments are complementary, with SO tightening the CPT-even bound by about 30% and LiteBIRD the CPT-odd ones by 40 to 45%. Exploiting the sensitivity to circular polarization provided by the non-idealities of its half-wave plate, LiteBIRD could also improve the CPT-even bound by more than an order of magnitude.

E-mail: [email protected], [email protected]

Introduction

Lorentz invariance is one of the fundamental symmetries of modern physics, underlying both the Standard Model of particle physics and general relativity. Several approaches to quantum gravity nonetheless suggest that this symmetry might be violated at very high energies, with residual violations potentially surviving down to observable scales [1, 2, 3]. Searching for such violations is therefore a powerful way to probe physics beyond the Standard Model.

A general and widely used framework for parametrizing possible deviations from exact Lorentz invariance is the Standard Model Extension (SME) [4, 5], an effective field theory that extends the Standard Model by adding all Lorentz-violating (LV) operators consistent with the underlying gauge symmetries, organized by their mass dimension. In the photon sector, the leading renormalizable operators are a CPT-odd, dimension-3 term and a CPT-even, dimension-4 term. Both modify the propagation of photons, with an effect that grows with the distance they travel, making astrophysical and cosmological sources particularly sensitive probes. Among these, the cosmic microwave background (CMB) is the most distant source of electromagnetic radiation available in nature and is linearly polarized by Thomson scattering at recombination and reionization [6]. Any modification of its polarization induced by these operators is therefore integrated over the longest possible path, so that even tiny Lorentz violations can leave a measurable imprint [7, 8]. Concretely, the CPT-odd operator induces a rotation of the linear polarization plane of CMB photons, an effect known as cosmic birefringence; its time component sources an isotropic rotation, while its space components source an anisotropic one. The isotropic rotation gives rise to parity-violating EBEB and TBTB correlations, absent in the standard cosmological model, while both induce a leakage of EE- into BB-mode polarization. Isotropic birefringence is particularly relevant today, since analyses of Planck and Wilkinson Microwave Anisotropy Probe (WMAP) data have found hints of a non-zero rotation angle [9, 10, 11], recently supported by a consistent measurement from the sixth data release (DR6) of the Atacama Cosmology Telescope (ACT) [12]. If confirmed to be of cosmological origin, and not due to a residual miscalibration of the detector polarization angles, such a signal would be evidence of parity-violating physics beyond the Standard Model. Proposed mechanisms include a pseudoscalar coupling between photons and an axion-like field [13, 14, 15, 16], which may constitute dark energy or dark matter [17, 18, 19, 20, 21] or form networks of domain walls and cosmic strings [22, 23, 24, 25], a vector Chern-Simons coupling between photons and the neutrino current in the presence of a large lepton asymmetry [26, 27, 28], and the LV electrodynamics considered in this work. The CPT-even operator, instead, converts linear into circular polarization, sourcing a non-zero VVVV spectrum. Circular polarization is not generated by scattering at recombination, and the mechanisms that can produce it within the standard cosmological model yield a negligible signal [29, 30]. A detection of VV modes in the CMB would therefore be a signature of beyond-Λ\LambdaCDM physics [31, 32, 33, 34, 35, 36, 37, 38, 39], including Lorentz violation.

In [8] (hereafter Paper I), we set up the framework to test LV electrodynamics with CMB polarization. Building on the effective susceptibility tensor formalism developed in [40], we mapped the LV operators onto phenomenological parameters that source isotropic and anisotropic birefringence and circular polarization in the CMB, and constrained them with data from Planck 2018 [41], BICEP/Keck 2018 [42] and ACT DR4 [43]. In this work, we update and extend the analysis of Paper I along several lines. First, we replace ACT DR4 with DR6 [44], and include data from the South Pole Telescope (SPT). From the third-generation receiver SPT-3G, we use the recent small-scale temperature and polarization measurements of the D1 data release [45] and the dedicated BB-mode measurements from the same dataset [46]. We complement these with the earlier BB-mode measurements from the previous-generation receiver SPTpol [47]. Second, for the first time in this type of analysis, we include the parity-violating EBEB spectrum from the Planck PR4 Lollipop likelihood [48, 49], which is expected to vanish in the standard cosmological model and is therefore a particularly clean probe of isotropic birefringence. Third, we use the latest circular polarization measurements from the Cosmology Large Angular Scale Surveyor (CLASS) at 90, 150 and 220 GHz [50]. Finally, we forecast the sensitivity of the ground-based Simons Observatory (SO) [51], which has already started science observations, and of the LiteBIRD satellite [52]. The two experiments probe complementary angular scales. SO measures temperature and polarization anisotropies from degree to arcminute scales, while LiteBIRD will target the largest angular scales over the full sky. Following the approach we introduced in [53], we show that the non-idealities of the LiteBIRD half-wave plate could provide competitive sensitivity to circular polarization, and hence to the CPT-even operator.

The paper is organized as follows. In § 2 we summarize the LV model, its impact on the CMB power spectra and the mapping between the phenomenological parameters and the SME coefficients. In § 3 we describe the datasets and the methodology adopted for current data and forecasts, and in § 4 we present the resulting constraints and recast them in terms of the SME coefficients. We conclude in § 5. Appendices § 7 and § 8 discuss the role of individual BB-mode datasets and the degeneracy with the polarization calibration parameters of ACT and SPT.

Model and CMB spectra

In this section we summarize the main aspects of the Lorentz-violating model relevant for our analysis and its impact on the CMB spectra, referring the reader to [8, 40] for more details. We work within the minimal SME [4, 5], considering renormalizable operators that extend Maxwell electrodynamics. The corresponding action reads

S=∫d4x−g[−14FμνFμν+12ϵαβμνAβ(kAF)αFμν−14(kF)αβμνFαβFμν](1)S = \int d^{4}x \sqrt{-g}\left[-\frac{1}{4}F_{\mu\nu}F^{\mu\nu}+\frac{1}{2}\epsilon^{\alpha\beta\mu\nu}A_{\beta}(k_{AF})_{\alpha}F_{\mu\nu}-\frac{1}{4}(k_{F})^{\alpha\beta\mu\nu}F_{\alpha\beta}F_{\mu\nu}\right] \tag*{(1)}

where the vector (kAF)α(k_{AF})_{\alpha} and the tensor (kF)αβμν(k_{F})^{\alpha\beta\mu\nu} encode the CPT-odd and CPT-even Lorentz-violating contributions, respectively. The corresponding operators have mass dimension d=3d = 3 and d=4d = 4, respectively, so that kAFk_{AF} has dimensions of mass while kFk_{F} is dimensionless. As shown in Paper I, these operators modify the propagation of CMB photons between the last scattering surface and today, leaving distinct imprints on the observed polarization.

The impact of these operators on the CMB power spectra is encoded in four phenomenological parameters, βAF,T2\beta_{AF,T}^{2}, βAF,S2\beta_{AF,S}^{2}, βF,E2\beta_{F,E}^{2} and βF,B2\beta_{F,B}^{2}; the first two are built from the time and space components of kAFk_{AF}, respectively, while the latter two are built from specific contractions of kFk_{F} (see section 2.2 and Paper I for more details). In terms of these parameters, the observed CMB spectra, CℓXYC_{\ell}^{XY} with X,Y∈{T,E,B,V}X,Y \in\{T,E,B,V\}, are related to the spectra in the absence of Lorentz violation, C~ℓXY\widetilde{C}_{\ell}^{XY}, by [8]

CℓTE=(1−Z2)C~ℓTE,(2)C_{\ell}^{TE} = \left(1-\frac{\mathcal{Z}}{2}\right)\widetilde{C}_{\ell}^{TE}, \tag*{(2)}
CℓEE=(1−Z)C~ℓEE+∑ℓ1Kℓ1ℓ11C~ℓ1EE+∑ℓ1Kℓ1ℓ22C~ℓ1BB,(3)C_{\ell}^{EE} = (1-\mathcal{Z})\widetilde{C}_{\ell}^{EE} + \sum_{\ell_{1}}\mathcal{K}_{\ell_{1}\ell}^{11}\widetilde{C}_{\ell_{1}}^{EE} + \sum_{\ell_{1}}\mathcal{K}_{\ell_{1}\ell}^{22}\widetilde{C}_{\ell_{1}}^{BB}, \tag*{(3)}
CℓBB=(1−Z)C~ℓBB+∑ℓ1Kℓ1ℓ11C~ℓ1BB+∑ℓ1Kℓ1ℓ22C~ℓ1EE,(4)C_{\ell}^{BB} = (1-\mathcal{Z})\widetilde{C}_{\ell}^{BB} + \sum_{\ell_{1}}\mathcal{K}_{\ell_{1}\ell}^{11}\widetilde{C}_{\ell_{1}}^{BB} + \sum_{\ell_{1}}\mathcal{K}_{\ell_{1}\ell}^{22}\widetilde{C}_{\ell_{1}}^{EE}, \tag*{(4)}
CℓEB=βAF,T2(C~ℓEE−C~ℓBB),(5)C_{\ell}^{EB} = \sqrt{\beta_{AF,T}^{2}}\left(\widetilde{C}_{\ell}^{EE}-\widetilde{C}_{\ell}^{BB}\right), \tag*{(5)}
CℓTB=βAF,T2C~ℓTE,(6)C_{\ell}^{TB} = \sqrt{\beta_{AF,T}^{2}}\widetilde{C}_{\ell}^{TE}, \tag*{(6)}
CℓVV=∑ℓ1Kℓ1ℓ33C~ℓ1EE+∑ℓ1Kℓ1ℓ44C~ℓ1BB,(7)C_{\ell}^{VV} = \sum_{\ell_{1}}\mathcal{K}_{\ell_{1}\ell}^{33}\widetilde{C}_{\ell_{1}}^{EE} + \sum_{\ell_{1}}\mathcal{K}_{\ell_{1}\ell}^{44}\widetilde{C}_{\ell_{1}}^{BB}, \tag*{(7)}
CℓEV=CℓBV=0,(8)C_{\ell}^{EV}=C_{\ell}^{BV}=0, \tag*{(8)}

where we have defined

Z=βAF,T2+βAF,S2+βF,E2+βF,B24,(9)\mathcal{Z}=\beta_{AF,T}^{2}+\beta_{AF,S}^{2}+\frac{\beta_{F,E}^{2}+\beta_{F,B}^{2}}{4}, \tag*{(9)}
∑ℓ1Kℓ1ℓ11C~ℓ1XX=βAF,S24ℓ+ℓ2C~ℓXX,(10)\sum_{\ell_{1}}\mathcal{K}_{\ell_{1}\ell}^{11}\widetilde{C}_{\ell_{1}}^{XX}=\beta_{AF,S}^{2}\frac{4}{\ell+\ell^{2}}\widetilde{C}_{\ell}^{XX}, \tag*{(10)}
∑ℓ1Kℓ1ℓ22C~ℓ1XX=βAF,T2C~ℓXX+βAF,S2[ℓ2−4ℓ(2ℓ+1)C~ℓ−1XX+(ℓ−1)(ℓ+3)(ℓ+1)(2ℓ+1)C~ℓ+1XX],(11)\sum_{\ell_{1}}\mathcal{K}_{\ell_{1}\ell}^{22}\widetilde{C}_{\ell_{1}}^{XX}=\beta_{AF,T}^{2}\widetilde{C}_{\ell}^{XX}+\beta_{AF,S}^{2}\left[\frac{\ell^{2}-4}{\ell(2\ell+1)}\widetilde{C}_{\ell-1}^{XX}+\frac{(\ell-1)(\ell+3)}{(\ell+1)(2\ell+1)}\widetilde{C}_{\ell+1}^{XX}\right], \tag*{(11)}
∑ℓ1Kℓ1ℓ33(44)C~ℓ1XX=βF,B(E)2[(ℓ−2)(ℓ−3)4(4ℓ2−1)C~ℓ−2XX+3(ℓ2+ℓ−2)2(4ℓ2+4ℓ−3)C~ℓXX+(ℓ+3)(ℓ+4)4(4ℓ2+8ℓ+3)C~ℓ+2XX]+βF,E(B)2[ℓ−22(2ℓ+1)C~ℓ−1XX+ℓ+32(2ℓ+1)C~ℓ+1XX].(12)\begin{aligned} \sum_{\ell_{1}}\mathcal{K}_{\ell_{1}\ell}^{33(44)}\widetilde{C}_{\ell_{1}}^{XX} ={}&\beta_{F,B(E)}^{2}\left[\frac{(\ell-2)(\ell-3)}{4(4\ell^{2}-1)}\widetilde{C}_{\ell-2}^{XX}+\frac{3(\ell^{2}+\ell-2)}{2(4\ell^{2}+4\ell-3)}\widetilde{C}_{\ell}^{XX}\right.\\ &\left.+\frac{(\ell+3)(\ell+4)}{4(4\ell^{2}+8\ell+3)}\widetilde{C}_{\ell+2}^{XX}\right]\\ &+\beta_{F,E(B)}^{2}\left[\frac{\ell-2}{2(2\ell+1)}\widetilde{C}_{\ell-1}^{XX}+\frac{\ell+3}{2(2\ell+1)}\widetilde{C}_{\ell+1}^{XX}\right]. \tag*{(12)} \end{aligned}

By inspecting eqs. (2)–(12), we can identify the effect of each class of operators. The CPT-odd parameters βAF,T2\beta_{AF,T}^{2} and βAF,S2\beta_{AF,S}^{2} source the well-known cosmic birefringence effect, inducing a mixing between the EEEE and BBBB spectra. In particular, βAF,T2\beta_{AF,T}^{2} produces isotropic birefringence, which also gives rise to non-vanishing EBEB and TBTB spectra (eqs. (5)–(6)), while βAF,S2\beta_{AF,S}^{2} generates anisotropic birefringence, coupling each multipole ℓ\ell to its neighbours ℓ±1\ell\pm1. The effect of the leakage of EE into BB modes is shown in the left panel of Figure 1 as two distinct features: one at ℓ≳300\ell\gtrsim300, accessible to ground-based BB-mode measurements at intermediate scales, and an additional excess at very large angular scales, present even in the absence of primordial tensor modes, which is potentially relevant for a future satellite experiment such as LiteBIRD. The CPT-even parameters βF,E2\beta_{F,E}^{2} and βF,B2\beta_{F,B}^{2} instead source a non-zero VVVV spectrum from linear polarization spectra, coupling multipoles between ℓ−2\ell-2 and ℓ+2\ell+2; notably, the VVVV spectrum is, in principle, the only one sensitive to βF,E2\beta_{F,E}^{2} and βF,B2\beta_{F,B}^{2} individually, since all the other spectra only depend on their sum (see section 2.1). The resulting VVVV spectrum is shown in the right panel of Figure 1, for values of the LV parameters corresponding to their 95% credible level upper limits from current temperature and linear-polarization data and from the LiteBIRD forecast. Finally, all the parameters rescale the TETE, EEEE and BBBB spectra through the common factor Z\mathcal{Z} defined in eq. (9). Through this rescaling, linear-polarization measurements also acquire sensitivity to the CPT-even parameters.

CMB B-mode and V-mode power spectra

Figure 1. Left: CMB BB-mode power spectrum for the ΛCDM\Lambda\mathrm{CDM} model (lensing only) and for models with Lorentz-violating parameters βAF,T2\beta_{AF,T}^{2}, βAF,S2\beta_{AF,S}^{2}, and βF2\beta_{F}^{2} each fixed to their 95% upper limit obtained from the SPA+BK18+SPTBB+SPTpolBB dataset combination (see § 3.1 for the definition of the dataset labels). Data points show BB-mode measurements from BK18 [42], SPT-3G D1 [46] and SPTpol [47]. SPTpol data points are restricted to ℓ>450\ell> 450 to avoid overlap with SPT-3G D1. Semi-transparent red diamonds show SPT-3G D1 data points in the multipole range covered by BK18; these are not used when BK18 is included in the analysis (see § 3.1 for more details). We also show the SO SATs and LiteBIRD noise curves, obtained by combining the frequency channels with inverse-noise weighting, as described in § 3.2. In particular, for SO SATs we adopt the baseline sensitivity and pessimistic 1/f1/f noise as described in [51]. Right: CMB VV-mode power spectrum for models with LV parameters fixed to their 95% upper limits obtained from current temperature and linear polarization data (SPA+BK18+SPTBB+SPTpolBB, black solid) and from the LiteBIRD TEB-only forecast (black dashed). Data points show the 95% CL upper limits on the VVVV spectrum from the combination of the CLASS 90, 150 and 220 GHz measurements used in this work [50], compared with previous limits from CLASS at 40 GHz [54] and from SPIDER at 150 GHz [55]. The noise curves in blue show the expected LiteBIRD sensitivity to VV modes for different values of the HWP phase shift, βHWP=1∘\beta_{\mathrm{HWP}} = 1^\circ, 5∘5^\circ, 10∘10^\circ, and 20∘20^\circ (see § 3.2 for more details).

The spectra presented in this section, and used throughout the rest of the analysis, are computed with CAMB_LV, a modified version of CAMB [56, 57] that extends the CAMB_CPT code used in Paper I.2

On the degeneracy between βF,E2\beta_{F,E}^{2} and βF,B2\beta_{F,B}^{2}

Although VV-mode polarization is, in principle, the only observable sensitive to βF,E2\beta_{F,E}^{2} and βF,B2\beta_{F,B}^{2} individually, we demonstrate that even for an ideal, cosmic-variance-limited experiment the VVVV power spectrum cannot separate them in practice, and only their sum is a well-constrained quantity.

This can be understood from the structure of eq. (7). Defining the two multipole-mixing kernels

A[X]ℓ≡(ℓ−2)(ℓ−3)4(4ℓ2−1)Xℓ−2+3(ℓ2+ℓ−2)2(4ℓ2+4ℓ−3)Xℓ+(ℓ+3)(ℓ+4)4(4ℓ2+8ℓ+3)Xℓ+2,(13)\mathcal{A}[X]_{\ell} \equiv\frac{(\ell-2)(\ell-3)}{4(4\ell^{2}-1)}X_{\ell-2}+\frac{3(\ell^{2}+\ell-2)}{2(4\ell^{2}+4\ell-3)}X_{\ell}+\frac{(\ell+3)(\ell+4)}{4(4\ell^{2}+8\ell+3)}X_{\ell+2}, \tag*{(13)}
B[X]ℓ≡ℓ−22(2ℓ+1)Xℓ−1+ℓ+32(2ℓ+1)Xℓ+1,(14)\mathcal{B}[X]_{\ell} \equiv\frac{\ell-2}{2(2\ell+1)}X_{\ell-1}+\frac{\ell+3}{2(2\ell+1)}X_{\ell+1}, \tag*{(14)}

eq. (7) can be recast, in terms of

βF2≡βF,E2+βF,B2andΔβF2≡βF,E2−βF,B2,(15)\beta_{F}^{2}\equiv\beta_{F,E}^{2}+\beta_{F,B}^{2}\quad\text{and}\quad\Delta\beta_{F}^{2}\equiv\beta_{F,E}^{2}-\beta_{F,B}^{2}, \tag*{(15)}

as

C~ℓVV=βF22(A+B)[C~ℓEE+C~ℓBB]+ΔβF22(B−A)[C~ℓEE−C~ℓBB],(16)\widetilde{C}_{\ell}^{VV}=\frac{\beta_{F}^{2}}{2}(\mathcal{A}+\mathcal{B})\left[\widetilde{C}_{\ell}^{EE}+\widetilde{C}_{\ell}^{BB}\right]+\frac{\Delta\beta_{F}^{2}}{2}(\mathcal{B}-\mathcal{A})\left[\widetilde{C}_{\ell}^{EE}-\widetilde{C}_{\ell}^{BB}\right], \tag*{(16)}

where A+B\mathcal{A}+\mathcal{B} acts on C~ℓEE+C~ℓBB\widetilde{C}_{\ell}^{EE}+\widetilde{C}_{\ell}^{BB}, and B−A\mathcal{B}-\mathcal{A} on C~ℓEE−C~ℓBB\widetilde{C}_{\ell}^{EE}-\widetilde{C}_{\ell}^{BB}. Since C~ℓBB≪C~ℓEE\widetilde{C}_{\ell}^{BB}\ll\widetilde{C}_{\ell}^{EE}, both terms in eq. (16) are dominated by essentially the same spectrum, C~ℓEE\widetilde{C}_{\ell}^{EE}. Moreover, A\mathcal{A} and B\mathcal{B} both mix each multipole with nearby ones, and on a smooth spectrum such as C~ℓEE\widetilde{C}_{\ell}^{EE} they give nearly the same result. The prefactor of ΔβF2\Delta\beta_{F}^{2}, B−A\mathcal{B}-\mathcal{A}, therefore nearly vanishes, while that of βF2\beta_{F}^{2}, A+B\mathcal{A}+\mathcal{B}, does not. The two operators differ appreciably only at the lowest multipoles, ℓ≲10\ell\lesssim10, which are also the scales where cosmic variance is largest. This explains why βF2\beta_{F}^{2} is well constrained while ΔβF2\Delta\beta_{F}^{2} is not.

To quantify this, we compare the two extreme cases at fixed βF2\beta_{F}^{2}, in which only one of the two parameters is non-zero, namely either βF2=βF,E2\beta_{F}^{2}=\beta_{F,E}^{2} or βF2=βF,B2\beta_{F}^{2}=\beta_{F,B}^{2}. For these two cases, we compute the cumulative signal-to-noise ratio

S/N(ℓmax⁡)=[∑ℓ=2ℓmax⁡(ΔCℓVVσℓ)2]1/2,(17)S/N(\ell_{\max})=\left[\sum_{\ell=2}^{\ell_{\max}}\left(\frac{\Delta C_{\ell}^{VV}}{\sigma_{\ell}}\right)^{2}\right]^{1/2}, \tag*{(17)}

where ΔCℓVV\Delta C_{\ell}^{VV} is the difference between the two VVVV spectra and σℓ=2/[(2ℓ+1)fsky] CℓVV\sigma_{\ell}=\sqrt{2/[(2\ell+1)f_{\mathrm{sky}}]}\,C_{\ell}^{VV} is the cosmic-variance uncertainty of the spectrum with βF,E2=βF,B2\beta_{F,E}^{2}=\beta_{F,B}^{2}. We show it in Figure 2 for fsky=1.0f_{\mathrm{sky}}=1.0. Since every VVVV spectrum at fixed composition is proportional to βF2\beta_{F}^{2}, βF2\beta_{F}^{2} cancels in this ratio and the result does not depend on the amplitude of the signal. The signal-to-noise ratio reaches only ≃0.3\simeq0.3 at ℓmax⁡≃10\ell_{\max} \simeq10 and does not increase further at higher multipoles. Even an ideal experiment would therefore be unable to distinguish the two extreme cases using the VVVV spectrum, which only captures the diagonal part of the VV-mode covariance. Additional information could in principle come from the correlations between different multipoles induced by the anisotropic components of kFk_F.

Figure 2. Cumulative signal-to-noise ratio, as a function of the maximum multipole ℓmax⁡\ell_{\max}, as defined in eq. (17). This is shown for a cosmic-variance limited experiment with fsky=1.0f_{\mathrm{sky}} = 1.0. The dotted line marks the 1σ1\sigma level.

For this reason, we retain only the term proportional to βF2\beta_F^2 in eq. (16), i.e. we set ΔβF2=0\Delta\beta_F^2 = 0, and report constraints on βF2\beta_F^2 throughout this work.

Mapping βAF,T2\beta_{AF,T}^2, βAF,S2\beta_{AF,S}^2 and βF2\beta_F^2 to the SME coefficients

In this section we provide the explicit mapping between the phenomenological β\beta parameters and the physical coefficients appearing in the action (1); we refer the reader to Paper I for the full derivation. The time and space components of kAFk_{AF}, usually recast as k(V)00(3)≡−4π(kˉAF)0k_{(V)00}^{(3)} \equiv-\sqrt{4\pi}(\bar{k}_{AF})^0 and ∣kAF∣|\mathbf{k}_{AF}|, are related to βAF,T2\beta_{AF,T}^2 and βAF,S2\beta_{AF,S}^2 by

∣k(V)00(3)∣=π2c(τ0−τLS)βAF,T2≃4.09×10−43βAF,T2 GeV,(18)\left|k_{(V)00}^{(3)}\right| = \frac{\sqrt{\pi}}{2c(\tau_0-\tau_{\mathrm{LS}})}\sqrt{\beta_{AF,T}^2} \simeq4.09 \times10^{-43}\sqrt{\beta_{AF,T}^2}\ \mathrm{GeV}, \tag*{(18)}
∣kAF∣≃2.00×10−43βAF,S2 GeV,(19)|\mathbf{k}_{AF}| \simeq2.00 \times10^{-43}\sqrt{\beta_{AF,S}^2}\ \mathrm{GeV}, \tag*{(19)}

where the bar denotes the line-of-sight average and c(τ0−τLS)≃13871 Mpcc(\tau_0-\tau_{\mathrm{LS}}) \simeq13871\ \mathrm{Mpc} is computed for the mean Λ\LambdaCDM parameters of the SPA+BK18+SPTBB+SPTpolBB combination (see § 3.1 for the definition of the dataset labels) with the LV parameters fixed to zero.3

relation for the CPT-even coefficient is instead frequency dependent4:

kF,E+B≡(2kˉF,E2+kˉF,B23)1/2≃4.38×10−29(νGHz)−1βF2(20)k_{F,E+B} \equiv\left(2\bar{k}_{F,E}^{2}+\frac{\bar{k}_{F,B}^{2}}{3}\right)^{1/2} \simeq4.38 \times10^{-29}\left(\frac{\nu}{\mathrm{GHz}}\right)^{-1}\sqrt{\beta_{F}^{2}} \tag*{(20)}

where kˉF,E\bar{k}_{F,E} and kˉF,B\bar{k}_{F,B} are specific line-of-sight-averaged contractions of (kF)αβμν(k_{F})^{\alpha\beta\mu\nu} (see Paper I for their explicit expressions), and ν\nu is replaced by an effective frequency νf\nu_f (see eq. (5.5) of Paper I) when combining datasets from different experiments, following the prescription of [58].

Methodology and datasets

We perform a Markov chain Monte Carlo (MCMC) analysis using Cobaya [59], jointly sampling the six Λ\LambdaCDM parameters5, the tensor-to-scalar ratio rr, and the Lorentz-violating parameters βAF,T2\beta_{AF,T}^{2}, βAF,S2\beta_{AF,S}^{2} and βF2\beta_{F}^{2}, for both current data and forecasts. We assume a spatially flat Universe with adiabatic initial conditions, one massive and two massless neutrino species with ∑mν=0.06 eV\sum m_{\nu}=0.06\ \mathrm{eV} [60, 61, 62], and effective number of relativistic species Neff=3.044N_{\mathrm{eff}}=3.044 [63, 64, 65, 66, 67, 68]. We adopt uniform priors βAF,T2,βAF,S2∈[0,0.01]\beta_{AF,T}^{2},\beta_{AF,S}^{2}\in[0,0.01] and βF2∈[0,0.1]\beta_{F}^{2}\in[0,0.1] on the LV parameters, and r∈[0,1]r\in[0,1] on the tensor-to-scalar ratio, while for the Λ\LambdaCDM parameters we follow [41]. Convergence of the chains is assessed via the Gelman-Rubin criterion [69], requiring R−1≲0.01R-1\lesssim0.01. The datasets and analysis methodology for current CMB experiments and the Simons Observatory and LiteBIRD forecasts are described below.

Current CMB data

We consider different dataset combinations, in order to assess the impact of different CMB observables on the constraints on Lorentz-violating parameters. The labels of the individual datasets are summarized in Table 1.

LabelData
P18Planck 2018 TTTT, TETE, EEEE [41, 70, 71, 72]; Planck PR4 lensing [73]
SPAPlanck 2018, ACT DR6 [44] and SPT-3G D1 [45, 74] TTTT, TETE, EEEE; Planck+ACT DR6 [73, 75, 76, 77] and SPT-3G D1 [78] lensing
BK18BICEP/Keck 2018 BBBB [42]
SPTBBSPT-3G D1 BBBB [46, 74, 79]
SPTpolBBSPTpol BBBB [47, 74, 79]
P20EBPlanck PR4 EEEE, BBBB, EBEB, replacing SRoll2 [48, 49]
CLASSCLASS 90/150/220 GHz combined VVVV [50]

Table 1. Labels of the individual datasets used in this work. Dataset combinations with the corresponding multipole cuts are described in § 3.1.

Temperature and linear polarization. As a reference point for comparison, we repeat the analysis of Paper I, with minor changes:6

  • P18+BK18: for Planck, we use large-scale temperature and EE-mode polarization data from the Commander [41, 70] and SRoll2 [71, 72] likelihoods, respectively, complemented at high multipoles by the Plik_lite likelihood [41]. We also include lensing reconstruction measurements from Planck PR4 [73]. For BB modes, we use data from BICEP/Keck 2018 (BK18) [42].

We then consider two additional combinations, obtained by adding new temperature and linear-polarization data from the ACT and SPT experiments, with the corresponding multipole cuts summarized below:

  • SPA+BK18: we combine Planck, ACT DR6 and SPT-3G D1 temperature, polarization and lensing data, which we collectively refer to as SPA, using the ACT-lite [44] and SPT-lite [45, 74] likelihoods. Following the strategy adopted in [44, 45], we restrict Planck to multipoles ℓ≤1000\ell\le1000 in temperature and ℓ≤600\ell\le600 in polarization to minimize the correlations with ACT, and we neglect correlations between SPT and the other datasets. We further include the joint Planck+ACT DR6 lensing likelihood [73, 75, 76, 77] and the MUSE lensing likelihood for SPT-3G D1 [78]. This combination allows us to assess the impact of improved small-scale measurements on the constraints;

  • SPA+BK18+SPTBBBB+SPTpolBBBB (baseline): we further add BB-mode measurements from SPT-3G D1 [46], restricted to 350<ℓ<450350 < \ell< 450, and from SPTpol [47], restricted to ℓ>450\ell> 450, to avoid overlap between the datasets (see left panel of Figure 1). For both we use the candl implementation [74] of the lite likelihoods, introduced in [79]. This combination allows us to assess the impact of additional BB-mode data.

_EB spectrum._ We include the parity-violating EB spectrum using the Lollipop likelihood [48, 49], based on the Planck PR4 data release, published in 2020 (hence the label P20). Since Lollipop provides the low-ℓ\ell EE, BB and EB spectra jointly, whenever we use the EB spectrum we consistently include the corresponding Lollipop EE and BB spectra as well. In these combinations, Lollipop therefore replaces the SRoll2 low-ℓ\ell E-mode likelihood. To maximize the information coming from these low-ℓ\ell polarization spectra while avoiding any double counting, we use Lollipop up to ℓ=150\ell= 150 and correspondingly raise the minimum multipole of the high-ℓ\ell Plik_lite EE spectrum to ℓmin⁡=151\ell_{\min} = 151. We further neglect the correlations between Planck and BK18 B-mode data, as done in [80]. Through eq. (5), the EB spectrum provides direct sensitivity to isotropic birefringence, i.e. to βAF,T2\beta^{2}_{AF,T}, complementing the indirect constraint coming from the E-to-B leakage in the BB spectrum.7 We include the EB spectrum in two dataset combinations:

  • P18+P20EBEB+BK18: the P18+BK18 combination together with the Lollipop spectra described above;

  • SPA+P20EBEB+BK18+SPTBBBB+SPTpolBBBB: the baseline combination together with the Lollipop spectra described above.

These combinations allow us to assess the constraining power of the direct parity-violating observable.

Circular polarization. In Paper I we showed that the VV-mode measurements from two years of CLASS observations at 40 GHz [81] and from SPIDER [55] do not improve the constraint on the CPT-even parameter βFV\beta^{V}_{F}. Here we use the latest CLASS results [50], which extend the measurements to 90, 150 and 220 GHz. We use the minimum-variance combination of the three frequency bands, in bins of Δℓ=20\Delta\ell= 20 over 30≤ℓ≤69030 \le\ell\le690.8 The limits from CLASS at 40 GHz, based on observations up to 2022 [54], and from SPIDER are about, respectively, one and three-to-four orders of magnitude weaker (see right panel of Figure 1), and we do not include them. Following Paper I, we build a simple Gaussian likelihood

−2ln⁡L=∑b(Db,thVV−Db,dataVV)2σb2,(21)-2\ln\mathcal{L}=\sum_b\frac{\left(\mathcal{D}^{VV}_{b,\mathrm{th}}-\mathcal{D}^{VV}_{b,\mathrm{data}}\right)^2}{\sigma_b^2}, \tag*{(21)}

where Db,thVV\mathcal{D}^{VV}_{b,\mathrm{th}} and Db,dataVV\mathcal{D}^{VV}_{b,\mathrm{data}} are the binned theoretical and measured VVVV bandpowers and σb\sigma_b the associated uncertainty. We include the VV-mode spectrum in one dataset combination:

  • SPA+BK18+SPTBB+SPTpolBB+CLASS: the baseline combination together with CLASS data. This combination allows us to assess the impact of circular polarization on the constraints.

Forecasts for future CMB experiments

We forecast the sensitivity of the Simons Observatory (SO) and the LiteBIRD satellite to LV parameters. For both experiments, we employ the Wishart likelihood [82, 83] as implemented in Cobaya in [84].9 which we extend to include VV modes. For Gaussian CMB fields it is exact on the full sky; partial sky coverage is approximated by scaling the effective number of degrees of freedom per multipole by fskyf_{\mathrm{sky}}. The likelihood reads

−2ln⁡L=∑ℓ(2ℓ+1)fsky[tr⁡(Cˉℓ−1C^ℓ)+ln⁡∣Cˉℓ∣∣C^ℓ∣−n],(22)-2\ln\mathcal{L}=\sum_{\ell}(2\ell+1)f_{\mathrm{sky}}\left[\operatorname{tr}\left(\bar{C}_{\ell}^{-1}\hat{C}_{\ell}\right)+\ln\frac{\left|\bar{C}_{\ell}\right|}{\left|\hat{C}_{\ell}\right|}-n\right], \tag*{(22)}

where Cˉℓ\bar{C}_{\ell} and C^ℓ\hat{C}_{\ell} are, respectively, the theoretical (signal plus noise) and observed covariance matrices of the nn fields, and ∣⋅∣|\cdot| denotes their determinant. Our analysis includes n=4n=4 fields, {T,E,B,V}\{T,E,B,V\}. We do not include the EBEB and TBTB cross-spectra, so that the covariance matrix is block-diagonal, with the TTTT, TETE and EEEE spectra forming one block and the BBBB and VVVV spectra entering independently, each over its own multipole range. Our forecasts on βAF,T\beta_{\mathrm{AF},T} are therefore directly comparable with the current-data combinations that do not include the EBEB spectrum.

When noise curves are not publicly available, they are modelled as white noise with a Gaussian beam (see e.g. [82, 83, 85]):

NℓXX=σX2exp⁡[ℓ(ℓ+1)θFWHM28ln⁡2](23)N_{\ell}^{XX} = \sigma_X^2 \exp\left[\frac{\ell(\ell+1)\theta_{\mathrm{FWHM}}^2}{8\ln2}\right] \tag*{(23)}

with X∈{T,E,B,V}X \in\{T, E, B, V\}, where σX\sigma_X denotes the instrumental sensitivity and θFWHM\theta_{\mathrm{FWHM}} the full-width half-maximum beam size. We assume that temperature and polarization instrumental noises are uncorrelated, so that NℓTE=0N_{\ell}^{TE} = 0. We adopt the following noise configurations:

  • Planck+SO: we model the noise properties of SO following the specifications provided by the collaboration [51, 86]. We include both the Small Aperture Telescopes (SATs), for BB-mode polarization, and the Large Aperture Telescope (LAT), for temperature and EE-mode polarization. For Planck, we construct the noise curves via eq. (23), assuming a single effective channel with 40 μK40\,\mu\mathrm{K}-arcmin in temperature and 80 μK80\,\mu\mathrm{K}-arcmin in EE-mode polarization, and a beam size θFWHM=7\theta_{\mathrm{FWHM}} = 7 arcmin, as in [53].10 The two experiments are combined as follows (see [53, 87]). At 2≤ℓ≤502 \le\ell\le50, over fsky=0.7f_{\mathrm{sky}} = 0.7, we use Planck alone. At higher multipoles, in the sky region observed by SO LAT (fsky=0.4f_{\mathrm{sky}} = 0.4) we adopt the combined SO LAT+Planck foreground-marginalized noise curves provided by the SO collaboration for the baseline noise level,11 used for 51≤ℓ≤300051 \le\ell\le3000. To account for the sky area not covered by SO LAT, we include an additional Planck-only contribution with fsky=0.3f_{\mathrm{sky}} = 0.3, used at 51≤ℓ≤250051 \le\ell\le2500. For BB-mode polarization, we use the SATs over 30≤ℓ≤30030 \le\ell\le300, with fsky=0.1f_{\mathrm{sky}} = 0.1. Since no component-separated noise curves are publicly available for the SATs, we compute the noise power spectrum of each frequency channel with the SO noise model, for baseline sensitivity and pessimistic 1/f1/f noise (see table 2 of [51]), and combine the channels with inverse-noise weighting. This corresponds to the ideal case without foreground residuals. We do not apply any delensing to the BB-mode spectrum. A partial removal of the lensing BB-mode contribution would enhance the constraining power on the tensor-to-scalar ratio rr and, potentially, on the LV parameters as well [51, 53].

  • LiteBIRD: we adopt the instrumental configuration of [52].12 We consider the frequency channels between 78 and 195 GHz, with sensitivities and beam widths taken from table 13 of [52], and assume σT=σE/B/2\sigma_T = \sigma_{E/B}/\sqrt{2}. The noise power spectra of the individual channels, obtained via eq. (23), are then coadded to obtain the noise curves used in the analysis. We do not include foregrounds in the LiteBIRD forecasts, which therefore correspond to the ideal case without foreground residuals. We adopt fsky=0.7f_{\mathrm{sky}} = 0.7 and include temperature and EE-mode polarization for 2≤ℓ≤6002 \le\ell\le600, as well as BB-mode

polarization for 2≤ℓ≤1502 \le\ell\le150 (see e.g. [52, 89]). The BB-mode noise curve is shown in the left panel of Figure 1, together with that of the SO SATs.

In addition to temperature and linear polarization, we explore the sensitivity of LiteBIRD to VV-mode polarization assuming the use of a rotating half-wave plate (HWP) as a polarization modulator.13 While an ideal HWP induces a phase shift of exactly π\pi between the two orthogonal components of the incident electromagnetic wave, yielding no sensitivity to circular polarization, realistic non-idealities parametrized by a departure βHWP\beta_{\mathrm{HWP}} from this ideal value provide sensitivity to VV-mode signals [53, 91]. In this work we neglect other effects of HWP non-ideality which can impact the VV-mode signal, such as differential optical load and cross-polarization [91]. The noise levels for linear and circular polarization are related to the total intensity noise σT\sigma_{\mathrm{T}}, as a function of βHWP\beta_{\mathrm{HWP}}, by (see [53])

σE/BσT=2cos⁡2(βHWP/2),σVσT=2sin⁡βHWP,(24)\frac{\sigma_{E/B}}{\sigma_{\mathrm{T}}} = \frac{\sqrt{2}}{\cos^{2}\left(\beta_{\mathrm{HWP}}/2\right)}, \qquad\frac{\sigma_{V}}{\sigma_{\mathrm{T}}} = \frac{\sqrt{2}}{\sin\beta_{\mathrm{HWP}}}, \tag*{(24)}

from which the noise power spectra follow via eq. (23). Note that, while a non-zero βHWP\beta_{\mathrm{HWP}} in principle also degrades the sensitivity to linear polarization, this effect is small for values of βHWP\beta_{\mathrm{HWP}} of the order of a few to a few tens of degrees [92, 93, 94] (see also the discussion in § 4.3). In our analysis we consider four representative values, βHWP=1∘\beta_{\mathrm{HWP}} = 1^\circ, 5∘5^\circ, 10∘10^\circ, and 20∘20^\circ, and include VV-mode polarization for 2≤ℓ≤6002 \le\ell\le600. The VV-mode noise curves, for the different values of βHWP\beta_{\mathrm{HWP}}, are shown in the right panel of Figure 1.

As our fiducial cosmology, we adopt the mean values of the Λ\LambdaCDM parameters obtained from our analysis with SPA+BK18+SPTBB+SPTpolBB, setting the tensor-to-scalar ratio and all the LV parameters to zero.

It is worth stressing that our forecasts are simplified and are meant to illustrate a best-case scenario. Apart from the component separation included in the SO LAT noise curves, we neglect foreground residuals and instrumental systematics. They are therefore not intended to represent the realistic sensitivity of SO and LiteBIRD.

Results

In this section we present our results on the LV parameters βAF,T2\beta^{2}_{\mathrm{AF,T}}, βAF,S2\beta^{2}_{\mathrm{AF,S}} and βF2\beta^{2}_{\mathrm{F}}. We report 95% credible level (CL) upper limits unless otherwise stated. § 4.1 reports the constraints from current data (Table 2), while § 4.2 and § 4.3 present the forecasts for SO and LiteBIRD, respectively (Table 3). Finally, in § 4.4 we recast these results in terms of the SME coefficients.

Datasetr×102r \times10^{2}βAF,T2×104\beta^{2}_{AF,T} \times10^{4}βAF,S2×104\beta^{2}_{AF,S} \times10^{4}βF2×102\beta^{2}_{F} \times10^{2}
P18+BK18<3.13< 3.13<6.77< 6.77<6.69< 6.69<4.39< 4.39
SPA+BK18<3.12< 3.12<6.36< 6.36<6.39< 6.39<4.70< 4.70
SPA+BK18+SPTBB+SPTpolBB<3.19< 3.19<3.59< 3.59<3.65< 3.65<4.72< 4.72
P18+P20EB+BK18<2.80< 2.80<7.92< 7.92<6.45< 6.45<4.80< 4.80
SPA+P20EB+BK18+SPTBB+SPTpolBB<2.96< 2.96<4.33< 4.33<3.46< 3.46<5.16< 5.16
SPA+BK18+SPTBB+SPTpolBB+CLASS<3.17< 3.17<3.66< 3.66<3.82< 3.82<4.13< 4.13

Table 2. Table 2. 95% CL upper limits on rr (in units of 10−210^{-2}), βAF,T2\beta^{2}_{AF,T}, βAF,S2\beta^{2}_{AF,S} (in units of 10−410^{-4}) and βF2\beta^{2}_{F} (in units of 10−210^{-2}) for the current dataset combinations discussed in § 4.1.

DatasetβAF,T2×104\beta^{2}_{AF,T} \times10^{4}βAF,S2×104\beta^{2}_{AF,S} \times10^{4}βF2×102\beta^{2}_{F} \times10^{2}
Planck+SO<2.48< 2.48<2.41< 2.41<1.19< 1.19
LiteBIRD TEB only<1.05< 1.05<1.38< 1.38<3.20< 3.20
LiteBIRD TEBV (βHWP=1∘\beta_{\mathrm{HWP}} = 1^\circ)<1.08< 1.08<1.42< 1.42<2.98< 2.98
LiteBIRD TEBV (βHWP=5∘\beta_{\mathrm{HWP}} = 5^\circ)<1.13< 1.13<1.50< 1.50<0.380< 0.380
LiteBIRD TEBV (βHWP=10∘\beta_{\mathrm{HWP}} = 10^\circ)<1.12< 1.12<1.50< 1.50<0.0954< 0.0954
LiteBIRD TEBV (βHWP=20∘\beta_{\mathrm{HWP}} = 20^\circ)<1.16< 1.16<1.48< 1.48<0.0246< 0.0246
SPA+BK18+SPTBB+SPTpolBB<3.59< 3.59<3.65< 3.65<4.72< 4.72

Table 3. 95% CL upper limits on βAF,T2\beta^{2}_{AF,T}, βAF,S2\beta^{2}_{AF,S} (in units of 10−410^{-4}) and βF2\beta^{2}_{F} (in units of 10−210^{-2}) for the Planck+SO and LiteBIRD forecasts. For the latter, we report results without (TEB) and with (TEBV) sensitivity to circular polarization at different values of the HWP phase shift βHWP\beta_{\mathrm{HWP}}. For comparison we also include the bounds obtained with current observations, SPA+BK18+SPTBB+SPTpolBB.

Constraints from current data

In this section, we discuss the impact of progressively adding the new datasets described in § 3.1 to Planck and BK18 data. We first include the small-scale temperature and EE-mode measurements from ACT and SPT, then the SPT-3G D1 and SPTpol BB-mode data, which yield our baseline combination, and the Planck EBEB spectrum. Finally, we discuss the impact of adding the CLASS VV modes.

Small-scale temperature and EE-mode polarization data. As a first step in our analysis, we test the impact of adding the small-scale temperature and EE-mode data from ACT and SPT. The 95% CL bounds on all the LV parameters are only marginally affected, changing by less than ∼7%\sim7\% with respect to P18+BK18 (Table 2 and left panel of Figure 3). For βAF,T2\beta^{2}_{AF,T} and βAF,S2\beta^{2}_{AF,S} this is expected, since these parameters are mainly constrained through the leakage of EE into BB modes probed by BK18, which is included in both combinations, while the additional EE-mode data are only sensitive to these parameters through the common rescaling factor Z\mathcal{Z} (eq. (9)). For βF2\beta^{2}_{F}, which in the absence of VV-mode data enters the spectra only through Z\mathcal{Z}, the new data could instead be expected to improve the constraint. This does not happen because the scale-independent rescaling by Z\mathcal{Z} is partially degenerate with the polarization calibration parameters sampled in the ACT and SPT likelihoods, namely the ACT polarization efficiency, pACTp_{\mathrm{ACT}}, and the SPT EE-mode calibration, ϵcalSPT\epsilon^{\mathrm{SPT}}_{\mathrm{cal}}, which therefore absorb most of the potential information on βF2\beta^{2}_{F}. In the baseline combination (SPA+BK18+SPTBB+SPTpolBB), βF2\beta^{2}_{F} has correlation coefficients of −0.58-0.58 and −0.38-0.38 with these two parameters, respectively, while the CPT-odd parameters are essentially uncorrelated with them, since their imprint on the BBBB spectrum cannot be mimicked by a calibration of the polarization spectra (see Figure 8 in appendix § 8). Fixing both calibration parameters to unity reduces the width of the βF2\beta^{2}_{F} posterior by about 15%. We perform this test on the baseline combination only, since the BB-mode data have a negligible impact on the constraint on βF2\beta^{2}_{F} (see appendices § 7 and § 8).

Figure 3. One- and two-dimensional posterior distributions of rr, βAF,T2\beta^{2}_{AF,T}, βAF,S2\beta^{2}_{AF,S}, and βF2\beta^{2}_{F} for different dataset combinations, compared with the P18+BK18 combination (black dashed line) in both panels. Left: impact of adding small-scale temperature and polarization data from ACT and SPT (green), and of adding SPT-3G D1 and SPTpol BB-mode data on top of it (red), and of further adding the CLASS VV-mode data (dark blue solid line). Right: impact of including the Lollipop EBEB spectrum on top of P18+BK18 (purple) and on top of our baseline dataset combination (orange).

Figure 8. Correlation coefficients between the LV parameters and the polarization calibration parameters of ACT and SPT, pACTp_{\mathrm{ACT}} and EcalSPTE_{\mathrm{cal}}^{\mathrm{SPT}}, for the baseline combination (left) and for the SPA combination without any BB-mode data (right).

BB-mode polarization data. Next, we add BB-mode data from SPT-3G D1 and SPTpol. The left panel of Figure 3 shows that these datasets improve the bounds on βAF,T2\beta^{2}_{AF,T} and βAF,S2\beta^{2}_{AF,S} by nearly a factor of 2 with respect to SPA+BK18, while leaving the constraint on βF2\beta^{2}_{F} unchanged. As we show in appendix § 7, this improvement is driven by SPT-3G D1, while SPTpol adds no constraining power and in fact slightly loosens the bounds. Moreover, removing BK18 from the baseline combination weakens the bounds on βAF,T2\beta^{2}_{AF,T} and βAF,S2\beta^{2}_{AF,S} by only ∼15%\sim15\%, indicating that current constraints on these parameters rely primarily on the intermediate-scale BB-mode measurements of SPT-3G D1, which probe the EE-to-BB leakage feature at ℓ≃300\ell\simeq300 (see left panel of Figure 1), rather than on the BB-mode measurements of BK18 at larger angular scales.

EE- and BB-mode polarization cross-correlation data. We then assess the constraining power of the EBEB spectrum, which we take from the Planck PR4 Lollipop likelihood. Unlike the observables considered so far, which probe isotropic birefringence only indirectly through the leakage of EE into BB modes, the EBEB spectrum is a direct tracer of this effect (see eq. (5)). As a validation of our implementation, we verify that the P18+P20EBEB+BK18 combination with the LV parameters fixed to zero yields r<0.032r < 0.032 at 95% CL, in agreement with [49].

The right panel of Figure 3 shows that adding the EBEB spectrum pulls the posterior of βAF,T2\beta^{2}_{AF,T} about 1σ1\sigma away from zero. At 68% CL, we find βAF,T2=3.4−3.2+1.0×10−4\beta^{2}_{AF,T} = 3.4^{+1.0}_{-3.2} \times10^{-4} for P18+P20EBEB+BK18 and βAF,T2=1.9−1.8+0.5×10−4\beta^{2}_{AF,T} = 1.9^{+0.5}_{-1.8} \times10^{-4} for SPA+P20EBEB+BK18+SPTBB+SPTpolBBBB, while we report the 95% CL upper limits in Table 2. Moreover, the bound on rr tightens with respect to the case with the LV parameters fixed to zero, since βAF,T2\beta^{2}_{AF,T} and rr are partially anticorrelated, both adding power to the BBBB spectrum. On the other hand, the bound on βF2\beta^{2}_{F} is slightly weakened. This is not due to the EBEB spectrum itself, but mostly to the replacement of the SRoll2 low-ℓ\ell EE-mode likelihood with Lollipop, which changes the constraint on the amplitude of the EEEE spectrum, and hence on Z\mathcal{Z}. Repeating the P18+P20EBEB+BK18 analysis with only the Lollipop EEEE spectrum, i.e. without the EBEB and BBBB spectra, we find a bound on βF2\beta^{2}_{F} that already accounts for most of this difference.

Our results for βAF,T2\beta^{2}_{AF,T} can be compared with existing measurements of the isotropic birefringence angle. A rotation of the CMB linear polarization plane by an isotropic angle α\alpha sources an EBEB spectrum

CℓEB=12sin⁡(4α)(C~ℓEE−C~ℓBB),(25)C_{\ell}^{EB} = \frac{1}{2}\sin(4\alpha)\left(\widetilde{C}_{\ell}^{EE} - \widetilde{C}_{\ell}^{BB}\right), \tag*{(25)}

where C~ℓEE\widetilde{C}_{\ell}^{EE} and C~ℓBB\widetilde{C}_{\ell}^{BB} are the spectra in the absence of rotation [95]. This holds regardless of the physical origin of the rotation. Our eq. (5) takes this form with βAF,T2=12sin⁡(4α)\sqrt{\beta^{2}_{AF,T}} = \frac{1}{2}\sin(4\alpha). Since βAF,T2\beta^{2}_{AF,T} is non-negative, our model only describes rotations with α≥0\alpha\ge0. A natural benchmark is the value α=0.36∘±0.11∘\alpha= 0.36^{\circ} \pm0.11^{\circ} measured in [11] from Planck PR4 data, the same release on which our EBEB likelihood is based, but with a substantially different methodology. We sample the Λ\LambdaCDM parameters jointly with all the LV phenomenological parameters, using the full information contained in the CMB spectra. However, we do not model any residual miscalibration of the detector polarization angles, and thus attribute the entire EBEB signal to a physical rotation. In [11], instead, the cosmological parameters are kept fixed to the Planck 2018 Λ\LambdaCDM best fit, and α\alpha is estimated from the EBEB cross-spectra of the Planck HFI frequency maps jointly with the miscalibration angle of each detector set. The degeneracy between them is broken by exploiting the polarized Galactic foreground emission, which is rotated by a miscalibration but not by cosmic birefringence [96, 97]. Galactic dust, however, may carry an intrinsic EBEB correlation, which would be absorbed into the miscalibration angles and bias α\alpha low. The value quoted above includes a correction for this effect, modelled from the TETE and TBTB spectra of dust measured at 353 GHz [98].

Figure 4 shows the posterior distributions of α\alpha from the P18+P20EBEB+BK18 and SPA+P20EBEB+BK18+SPTBBBB+SPTpolBBBB combinations, compared with the measurement of [11]. Since βAF,T2=12sin⁡(4α)\sqrt{\beta^{2}_{AF,T}} = \frac{1}{2}\sin(4\alpha), a uniform prior on βAF,T2\beta^{2}_{AF,T} corresponds to a prior on α\alpha proportional to sin⁡(8α)\sin(8\alpha), which suppresses small rotation angles. For a consistent comparison, we therefore reweight each sample of our chains by 1/sin⁡(8α)1/\sin(8\alpha), which is equivalent to adopting a uniform prior on α\alpha. The two posteriors peak at about 0.39∘0.39^\circ and 0.27∘0.27^\circ, respectively, within the 1σ1\sigma region of this measurement. This comparison therefore shows that the mild preference for a non-zero isotropic rotation found in Planck PR4 data persists when the cosmological parameters are varied jointly with all the LV parameters. This result should nonetheless be interpreted with caution, since our constraint is sensitive to the total EBEB signal, including any residual miscalibration of the polarization angles and any residual foreground EBEB correlation. Disentangling a cosmological rotation from these contributions will require a joint modelling of Lorentz violation and polarization angle miscalibration.

Figure 4. One-dimensional posterior distributions of the isotropic birefringence angle α\alpha, derived from βAF,T2\beta^{2}_{AF,T}, for the P18+P20EBEB+BK18 (purple) and SPA+P20EBEB+BK18+SPTBBBB+SPTpolBBBB (orange) dataset combinations, reweighted to a uniform prior on α\alpha. The black line and the grey band show the central value and the ±1σ\pm1\sigma region of α=0.36∘±0.11∘\alpha= 0.36^\circ\pm0.11^\circ, measured from Planck PR4 data in [11], which includes a correction for the intrinsic EBEB correlation of Galactic dust.

Circular polarization data. Finally, we include the CLASS VV-mode measurements. As discussed in § 2.1, these data are only sensitive to the sum βF2\beta_{\mathrm{F}}^2, which in the previous combinations is constrained only through the rescaling factor Z\mathcal{Z} (eq. (9)). Adding them to the baseline combination leaves the bounds on βAF,T2\beta_{\mathrm{AF,T}}^2 and βAF,S2\beta_{\mathrm{AF,S}}^2 essentially unchanged, while tightening the bound on βF2\beta_{\mathrm{F}}^2 by ∼13%\sim13\% (Table 2 and left panel of Figure 3). This improvement arises because, at intermediate angular scales, the CLASS upper limits now lie close to the largest VVVV spectrum compatible with temperature and linear-polarization data (right panel of Figure 1), so that VV-mode data begin to provide constraining power on βF2\beta_{\mathrm{F}}^2. Circular polarization also probes this parameter in a cleaner way. While the rescaling by Z\mathcal{Z} is partially degenerate with the polarization calibration parameters of ACT and SPT, the VVVV spectrum sourced by the CPT-even operator cannot be mimicked by a miscalibration of the linear polarization. Including CLASS data therefore reduces the anticorrelation of βF2\beta_{\mathrm{F}}^2 with these calibration parameters (see appendix § 8).

Simons Observatory: impact of small-scale temperature and EE-mode data

We first discuss the forecast sensitivity of Planck+SO to the LV parameters, following the methodology described in § 3.2. As shown in Table 3 and Figure 5, the largest improvement with respect to current data is on βF2\beta_{\mathrm{F}}^2, whose bound tightens by a factor of ∼4\sim4. This is expected, since in the absence of VV-mode sensitivity the constraint on βF2\beta_{\mathrm{F}}^2 comes entirely from the rescaling of the TETE and EEEE spectra through Z\mathcal{Z} (eq. (9)), and therefore benefits directly from the high-precision small-scale temperature and EE-mode measurements of SO LAT. Nevertheless, we note that our forecasts do not include polarization calibration uncertainties, which, as found for current data, are partially degenerate with βF2\beta_{\mathrm{F}}^2 (see appendix § 8), so that the forecast improvement on this parameter may be somewhat optimistic. Comparing the widths of the posterior distributions, the improvement with respect to current data analysed with the ACT and SPT calibration parameters fixed is only slightly smaller, roughly a factor of 3.5.

Figure 5. One- and two-dimensional posterior distributions of βAF,T2\beta_{AF,T}^{2}, βAF,S2\beta_{AF,S}^{2} and βF2\beta_{F}^{2} forecast for Planck+SO (brown), LiteBIRD TEB only (yellow) and LiteBIRD TEBV with βHWP=10∘\beta_{\mathrm{HWP}} = 10^\circ (blue), compared with the constraints from current data, SPA+BK18+SPTBB+SPTpolBB (black dashed). The forecasts assume a fiducial cosmology with r=0r = 0 and all LV parameters set to zero (see § 3.2).

The bounds on βAF,T2\beta_{\mathrm{AF,T}}^2 and βAF,S2\beta_{\mathrm{AF,S}}^2, instead, improve by a factor of ∼1.5\sim1.5 with respect to current data. This improvement comes from the higher sensitivity of the SO SATs with respect to BK18 over the same range of angular scales, 30≤ℓ≤30030 \le\ell\le300. These data, however, do not cover the EE-to-BB leakage feature at ℓ≳300\ell\gtrsim300, where most of the information on these two parameters is contained (see § 4.1 and appendix § 7). BB-mode measurements at smaller angular scales than those probed by the SATs, e.g. with SO LAT [51], would therefore further tighten the bounds on βAF,T2\beta_{\mathrm{AF,T}}^2 and βAF,S2\beta_{\mathrm{AF,S}}^2. Although rr is always varied in the analysis, we do not show it in Table 3 and Figure 5, since its constraint is driven by the standard BB-mode sensitivity of the experiment. For reference, when the LV parameters are varied, we find r<3.5×10−3r < 3.5 \times10^{-3} at 95% CL, about an order of magnitude tighter than with current data (see Table 2).14

LiteBIRD: the role of future VV-mode measurements

We now turn to the forecast sensitivity of the LiteBIRD satellite, first considering a TEBTEB-only configuration, i.e. without any sensitivity to circular polarization. As shown in Table 3 and Figure 5, LiteBIRD improves the bounds on βAF,T2\beta_{\mathrm{AF,T}}^2 and βAF,S2\beta_{\mathrm{AF,S}}^2 by a factor of ∼3.4\sim3.4 and ∼2.6\sim2.6 with respect to current data, and by a factor of ∼2.4\sim2.4 and ∼1.7\sim1.7 with respect to

Planck+SO. This improvement is driven by the high-sensitivity B-mode measurements of LiteBIRD at the largest angular scales, which are not accessible to SO and where the Planck noise is too high to be informative. There, the EE-to-BB leakage produces an excess in the BBBB spectrum even for r=0r = 0 (see left panel of Figure 1). Moreover, at these scales the leakage sourced by βAF,S2\beta^{2}_{AF,S} is weaker than that sourced by βAF,T2\beta^{2}_{AF,T}. In eq. (11), βAF,T2\beta^{2}_{AF,T} multiplies C~ℓEE\widetilde{C}^{EE}_{\ell} with a unit coefficient, while βAF,S2\beta^{2}_{AF,S} multiplies C~ℓ−1EE\widetilde{C}^{EE}_{\ell-1} and C~ℓ+1EE\widetilde{C}^{EE}_{\ell+1} with weights (ℓ2−4)/[ℓ(2ℓ+1)](\ell^{2} - 4)/[\ell(2\ell+ 1)] and (ℓ−1)(ℓ+3)/[(ℓ+1)(2ℓ+1)](\ell- 1)(\ell+ 3)/[(\ell+ 1)(2\ell+ 1)], whose sum approaches unity only at high multipoles and drops to 1/31/3 at ℓ=2\ell= 2. This explains the larger improvement on βAF,T2\beta^{2}_{AF,T}. The bound on βF2\beta^{2}_{F}, instead, improves only by a factor of ∼1.5\sim1.5 with respect to current data, and is nearly 3 times weaker than the Planck+SO one. As for current data, without VV-mode sensitivity the constraint on βF2\beta^{2}_{F} comes entirely from the rescaling of the TETE and EEEE spectra through Z\mathcal{Z}, to which LiteBIRD, limited to ℓ≤600\ell\le600, is less sensitive than SO. In the TEB-only case, the two experiments are therefore complementary, with SO dominating the constraint on the CPT-even parameter and LiteBIRD those on the CPT-odd ones.

Including VV-mode sensitivity through a non-ideal HWP changes this picture for βF2\beta^{2}_{F}. As βHWP\beta_{\mathrm{HWP}} increases from 1∘1^\circ to 20∘20^\circ, the bound on βF2\beta^{2}_{F} tightens by two orders of magnitude, eventually becoming comparable to those on βAF,T2\beta^{2}_{AF,T} and βAF,S2\beta^{2}_{AF,S}. This trend can be understood from the right panel of Figure 1: larger values of βHWP\beta_{\mathrm{HWP}} lower the VV-mode noise, improving the signal-to-noise ratio of the predicted VVVV spectrum and hence the sensitivity to βF2\beta^{2}_{F}. For βHWP=1∘\beta_{\mathrm{HWP}} = 1^\circ, VV-mode data tighten the bound by only ∼7%\sim7\% with respect to the TEB-only case, a contribution similar to that of current CLASS data (see § 4.1). Already for βHWP=5∘\beta_{\mathrm{HWP}} = 5^\circ, however, the LiteBIRD bound on βF2\beta^{2}_{F} is about an order of magnitude tighter than the current one including CLASS, and a factor of ∼3\sim3 tighter than the Planck+SO one. The bounds on βAF,T2\beta^{2}_{AF,T} and βAF,S2\beta^{2}_{AF,S}, on the other hand, remain stable within ∼10%\sim10\% across all values of βHWP\beta_{\mathrm{HWP}} considered, as do the uncertainties on the standard Λ\LambdaCDM+rr parameters, with a 95% CL upper limit r≲4×10−4r \lesssim4 \times10^{-4} in all cases when the LV parameters are varied.15 A non-ideal HWP therefore improves the constraint on βF2\beta_{F}^{2} without degrading the sensitivity to the other parameters, as already noted in [53].

Implications for the minimal SME coefficients

In this section, we recast the constraints on the phenomenological parameters βAF,T2\beta_{AF,T}^{2}, βAF,S2\beta_{AF,S}^{2} and βF2\beta_{F}^{2} presented above in terms of the fundamental coefficients appearing in the SME action, eq. (1), using the mapping described in section 2.2. The results are reported in Table 4. For the baseline combination, SPA+BK18+SPTBB+SPTpolBB, we find at 95% CL:

Dataset∣k(V)00(3)∣×1045\lvert k^{(3)}_{(V)00}\rvert\times10^{45} (GeV)∣kAF∣×1045\lvert\mathbf{k}_{AF}\rvert\times10^{45} (GeV)kF,E+B×1032k_{F,E+B} \times10^{32}
SPA+BK18+SPTBBBB+SPTpolBBBB<7.75< 7.75<3.82< 3.82<5.99< 5.99
SPA+P20EBEB+BK18+SPTBBBB+SPTpolBBBB<8.51< 8.51<3.72< 3.72<6.27< 6.27
SPA+BK18+SPTBBBB+SPTpolBBBB+CLASS<7.82< 7.82<3.91< 3.91<5.61< 5.61
Planck+SO<6.44< 6.44<3.10< 3.10<4.11< 4.11
LiteBIRD TEB only<4.19< 4.19<2.35< 2.35<5.61< 5.61
LiteBIRD TEBV (βHWP=20∘\beta_{\mathrm{HWP}} = 20^\circ)<4.40< 4.40<2.43< 2.43<0.492< 0.492

Table 4. 95% CL upper limits on the SME coefficients ∣k(V)00(3)∣\lvert k^{(3)}_{(V)00}\rvert and ∣kAF∣\lvert\mathbf{k}_{AF}\rvert (in units of 10−4510^{-45} GeV) and kF,E+Bk_{F,E+B} (in units of 10−3210^{-32}), obtained by recasting the bounds on βAF,T2\beta^{2}_{AF,T}, βAF,S2\beta^{2}_{AF,S} and βF2\beta^{2}_{F} reported in tables Table 2 and Table 3 through eqs. (18)–(20). For kF,E+Bk_{F,E+B}, the frequency ν\nu in eq. (20) is replaced by an effective frequency νf\nu_f, which is that of Planck for current data, including the combination with CLASS, and that of SO and LiteBIRD for the respective forecasts (see main text for more details).

∣κ(V)00(3)∣<7.75×10−45 GeV,(26)\left|\kappa_{(V)00}^{(3)}\right| < 7.75 \times10^{-45}\ \mathrm{GeV}, \tag*{(26)}
∣kAF∣<3.82×10−45 GeV,(27)\left|k_{\mathrm{AF}}\right| < 3.82 \times10^{-45}\ \mathrm{GeV}, \tag*{(27)}
kF,E+B<5.99×10−32(νf158.8 GHz)−1.(28)k_{\mathrm{F,E+B}} < 5.99 \times10^{-32} \left(\frac{\nu_f}{158.8\ \mathrm{GHz}}\right)^{-1}. \tag*{(28)}

obtained in [53] with a similar approach, based only on white noise and beam, for an analogous extension of Λ\LambdaCDM.

Note that the bound on kF,E+Bk_{F,E+B} is normalized to the effective frequency of the Planck channels, νf=158.8\nu_f = 158.8 GHz, since the constraint on βF2\beta_F^2 is driven by EE and TE Planck data.16 This value is computed using the sensitivities and nominal bandwidths of the 100, 143 and 217 GHz channels [99]. Following Paper I, the Planck channels are weighted by their temperature sensitivities, since the constraint on βF2\beta_F^2 comes also from the TE spectrum. Weighting them by their polarization sensitivities would instead give νf≃151.3\nu_f \simeq151.3 GHz, loosening the bound on kF,E+Bk_{F,E+B} by about 5%.

Compared to the constraints of Paper I, obtained with Planck+BK18+ACT DR4 data, the bounds on both CPT-odd coefficients are reduced by ∼25%\sim25\%. These constraints represent the strongest bounds on the CPT-odd coefficients of the minimal SME to date, considering both CMB and non-CMB probes (see [100] for a compilation of existing constraints). When the Planck PR4 EB spectrum is included, the posterior of ∣k(V)00(3)∣\lvert k^{(3)}_{(V)00}\rvert peaks away from zero, reflecting the mild preference for a non-zero isotropic rotation discussed in § 4.1, so that the 95% CL upper limit shifts to 8.51×10−458.51 \times10^{-45} GeV. The bound on ∣kAF∣\lvert\mathbf{k}_{AF}\rvert is instead essentially unchanged. Looking at the forecasts, Planck+SO would further reduce the bounds on the CPT-odd SME coefficients by ∼20%\sim20\% with respect to current data, while LiteBIRD in the TEB-only configuration would reduce them by ∼45%\sim45\% for ∣k(V)00(3)∣\lvert k^{(3)}_{(V)00}\rvert and ∼40%\sim40\% for ∣kAF∣\lvert\mathbf{k}_{AF}\rvert. Including V-mode sensitivity through a non-ideal HWP has no relevant impact on these coefficients, as the bounds change by at most ∼5%\sim5\% with respect to the TEB-only case for all the values of βHWP\beta_{\mathrm{HWP}} considered.

For the CPT-even coefficient, our baseline bound on kF,E+Bk_{F,E+B} is comparable to that of Paper I. Including CLASS data tightens it by about 7%. Note that the bound we report with CLASS is normalized to the effective frequency of Planck, which continues to provide most of the constraining power on kF,E+Bk_{F,E+B}. Astrophysical sources emitting more energetic photons provide much stronger bounds on the CPT-even coefficients, from the 10−3410^{-34}–10−3510^{-35} level with optical polarimetry [101, 102] down to ∼10−37\sim10^{-37} with gamma-ray burst polarimetry [103, 104], since the effect of this operator grows with the photon energy. These works, however, constrain different combinations of the components of (kF)αβμν(k_F)^{\alpha\beta\mu\nu} from the one probed by our analysis, kF,E+Bk_{F,E+B}, so the two kinds of bounds are not directly comparable.

For the forecasts, the effective frequencies are computed from the channels used to construct the noise curves of each experiment. For SO, we use νf=116.2\nu_f = 116.2 GHz, obtained from the six LAT channels (27, 39, 93, 145, 225 and 280 GHz) with the baseline polarization sensitivities of [51] and the passbands taken from [105]. For LiteBIRD, we use νf=139.6\nu_f = 139.6 GHz, obtained from the seven channels between 78 and 195 GHz (see § 3.2). As reported in Table 4, Planck+SO would reduce the bound on kF,E+Bk_{F,E+B} by ∼30%\sim30\% with respect to current data, while LiteBIRD in the TEB-only configuration would reduce it by only ∼5%\sim5\%, remaining weaker than Planck+SO. Including V-mode sensitivity changes this picture. For βHWP=20∘\beta_{\mathrm{HWP}} = 20^\circ, the bound improves by a factor of ∼12\sim12 with respect to current data and by a factor of ∼8\sim8 with respect to Planck+SO.

Conclusions

In this work, we have updated the CMB constraints on Lorentz-violating electrodynamics within the minimal Standard Model Extension, considering the CPT-odd, dimension-3 operator, which sources cosmic birefringence, and the CPT-even, dimension-4 operator, which converts linear into circular polarization. Combining Planck, ACT DR6 and SPT-3G D1 temperature and EE-mode polarization data with BB-mode measurements from BICEP/Keck, SPT-3G D1 and SPTpol, we obtain ∣k(V)00(3)∣<7.75×10−45\lvert k^{(3)}_{(V)00}\rvert< 7.75 \times10^{-45} GeV and ∣kAF∣<3.82×10−45\lvert\mathbf{k}_{AF}\rvert< 3.82 \times10^{-45} GeV at 95% CL on the coefficients sourcing isotropic and anisotropic birefringence, respectively. These are the strongest bounds to date on these coefficients from any probe [100]. With respect to our previous analysis [8], these bounds improve by about 25%, thanks to the precision of the SPT-3G D1 BB-mode measurements at ℓ≥300\ell\ge300, where the leakage of EE into BB modes induced by these operators is significant. More precise BB-mode data at these scales would therefore further tighten these constraints. At these scales, the BB-mode spectrum is dominated by gravitational lensing, so that the EE-to-BB leakage induced by Lorentz violation may also be relevant for lensing analyses, either as a contaminant or, for anisotropic birefringence, through the induced coupling between multipoles.

For the first time in this kind of analysis, we have also included the Planck PR4 EBEB spectrum, which directly traces isotropic birefringence. These data yield a mild preference for a non-zero k(V)00(3)k^{(3)}_{(V)00}, corresponding to a birefringence angle about 1σ1\sigma away from zero and consistent, although with a lower significance, with the one measured from the same Planck release with a different methodology in [11]. Since our analysis attributes the entire EBEB signal to a physical rotation, assessing the origin of this signal will require modelling the mis-calibration of the polarization angles jointly with Lorentz violation, together with a better characterization of the polarized Galactic emission used to calibrate them [96, 98]. More generally, several ground-based experiments, including BICEP/Keck and SPT-3G, calibrate the global polarization angle by nulling the observed EBEB and TBTB spectra [42, 46]. This procedure removes any isotropic rotation from the maps, including a cosmological one, so that the corresponding BB-mode data mainly retain sensitivity to anisotropic birefringence. Planck and ACT, instead, do not null the EBEB spectrum [44, 106], so that their data retain sensitivity to an isotropic rotation, within the uncertainty on the orientation of their polarization-sensitive detectors [106]. A consistent treatment of the polarization angle calibration across datasets will therefore be essential to robustly constrain the isotropic component.

For the CPT-even coefficient, temperature and linear-polarization data give kF,E+B<5.99×10−32k_{F,E+B} < 5.99 \times10^{-32} at 95% CL. Including the latest CLASS circular polarization measurements at 90, 150 and 220 GHz [50], we tighten this bound by about 7%. This is the first time that CMB VV-mode data contribute to constraining this coefficient. In this respect, we defer to future work the inclusion of the recent data from the POLARBEAR collaboration [107], which probe angular scales up to ℓ=2000\ell= 2000. Without circular polarization measurements, the constraint relies on the amplitude rescaling of the linear-polarization spectra, which is degenerate with their calibration factors. An accurate external calibration of the polarization amplitude would therefore make this bound more robust. Furthermore, a multi-frequency analysis, sampling kF,E+Bk_{F,E+B} directly rather than relying on an effective frequency, would also fully exploit the frequency dependence of the CPT-even effect.

We have then forecast the sensitivity of Simons Observatory (SO) and LiteBIRD to these coefficients. These forecasts neglect foreground residuals in the SO BB-mode data and in all LiteBIRD data, and should therefore be regarded as optimistic. Using linear polarization data only, we find that the two experiments are complementary, with SO tightening the CPT-even bound by about 30% and LiteBIRD the CPT-odd ones by 40 to 45%. For LiteBIRD, we have further considered a configuration including a rotating half-wave plate (HWP), whose non-idealities, parametrized by a departure of its phase shift from the ideal value, provide sensitivity to VV modes [53, 91]. We have explored four representative values of this departure, between 1° and 20°. The CPT-even bound improves rapidly with the phase shift, by more than one order of magnitude for the largest value, without degrading the other constraints. HWP non-idealities, usually treated as a systematic to be mitigated, can thus become a resource for testing fundamental physics. We have not considered here the new LiteBIRD setup proposed in the revised mission concept [88]. While we expect the constraints on the CPT-odd coefficients to be largely unaffected, its improved angular resolution could strengthen the CPT-even bound from linear polarization. We defer a quantitative assessment to future work.

A natural extension of this work is to go beyond the CMB power spectra, which are only sensitive to quadratic combinations of the SME coefficients. The direction-dependent components of kAFk_{AF} and kFk_F induce correlations between different multipoles, ℓ≠ℓ′\ell\ne\ell', which carry information that power spectra cannot access. For kAFk_{AF}, power spectra constrain the amplitude of its spatial part but not its direction, which sets a dipolar modulation of the rotation angle across the sky and could be measured with existing estimators of anisotropic birefringence, see e.g. [108, 109]. For kFk_F, to our knowledge, no analogous estimator for the conversion of linear into circular polarization has been developed yet. Such an estimator would give access to further combinations of the components of kFk_F, separating the two contractions that power spectra only constrain through kF,E+Bk_{F,E+B} (eq. (20)). We leave such map-level analyses, which will be key to fully exploiting upcoming CMB polarization data as a probe of Lorentz invariance, to future work.

Acknowledgments

We thank Lennart Balkenhol for his support with the SPT data and the use of the candl likelihood framework. We also thank Paolo Campeti, Eric Hivon and Ali Rida Khalife for useful discussions. We acknowledge CINECA for the availability of high performance computing resources and support through the CINECA-INFN agreement for the InDark project. We acknowledge support by Commissione Scientifica Nazionale 4 (CSN4) and the Research Network InDark funded by Istituto Nazionale di Fisica Nucleare (INFN), and by the ASI (Italian Space Agency) Grant 2020-9-HH.0 (participation in LiteBIRD, phase A). We acknowledge the use of numpy [110], matplotlib [111] and getdist [112] software packages. L.C. acknowledges the financial support provided by Fundação para a Ciência e Tecnologia (FCT), I.P., through the Strategic Funding UID/04650/2025 and UID/04564/2025 and national funds with DOI identifiers 10.54499/2023.11681.PEX, 10.54499/2024.00249.CERN funded by measure RE-C06-i06.m02 – “Reinforcement of funding for International Partnerships in Science, Technology and Innovation” of the Recovery and Resilience Plan – RRP, within the framework of the financing contract signed between the Recover Portugal Mission Structure (EMRP) and the Foundation for Science and Technology I.P. (FCT), as an intermediate beneficiary, as well as the advanced computing projects 2024.00249.CERN.F1 and 2024.07037.CPCA.A1. MG, ML and LP are funded by the European Union (ERC, RELiCS, project number 101116027). Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Council Executive Agency. Neither the European Union nor the granting authority can be held responsible for them. This is not an official Simons Observatory or LiteBIRD collaboration paper.

Impact of BB-mode datasets on the constraints

In § 4.1 we showed that adding SPT-3G D1 and SPTpol BB-mode data to the SPA+BK18 combination improves the bounds on βAF,T2\beta_{AF,T}^{2} and βAF,S2\beta_{AF,S}^{2} by nearly a factor of 2. Here we assess the role of BB-mode data in more detail, first by comparing the baseline combination with the SPA combination alone, and then by removing in turn SPTpol, SPT-3G D1 or BK18 data from the baseline combination.

Without any BB-mode data, the bounds on βAF,T2\beta_{AF,T}^{2} and βAF,S2\beta_{AF,S}^{2} are about 20 times looser than for the baseline combination (Table 5). In this case the CPT-odd parameters are constrained essentially only through the scale-independent rescaling of the TETE and EEEE spectra by Z\mathcal{Z} (eq. (9)). The comparison with the baseline combination therefore shows that almost all the information on these parameters comes from the EE-to-BB leakage probed by BB-mode data. Adding BK18 alone already tightens these bounds by a factor of ∼12\sim12 (Table 2).

DatasetβAF,T2×104\beta_{AF,T}^{2} \times10^{4}βAF,S2×104\beta_{AF,S}^{2} \times10^{4}βF2×102\beta_{F}^{2} \times10^{2}
SPA+BK18+SPTBB+SPTpolBB<3.59< 3.59<3.65< 3.65<4.72< 4.72
SPA+BK18+SPTBB<3.31< 3.31<3.31< 3.31<4.87< 4.87
SPA+BK18+SPTpolBB<6.50< 6.50<6.52< 6.52<4.80< 4.80
SPA+SPTBB+SPTpolBB<4.18< 4.18<4.23< 4.23<4.64< 4.64
SPA<79.8< 79.8<77.8< 77.8<3.84< 3.84

Table 5. 95% CL upper limits on βAF,T2\beta_{AF,T}^{2}, βAF,S2\beta_{AF,S}^{2} (in units of 10−410^{-4}) and βF2\beta_{F}^{2} (in units of 10−210^{-2}), for the baseline combination, for the combinations obtained by removing SPTpol, SPT-3G D1 or BK18 BB-mode data individually, and for the SPA combination without any BB-mode data.

The further improvement obtained by adding SPT BB-mode data to SPA+BK18 is almost entirely driven by SPT-3G D1 (Figure 6 and Table 5). The SPA+BK18+SPTBBBB combination yields bounds comparable to, and in fact slightly tighter than, those of the baseline combination, whereas SPA+BK18+SPTpolBBBB gives bounds roughly a factor of 2 looser on both parameters, close to those of SPA+BK18 (Table 2). This is likely due to the lowest-multipole SPTpol band power, which lies above the Λ\LambdaCDM prediction and therefore prefers a slightly larger EE-to-BB leakage (see left panel of Figure 1).

Figure 6. One- and two-dimensional posterior distributions of βAF,T2\beta_{AF,T}^{2}, βAF,S2\beta_{AF,S}^{2} and βF2\beta_{F}^{2} for the baseline combination SPA+BK18+SPTBB+SPTpolBB compared with the combinations obtained by removing SPTpol (SPA+BK18+SPTBB), SPT-3G D1 (SPA+BK18+SPTpolBB) or BK18 (SPA+SPTBB+SPTpolBB) BB-mode data.

We then remove BK18 while keeping both SPT BB-mode datasets. In this case we also include the SPT-3G D1 band powers that overlap with BK18, which are excluded from the baseline combination (semi-transparent points in the left panel of Figure 1). The bounds on βAF,T2\beta_{AF,T}^{2} and βAF,S2\beta_{AF,S}^{2} loosen by only ∼15%\sim15\% with respect to the baseline combination, and remain tighter than those from SPA+BK18. With current data, the sensitivity to these parameters is therefore dominated by the EE-to-BB leakage at ℓ≳300\ell\gtrsim300 probed by SPT-3G D1, rather than by the BB-mode measurements of BK18 at larger angular scales.

The bound on βF2\beta_{F}^{2}, on the other hand, is essentially unaffected by which BB-mode dataset is included, since in the absence of VV-mode data this parameter is constrained through the rescaling of the TETE and EEEE spectra by Z\mathcal{Z} rather than through the EE-to-BB leakage (see § 4.1). Nevertheless, without any BB-mode data, the bound on βF2\beta_{F}^{2} is tighter. In this case all three LV parameters enter the spectra only through their weighted sum Z\mathcal{Z}, and, since their priors are positive, the larger values now allowed for the CPT-odd parameters leave less room for βF2/4\beta_{F}^{2}/4.

On the degeneracy with polarization calibration parameters

As discussed in § 4.1, in the absence of VV-mode data βF2\beta^{2}_{F} enters the spectra only through the scale-independent rescaling of the TETE and EEEE spectra by Z\mathcal{Z} (eq. (9)), which is partially degenerate with the polarization calibration parameters sampled in the ACT DR6 and SPT-3G D1 likelihoods, pACTp_{\mathrm{ACT}} and EcalSPTE^{\mathrm{SPT}}_{\mathrm{cal}}. In this appendix we examine this degeneracy in more detail and quantify its impact on the constraints.

Figure 7 shows the posterior distributions of the LV parameters and of the two calibration parameters for the baseline combination, and Figure 8 (left) reports the corresponding correlation coefficients. βF2\beta^{2}_{F} is anticorrelated with both pACTp_{\mathrm{ACT}} and EcalSPTE^{\mathrm{SPT}}_{\mathrm{cal}}, with correlation coefficients of −0.58-0.58 and −0.38-0.38, respectively. Opening the LV parameter space also broadens the posteriors of both calibration parameters and shifts them towards lower values with respect to the Λ\LambdaCDM+rr analysis, as expected if they compensate for a non-zero Z\mathcal{Z}. In contrast, the correlation between the calibration parameters and βAF,T2\beta^{2}_{AF,T} or βAF,S2\beta^{2}_{AF,S} is negligible. The latter also source a leakage of EE into BB modes, whose amplitude and scale dependence cannot be mimicked by a calibration of the polarization spectra, so that the BB-mode data break the degeneracy. Removing all BB-mode data confirms this, since for the SPA combination alone βAF,T2\beta^{2}_{AF,T} and βAF,S2\beta^{2}_{AF,S} become anticorrelated with both calibration parameters, like βF2\beta^{2}_{F} (Figure 8, right panel).

Figure 7. One- and two-dimensional posterior distributions of βAF,T2\beta^{2}_{AF,T}, βAF,S2\beta^{2}_{AF,S} and βF2\beta^{2}_{F}, together with the polarization calibration parameters of ACT and SPT (pACTp_{\mathrm{ACT}} and EcalSPTE^{\mathrm{SPT}}_{\mathrm{cal}}, respectively), for the baseline combination, SPA+BK18+SPTBBBB+SPTpolBBBB. For comparison, we also show the polarization calibration parameters obtained in the Λ\LambdaCDM+rr analysis.

To quantify the impact of this degeneracy, we repeat the analysis of the baseline combination fixing both calibration parameters to unity (Figure 9). The bounds on βAF,T2\beta^{2}_{AF,T} and βAF,S2\beta^{2}_{AF,S} remain essentially unchanged, while the width of the βF2\beta^{2}_{F} posterior is reduced by about 15%, showing that, once their calibration is fixed, ACT and SPT polarization data contribute to the constraint on Z\mathcal{Z}. The posterior also moves away from zero, with a mild preference for βF2>0\beta^{2}_{F}>0 at 68% CL, consistent with EcalSPTE^{\mathrm{SPT}}_{\mathrm{cal}} favouring values slightly above unity in the Λ\LambdaCDM+rr analysis (Figure 7).

Figure 9. One- and two-dimensional posterior distributions of βAF,T2\beta^{2}_{AF,T}, βAF,S2\beta^{2}_{AF,S} and βF2\beta^{2}_{F} for the baseline combination, with the polarization calibration parameters of ACT and SPT varied (red) or fixed to unity (dark red lines), compared with the P18+BK18 combination (black dashed).

Similar considerations apply to Planck, whose polarization efficiencies are not sampled in the Plik_lite likelihood, but fixed to the values obtained by fitting the EEEE spectrum of each frequency channel to the Λ\LambdaCDM model that best fits TTTT [70]. Any rescaling of the polarization spectra by Z\mathcal{Z} would therefore have been absorbed into these efficiencies, and the Planck bound on βF2\beta_{\mathrm{F}}^{2}, which dominates the constraint from current data, does not account for their uncertainty. As also noted in [44], fixing the polarization efficiency can potentially disfavour extensions of Λ\LambdaCDM that modify the amplitude of polarization relative to temperature.

Without VV-mode data, the polarization calibration is therefore a key limitation for the constraint on βF2\beta_{\mathrm{F}}^{2}, which would benefit from an accurate external calibration of the polarization amplitude. Circular polarization can instead break this degeneracy, since the VVVV spectrum sourced by the CPT-even operator cannot be mimicked by a miscalibration of the linear polarization. Current CLASS data already begin to do so, as including them in the baseline combination reduces the correlation coefficients of βF2\beta_{\mathrm{F}}^{2} with pACTp_{\mathrm{ACT}} and EcalSPTE_{\mathrm{cal}}^{\mathrm{SPT}} from −0.58-0.58 and −0.38-0.38 to −0.49-0.49 and −0.34-0.34, respectively.

References

  1. [1]D. Mattingly, Modern tests of Lorentz invariance, Living Rev. Rel. 8 (2005) 5, [gr-qc/0502097].
  2. [2]G. Amelino-Camelia, Quantum-Spacetime Phenomenology, Living Rev. Rel. 16 (2013) 5, [0806.0339].
  3. [3]A. Addazi et al., Quantum gravity phenomenology at the dawn of the multi-messenger era—A review, Prog. Part. Nucl. Phys. 125 (2022) 103948, [2111.05659].DOI
  4. [4]D. Colladay and V. A. Kostelecký, Lorentz violating extension of the standard model, Phys. Rev. D 58 (1998) 116002, [hep-ph/9809521].
  5. [5]V. A. Kostelecký and M. Mewes, Signals for Lorentz violation in electrodynamics, Phys. Rev. D 66 (2002) 056005, [hep-ph/0205211].
  6. [6]W. Hu and S. Dodelson, Cosmic Microwave Background Anisotropies, Ann. Rev. Astron. Astrophys. 40 (2002) 171–216, [astro-ph/0110441].
  7. [7]V. A. Kostelecký and M. Mewes, Lorentz-violating electrodynamics and the cosmic microwave background, Phys. Rev. Lett. 99 (2007) 011601, [astro-ph/0702379].
  8. [8]L. Caloni, S. Giardiello, M. Lembo, M. Gerbino, G. Gubitosi, M. Lattanzi et al., Probing Lorentz-violating electrodynamics with CMB polarization, JCAP 03 (2023) 018, [2212.04867].DOI
  9. [9]Y. Minami and E. Komatsu, New Extraction of the Cosmic Birefringence from the Planck 2018 Polarization Data, Phys. Rev. Lett. 125 (2020) 221301, [2011.11254].DOI
  10. [10]J. R. Eskilt and E. Komatsu, Improved constraints on cosmic birefringence from the WMAP and Planck cosmic microwave background polarization data, Phys. Rev. D 106 (2022) 063503, [2205.13962].DOI
  11. [11]P. Diego-Palazuelos et al., Cosmic Birefringence from the Planck Data Release 4, Phys. Rev. Lett. 128 (2022) 091302, [2201.07682].DOI
  12. [12]P. Diego-Palazuelos and E. Komatsu, Cosmic birefringence from the Atacama Cosmology Telescope Data Release 6, Phys. Rev. D 113 (2026) L101302, [2509.13654].
  13. [13]S. M. Carroll, G. B. Field and R. Jackiw, Limits on a Lorentz and Parity Violating Modification of Electrodynamics, Phys. Rev. D 41 (1990) 1231.
  14. [14]S. M. Carroll, Quintessence and the rest of the world, Phys. Rev. Lett. 81 (1998) 3067–3070, [astro-ph/9806099].DOI
  15. [15]M. Li and X. Zhang, Cosmological CPT violating effect on CMB polarization, Phys. Rev. D 78 (2008) 103516, [0810.0403].DOI
  16. [16]M. Pospelov, A. Ritz and C. Skordis, Pseudoscalar perturbations and polarization of the cosmic microwave background, Phys. Rev. Lett. 103 (2009) 051302, [0808.0673].
  17. [17]T. Fujita, K. Murai, H. Nakatsuka and S. Tsujikawa, Detection of isotropic cosmic birefringence and its implications for axionlike particles including dark energy, Phys. Rev. D 103 (2021) 043509, [2011.11894].DOI
  18. [18]I. Obata, Implications of the cosmic birefringence measurement for the axion dark matter search, JCAP 09 (2022) 062, [2108.02150].DOI
  19. [19]S. Gasparotto and I. Obata, Cosmic birefringence from monodromic axion dark energy, JCAP 08 (2022) 025, [2203.09409].
  20. [20]K. Murai, F. Naokawa, T. Namikawa and E. Komatsu, Isotropic cosmic birefringence from early dark energy, Phys. Rev. D 107 (2023) L041302, [2209.07804].
  21. [21]S. Gasparotto and E. I. Sfakianakis, Cosmic birefringence from the Axiverse, JCAP 11 (2023) 017, [2306.16355].arxiv.org/abs/2306.16355
  22. [22]F. Takahashi and W. Yin, Kilobyte Cosmic Birefringence from ALP Domain Walls, JCAP 04 (2021) 007, [2012.11576].DOI
  23. [23]N. Kijitama, F. Kozai, F. Takahashi and W. Yin, Power spectrum of domain-wall network, and its implications for isotropic and anisotropic cosmic birefringence, JCAP 10 (2022) 043, [2205.05083].DOI
  24. [24]M. Jain, A. J. Long and M. A. Amin, CMB birefringence from ultralight-axion string networks, JCAP 05 (2021) 055, [2103.10962].DOI
  25. [25]R. Z. Ferreira, S. Gasparotto, T. Hiramatsu, I. Obata and O. Pujolas, Axionic defects in the CMB: birefringence and gravitational waves, JCAP 05 (2024) 066, [2312.14104].
  26. [26]C. Q. Geng, S. H. Ho and J. N. Ng, Neutrino number asymmetry and cosmological birefringence, JCAP 09 (2007) 010, [0706.0080].DOI
  27. [27]R.-P. Zhou, D. Huang and C.-Q. Geng, Cosmic birefringence from neutrino and dark matter asymmetries, JCAP 07 (2023) 053, [2302.11140].DOI
  28. [28]K. Kasai and I. Obata, Cosmic Birefringence from Large Lepton Asymmetry, 2610.09748.
  29. [29]P. Montero-Camacho and C. M. Hirata, Exploring circular polarization in the CMB due to conventional sources of cosmic birefringence, JCAP 08 (2018) 040, [1803.04505].
  30. [30]S. King and P. Lubin, Circular polarization of the CMB: Foregrounds and detection prospects, Phys. Rev. D 94 (2016) 023501, [1606.04112].DOI
  31. [31]S. Alexander, J. Ochoa and A. Kosowsky, Generation of Circular Polarization of the Cosmic Microwave Background, Phys. Rev. D 79 (2009) 063524, [0810.2355].
  32. [32]D. Ejlli, Magneto-optic effects of the cosmic microwave background, Nucl. Phys. B 935 (2018) 83–128, [1607.02094].DOI
  33. [33]S. Tizchang, S. Batebi, M. Haghighat and R. Mohammadi, Cosmic microwave background polarization in non-commutative space-time, Eur. Phys. J. C 76 (2016) 478, [1605.09045].DOI
  34. [34]D. Ejlli, Millicharged fermion vacuum polarization in a cosmic magnetic field and generation of CMB elliptic polarization, Phys. Rev. D 96 (2017) 023540, [1704.01894].DOI
  35. [35]K. Inomata and M. Kamionkowski, Circular polarization of the cosmic microwave background from vector and tensor perturbations, Phys. Rev. D 99 (2019) 043501, [1811.04957].DOI
  36. [36]N. Bartolo, A. Hoseinpour, S. Matarrese, G. Orlando and M. Zarei, CMB Circular and B-mode Polarization from New Interactions, Phys. Rev. D 100 (2019) 043516, [1903.04578].
  37. [37]A. Hoseinpour, M. Zarei, G. Orlando, N. Bartolo and S. Matarrese, CMB V modes from photon-photon forward scattering revisited, Phys. Rev. D 102 (2020) 063501, [2006.14418].DOI
  38. [38]E. Bavarsad, M. Haghighat, Z. Rezaei, R. Mohammadi, I. Motie and M. Zarei, Generation of circular polarization of the CMB, Phys. Rev. D 81 (2010) 084035, [0912.2993].
  39. [39]A. Kushwaha and R. K. Jain, Probing axion photon conversion via circular polarization imprints in the CMB V-mode observations, Phys. Rev. D 113 (2026) 123511, [2510.22592].
  40. [40]M. Lembo, M. Lattanzi, L. Pagano, A. Gruppuso, P. Natoli and F. Forastieri, Cosmic Microwave Background Polarization as a Tool to Constrain the Optical Properties of the Universe, Phys. Rev. Lett. 127 (2021) 011301, [2007.08486].DOI
  41. [41]Planck collaboration, N. Aghanim et al., Planck 2018 results. VI. Cosmological parameters, Astron. Astrophys. 641 (2020) A6, [1807.06209].
  42. [42]BICEP, Keck collaboration, P. A. R. Ade et al., Improved Constraints on Primordial Gravitational Waves using Planck, WMAP, and BICEP/Keck Observations through the 2018 Observing Season, Phys. Rev. Lett. 127 (2021) 151301, [2110.00483].
  43. [43]ACT collaboration, S. Aiola et al., The Atacama Cosmology Telescope: DR4 Maps and Cosmological Parameters, JCAP 12 (2020) 047, [2007.07288].
  44. [44]Atacama Cosmology Telescope collaboration, T. Louis et al., The Atacama Cosmology Telescope: DR6 power spectra, likelihoods and ΛCDM parameters, JCAP 11 (2025) 062, [2503.14452].
  45. [45]SPT-3G collaboration, E. Camphuis et al., SPT-3G D1: CMB temperature and polarization power spectra and cosmology from 2019 and 2020 observations of the SPT-3G main field, Phys. Rev. D 113 (2026) 083504, [2506.20707].DOI
  46. [46]SPT-3G collaboration, J. A. Zebrowski et al., SPT-3G D1: Constraints on inflationary gravitational waves with two years of SPT-3G data, Phys. Rev. D 112 (2025) 123520, [2505.02827].
  47. [47]SPT collaboration, J. T. Sayre et al., Measurements of B-mode Polarization of the Cosmic Microwave Background from 500 Square Degrees of SPTpol Data, Phys. Rev. D 101 (2020) 122003, [1910.05748].
  48. [48]M. Tristram et al., Planck constraints on the tensor-to-scalar ratio, Astron. Astrophys. 647 (2021) A128, [2010.01139].DOI
  49. [49]M. Tristram et al., Improved limits on the tensor-to-scalar ratio using BICEP and Planck data, Phys. Rev. D 105 (2022) 083524, [2112.07961].
  50. [50]T. Essinger-Hileman et al., Improved Constraints on Cosmic Microwave Background Circular Polarization with CLASS, 2609.40125.
  51. [51]Simons Observatory collaboration, P. Ade et al., The Simons Observatory: Science goals and forecasts, JCAP 02 (2019) 056, [1808.07445].
  52. [52]LiteBIRD collaboration, E. Allys et al., Probing Cosmic Inflation with the LiteBIRD Cosmic Microwave Background Polarization Survey, PTEP 2023 (2023) 042F01, [2202.02773].
  53. [53]N. Raffuzzi, M. Lembo, S. Giardiello, M. Gerbino, M. Lattanzi, P. Natoli et al., Unveiling V modes: enhancing CMB sensitivity to BSM physics with a non-ideal half-wave plate, JCAP 03 (2025) 009, [2409.10424].
  54. [54]J. R. Eimer et al., CLASS Angular Power Spectra and Map-component Analysis for 40 GHz Observations through 2022, Astrophys. J. 963 (2024) 92, [2309.00675].
  55. [55]SPIDER collaboration, J. M. Nagy et al., A New Limit on CMB Circular Polarization from SPIDER, Astrophys. J. 844 (2017) 151, [1704.00215].
  56. [56]A. Lewis, A. Challinor and A. Lasenby, Efficient computation of CMB anisotropies in closed FRW models, Astrophys. J. 538 (2000) 473–476, [astro-ph/9911177].DOI
  57. [57]C. Howlett, A. Lewis, A. Hall and A. Challinor, CMB power spectrum parameter degeneracies in the era of precision cosmology, JCAP 04 (2012) 027, [1201.3654].DOI
  58. [58]G. Gubitosi and F. Paci, Constraints on cosmological birefringence energy dependence from CMB polarization data, JCAP 02 (2013) 020, [1211.3321].DOI
  59. [59]J. Torrado and A. Lewis, Cobaya: Code for Bayesian Analysis of hierarchical physical models, JCAP 05 (2021) 057, [2005.05290].DOI
  60. [60]P. F. de Salas, D. V. Forero, S. Gariazzo, P. Martínez-Miravé, O. Mena, C. A. Ternes et al., 2020 global reassessment of the neutrino oscillation picture, JHEP 02 (2021) 071, [2006.11237].DOI
  61. [61]I. Esteban, M. C. Gonzalez-Garcia, M. Maltoni, T. Schwetz and A. Zhou, The fate of hints: updated global analysis of three-flavour neutrino oscillations, JHEP 09 (2020) 178, [2007.14792].DOI
  62. [62]F. Capozzi, E. Di Valentino, E. Lisi, A. Marrone, A. Melchiorri and A. Palazzo, Unfinished fabric of the three neutrino paradigm, Phys. Rev. D 104 (2021) 083031, [2107.00532].arxiv.org/abs/2107.00532
  63. [63]G. Mangano, G. Miele, S. Pastor and M. Peloso, A Precision calculation of the effective number of cosmological neutrinos, Phys. Lett. B 534 (2002) 8–16, [astro-ph/0111408].
  64. [64]J. J. Bennett, G. Buldgen, M. Drewes and Y. Y. Y. Wong, Towards a precision calculation of the effective number of neutrinos N_eff in the Standard Model I: the QED equation of state, JCAP 03 (2020) 003, [1911.04504].DOI
  65. [65]J. J. Bennett, G. Buldgen, P. F. De Salas, M. Drewes, S. Gariazzo, S. Pastor et al., Towards a precision calculation of N_eff in the Standard Model II: Neutrino decoupling in the presence of flavour oscillations and finite-temperature QED, JCAP 04 (2021) 073, [2012.02726].DOI
  66. [66]K. Akita and M. Yamaguchi, A precision calculation of relic neutrino decoupling, JCAP 08 (2020) 012, [2005.07047].DOI
  67. [67]J. Froustey, C. Pitrou and M. C. Volpe, Neutrino decoupling including flavour oscillations and primordial nucleosynthesis, JCAP 12 (2020) 015, [2008.01074].DOI
  68. [68]M. Drewes, Y. Georis, M. Klasen, L. P. Wiggering and Y. Y. Y. Wong, Towards a precision calculation of N_eff in the Standard Model. Part III. Improved estimate of NLO contributions to the collision integral, JCAP 06 (2024) 032, [2402.18481].DOI
  69. [69]A. Gelman and D. B. Rubin, Inference from Iterative Simulation Using Multiple Sequences, Statist. Sci. 7 (1992) 457–472.DOI
  70. [70]Planck collaboration, N. Aghanim et al., Planck 2018 results. V. CMB power spectra and likelihoods, Astron. Astrophys. 641 (2020) A5, [1907.12875].
  71. [71]L. Pagano, J. M. Delouis, S. Mottet, J. L. Puget and L. Vibert, Reionization optical depth determination from Planck HFI data with ten percent accuracy, Astron. Astrophys. 635 (2020) A99, [1908.09856].DOI
  72. [72]J. M. Delouis, L. Pagano, S. Mottet, J. L. Puget and L. Vibert, Sroll2: an improved mapmaking approach to reduce large-scale systematic effects in the Planck High Frequency Instrument legacy maps, Astron. Astrophys. 629 (2019) A38, [1901.11386].DOI
  73. [73]J. Carron, M. Mirmelstein and A. Lewis, CMB lensing from Planck PR4 maps, JCAP 09 (2022) 039, [2206.07773].
  74. [74]L. Balkenhol, C. Trendafilova, K. Benabed and S. Galli, candl: cosmic microwave background analysis with a differentiable likelihood, Astron. Astrophys. 686 (2024) A10, [2401.13433].DOI
  75. [75]ACT collaboration, M. S. Madhavacheril et al., The Atacama Cosmology Telescope: DR6 Gravitational Lensing Map and Cosmological Parameters, Astrophys. J. 962 (2024) 113, [2304.05203].
  76. [76]ACT collaboration, F. J. Qu et al., The Atacama Cosmology Telescope: A Measurement of the DR6 CMB Lensing Power Spectrum and Its Implications for Structure Growth, Astrophys. J. 962 (2024) 112, [2304.05202].
  77. [77]ACT collaboration, N. MacCrann et al., The Atacama Cosmology Telescope: Mitigating the Impact of Extragalactic Foregrounds for the DR6 Cosmic Microwave Background Lensing Analysis, Astrophys. J. 966 (2024) 138, [2304.05196].
  78. [78]SPT-3G collaboration, F. Ge et al., Cosmology from CMB lensing and delensed EE power spectra using 2019–2020 SPT-3G polarization data, Phys. Rev. D 111 (2025) 083534, [2411.06000].
  79. [79]L. Balkenhol, A. Coerver, C. L. Reichardt and J. A. Zebrowski, Probing Anisotropic Cosmic Birefringence with Foreground-Marginalised SPT B-mode Likelihoods, Open J. Astrophys. 8 (2025) 147459, [2510.07928].DOI
  80. [80]G. Galloni, S. Henrot-Versillé and M. Tristram, Robust constraints on tensor perturbations from cosmological data: A comparative analysis from Bayesian and frequentist perspectives, Phys. Rev. D 110 (2024) 063511, [2405.04455].DOI
  81. [81]I. L. Padilla, J. R. Eimer, Y. Li, G. E. Addison, A. Ali, J. W. Appel et al., Two-year cosmology large angular scale surveyor (class) observations: A measurement of circular polarization at 40 ghz, Astrophys. J. 889 (jan, 2020) 105, [1911.00391].DOI
  82. [82]L. Perotto, J. Lesgourgues, S. Hannestad, H. Tu and Y. Y. Y. Wong, Probing cosmological parameters with the CMB: Forecasts from full Monte Carlo simulations, JCAP 10 (2006) 013, [astro-ph/0606227].DOI
  83. [83]M. Gerbino, M. Lattanzi, M. Migliaccio, L. Pagano, L. Salvati, L. Colombo et al., Likelihood methods for CMB experiments, Front. in Phys. 8 (2020) 15, [1909.09375].
  84. [84]M. Rashkovetskyi, J. B. Muñoz, D. J. Eisenstein and C. Dvorkin, Small-scale clumping at recombination and the Hubble tension, Phys. Rev. D 104 (2021) 103517, [2108.02747].DOI
  85. [85]W. L. K. Wu, J. Errard, C. D. Dvorkin, C. L. Kuo, A. T. Lee, P. McDonald et al., A Guide to Designing Future Ground-based Cosmic Microwave Background Experiments, Astrophys. J. 788 (2014) 138, [1402.4108].DOI
  86. [86]Simons Observatory collaboration, I. Abril-Cabezas et al., The Simons Observatory: forecasted constraints on primordial gravitational waves with the expanded array of Small Aperture Telescopes, JCAP 04 (2026) 051, [2512.15833].
  87. [87]L. Caloni, R. Z. Ferreira, L. Sousa and C. Winckler, Cosmic strings and domain walls: the impact of CMB B-mode data, 2602.20050.DOI
  88. [88]LiteBIRD collaboration, K. Aizawa et al., LiteBIRD Mission Overview after Mission Reformation, in SPIE Astronomical Telescopes + Instrumentation 2026, 7, 2026. 2607.25285.
  89. [89]LiteBIRD collaboration, P. Campeti et al., LiteBIRD science goals and forecasts. A case study of the origin of primordial gravitational waves using large-scale CMB polarization, JCAP 06 (2024) 008, [2312.00717].
  90. [90]S. Micheli et al., Systematic effects induced by half-wave plate differential optical load and TES nonlinearity for LiteBIRD, Proc. SPIE Int. Soc. Opt. Eng. 13102 (2024) 131022R, [2407.15294].arxiv.org/abs/2407.15294
  91. [91]S. Giardiello et al., Detailed study of HWP non-idealities and their impact on future measurements of CMB polarization anisotropies from space, Astron. Astrophys. 658 (2022) A15, [2106.08031].DOI
  92. [92]LiteBIRD collaboration, N. Raffuzzi et al., Mitigating half-wave plate systematics at the map-making level: calibration requirements for LiteBIRD, JCAP 08 (2026) 053, [2602.02410].
  93. [93]K. Komatsu, H. Ishino, N. Katayama, T. Matsumura and Y. Sakurai, Design of a frequency-independent optic axis Pancharatnam-based achromatic half-wave plate, Journal of Astronomical Telescopes, Instruments, and Systems 7 (2021) 034005.DOI
  94. [94]G. Pisano et al., Development and application of metamaterial-based half-wave plates for the NIKA and NIKA2 polarimeters, Astron. Astrophys. 658 (2022) A24, [2006.12081].DOI
  95. [95]A. Lue, L.-M. Wang and M. Kamionkowski, Cosmological signature of new parity violating interactions, Phys. Rev. Lett. 83 (1999) 1506-1509, [astro-ph/9812088].DOI
  96. [96]Y. Minami, H. Ochi, K. Ichiki, N. Katayama, E. Komatsu and T. Matsumura, Simultaneous determination of the cosmic birefringence and miscalibrated polarization angles from CMB experiments, PTEP 2019 (2019) 083E02, [1904.12440].arxiv.org/abs/1904.12440
  97. [97]Y. Minami, Determination of miscalibrated polarization angles from observed cosmic microwave background and foreground EB power spectra: Application to partial-sky observation, PTEP 2020 (2020) 063E01, [2002.03572].DOI
  98. [98]S. E. Clark, C.-G. Kim, J. C. Hill and B. S. Hensley, The Origin of Parity Violation in Polarized Dust Emission and Implications for Cosmic Birefringence, Astrophys. J. 919 (2021) 53, [2105.00120].DOI
  99. [99]Planck collaboration, N. Aghanim et al., Planck 2018 results. I. Overview and the cosmological legacy of Planck, Astron. Astrophys. 641 (2020) A1, [1807.06205].
  100. [100]V. A. Kostelecky and N. Russell, Data Tables for Lorentz and CPT Violation, 0801.0287.
  101. [101]A. S. Friedman, R. Gerasimov, D. Leon, W. Stevens, D. Tytler, B. G. Keating et al., Improved constraints on anisotropic birefringent Lorentz invariance and CPT violation from broadband optical polarimetry of high redshift galaxies, Phys. Rev. D 102 (2020) 043008, [2003.00647].DOI
  102. [102]R. Gerasimov, P. Bhoj and F. Kislat, New Constraints on Lorentz Invariance Violation from Combined Linear and Circular Optical Polarimetry of Extragalactic Sources, Symmetry 13 (2021) 880, [2104.00238].DOI
  103. [103]V. A. Kostelecky and M. Mewes, Sensitive polarimetric search for relativity violations in gamma-ray bursts, Phys. Rev. Lett. 97 (2006) 140401, [hep-ph/0607084].
  104. [104]V. A. Kostelecky and M. Mewes, Constraints on relativity violations from gamma-ray bursts, Phys. Rev. Lett. 110 (2013) 201601, [1301.5367].
  105. [105]Simons Observatory collaboration, M. Abitbol et al., The Simons Observatory: science goals and forecasts for the enhanced Large Aperture Telescope, JCAP 08 (2025) 034, [2503.00636].
  106. [106]Planck collaboration, N. Aghanim et al., Planck intermediate results. XLIX. Parity-violation constraints from polarization data, Astron. Astrophys. 596 (2016) A110, [1605.08833].DOI
  107. [107]POLARBEAR collaboration, K. Arnold et al., Millimeter-Wavelength Constraints on Circular Polarization of the Cosmic Microwave Background at Arcminute Scales with the Simons Array, 2611.13345.
  108. [108]G. Zagatti, M. Bortolami, A. Gruppuso, P. Natoli, L. Pagano and G. Fabbian, Planck constraints on cosmic birefringence and its cross-correlation with the CMB, JCAP 05 (2024) 034, [2401.11973].
  109. [109]V. Gluscevic, M. Kamionkowski and A. Cooray, De-Rotation of the Cosmic Microwave Background Polarization: Full-Sky Formalism, Phys. Rev. D 80 (2009) 023510, [0905.1687].
  110. [110]C. R. Harris et al., Array programming with NumPy, Nature 585 (2020) 357–362, [2006.10256].DOI
  111. [111]J. D. Hunter, Matplotlib: A 2D Graphics Environment, Comput. Sci. Eng. 9 (2007) 90–95.DOI
  112. [112]A. Lewis, GetDist: a Python package for analysing Monte Carlo samples, 1910.13970.

Paper details

vlnc:W_tjk3r543
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 1

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents