Hemispherical Power Asymmetry (HPA)

• Hemispherical Power Asymmetry is a cosmic anomaly marked by a 7% dipolar excess in CMB power between hemispheres, violating statistical isotropy.
• Advanced local-variance estimators and dipolar modulation analyses from WMAP and Planck robustly characterize the anomaly’s scale dependence and preferred galactic direction.
• The observed asymmetry challenges ΛCDM, prompting theoretical models like modulated inflation and pre-inflationary inhomogeneities to explain its cosmological implications.

Hemispherical Power Asymmetry (HPA) is a violation of statistical isotropy in the temperature and polarization fluctuations of the cosmic microwave background (CMB), characterized by a significant excess of power in one half of the sky compared to the opposite hemisphere. First identified in WMAP data and subsequently confirmed by Planck, this anomaly manifests as a dipolar modulation of the CMB two-point function on large angular scales, with an observed amplitude of approximately 7% and a preferred direction in galactic coordinates around $(\ell, b) \approx (224^\circ, -22^\circ)$. The persistence, scale dependence, and cosmological robustness of HPA in both CMB intensity and (more tentatively) polarization challenge the standard ΛCDM expectation of statistical isotropy and motivate a broad spectrum of theoretical and phenomenological investigations.

1. Phenomenology and Statistical Characterization

The basic phenomenological description of HPA is a modulated temperature field,

$$\Delta T_{\rm obs}(\hat n) = [1 + A\,(\hat p\cdot\hat n)]\,\Delta T_{\rm iso}(\hat n),$$

where $A$ is the dipole modulation amplitude ($A \sim 0.07$ at large scales), $\hat p$ is the preferred axis, and $\Delta T_{\rm iso}$ is a statistically isotropic realization. The modulation induces a hemispherical gradient in the local power spectrum and can equivalently be cast as a dipole in real-space local-variance maps or as off-diagonal correlations between $a_{\ell m}$ coefficients in harmonic space, particularly between $\ell$ and $\ell \pm 1$ multipoles. The same model applies (modulo noise and masking) to the E-mode polarization field, although the significance in current polarization data is lower due to residual systematics and higher noise (Gimeno-Amo et al., 2023, Aluri et al., 2017).

Local-variance estimators (LVE) have played a central role in quantifying HPA. In this approach, the variance of the temperature field is computed over a tessellation of sky patches (discs of radius $R$), producing maps whose dipole component is proportional to $A$ and whose preferred direction yields the HPA axis. Large-sky simulations are used to calibrate the estimator and assess significance, with Planck PR3/PR4 and WMAP data showing consistent results: a significant dipole at disc radii $R \sim 4^\circ$–$14^\circ$, corresponding to the multipole range $\ell \sim 10$–$45$ (Sanyal et al., 20 Jan 2026, Sanyal et al., 2024, Akrami et al., 2014).

Dataset	$A$ (approx.)	Dipole direction $(\ell, b)$	Significance ($p$)
Planck PR4 Temperature	0.07	$(215^\circ,-30^\circ)$	$<0.2\%$
WMAP 9 Temperature	0.07	$(219^\circ,-24^\circ)$	$0.5$–$2\%$
Planck PR4 E-mode	0.09	$(234^\circ,-14^\circ)$	$2.8\%$ (SEVEM)

The amplitude and direction show negligible frequency dependence among Planck and WMAP frequency bands, and persist under alternative masking, filtering, and data-processing pipelines, arguing against foreground or instrumental artifacts (Sanyal et al., 20 Jan 2026).

2. Scale Dependence and Observational Constraints

Observational evidence overwhelmingly indicates that HPA is a large-scale-only phenomenon. The modulation amplitude $A(\ell)$ is scale-dependent, peaking at $A \sim 0.07$ for $\ell \lesssim 64$, and dropping sharply at higher multipoles. At $\ell \gtrsim 600$ (corresponding to physical scales $\lesssim 10$ Mpc), Planck and quasar clustering data constrain $A(k) < 0.0045$ at 95% C.L., ruling out significant HPA on small scales (Flender et al., 2013, Sanyal et al., 2024, Li et al., 2019).

Parameterizations of the scale dependence include a power-law: $$A(\ell) = A_0 \left( \frac{\ell_0}{\ell} \right)^{n},$$ with best-fit values $A_0 \sim 0.086 \pm 0.009$ at $\ell_0 = 10$ and $n = 0.31 \pm 0.16$ from PR4 and WMAP multi-frequency analyses (Sanyal et al., 20 Jan 2026).

Open questions remain regarding the precise functional form of $A(\ell)$, but all viable physical models must reconcile CMB large-scale HPA with null or extremely tight bounds on quasar and small-scale CMB power asymmetry (Flender et al., 2013, Li et al., 2019).

3. Physical Mechanisms and Inflationary Model Building

Multiple early-universe and inflationary scenarios have been proposed to explain the observed HPA:

• Pre-inflationary topological defects: Spontaneous breaking of a global U(1) symmetry prior to inflation can yield long-wavelength phase gradients (cosmic strings). If such a defect is present near our observable patch, the phase field imprints a large-scale dipole in the post-inflationary pseudo-Goldstone boson $\sigma$, leading to a scale-dependent power asymmetry after boson oscillation and decay. The modulation amplitude is predicted to fall as $A(k) \sim e^{-k l}/k$ (where $l$ is the comoving distance to the string), fitting both the large-scale amplitude and the small-scale constraints (Yang et al., 2016, Kohri et al., 2013).
• Modulation of inflationary sound speed: Allowing the sound speed $c_s$ of the inflaton to carry a super-Hubble perturbation (e.g., from a light spectator field), the $\delta N$-formalism predicts a locally modulated curvature power spectrum. The resultant amplitude $A(k)$ is scale-dependent and sensitive to primordial non-Gaussianity constraints (Wang et al., 2015, Cai et al., 2013).
• Modulated reheating: A subdominant, scale-dependent contribution from modulated reheating seeded by a tachyonically-growing field with intrinsic hemispherical variation naturally produces a red-tilted HPA, satisfying non-Gaussianity and power spectrum bounds (McDonald, 2013).
• Pre-inflationary metric inhomogeneity: A localized, long-wavelength deviation from initial FLRW homogeneity perturbs the in-in evolution of adiabatic perturbations, yielding a directional modulation in the primordial power spectrum. Off-diagonal correlations between multipoles $\ell$ and $\ell\pm 1$ provide a direct test (Gandhi et al., 29 Sep 2025).
• Tensor modulation: Spatial variation of the tensor-to-scalar ratio $r$ leads to polarization and temperature power asymmetry but cannot reproduce the full observed temperature HPA amplitude unless $r$ is implausibly large (Chluba et al., 2014).

Phenomenological models, such as the two-component adiabatic power spectrum or generic modulation field models, provide a flexible language for analyzing HPA data but are not tied to a specific high-energy mechanism (McDonald, 2014, Li et al., 2019).

4. Implications for Cosmological Parameters and Statistical Isotropy

The presence of HPA implies a preferred cosmic direction, violating one of the pillars of the Cosmological Principle. Robust evidence against foregrounds or local systematics includes its appearance in both WMAP and Planck, across cleaning methods and frequency bands, and alignment persisting under varied masking strategies (Sanyal et al., 2024, Akrami et al., 2014).

HPA can bias estimates of cosmological parameters in the two hemispheres. For instance, in WMAP9, $n_s$ differed significantly between high- and low-power hemispheres ($n_s = 0.989 \pm 0.024$ vs. $n_s = 0.959 \pm 0.022$), although global parameter estimates remain robust due to cosmic variance (Axelsson et al., 2013). Dipolar power modulation generically produces off-diagonal $a_{\ell m}$ covariance and a small but potentially measurable shift in $n_s$ and its running between hemispheres (McDonald, 2014).

Table: Effect of HPA on selected $\Lambda$CDM parameters (WMAP9)

Hemisphere	$n_s$	$\ln(10^{10}A_s)$	$\Omega_b h^2$
High-power	$0.989 \pm 0.024$	$3.109 \pm 0.075$	$0.0227 \pm 0.0006$
Low-power	$0.959 \pm 0.022$	$3.036 \pm 0.073$	$0.0219 \pm 0.0006$

The null detection of HPA in small-scale data imposes stringent constraints on inflationary models; any viable explanation must guarantee a sharp drop in $A(k)$ above $k\sim0.04$ Mpc$^{-1}$ (Flender et al., 2013). Observed alignments of the HPA axis with neither the ecliptic nor with systematic directions further support a cosmological origin.

5. Extensions to Polarization and Large-Scale Structure

HPA analyses have been extended to CMB E-mode polarization, with PR3/PR4 Planck data suggesting a marginally significant ($p \sim 1$–$3\%$) large-scale E-mode dipole, with direction broadly consistent with the temperature HPA, but with some preference for the kinetic dipole axis in Commander maps (Gimeno-Amo et al., 2023, Aluri et al., 2017). The amplitude in Planck PR4 is $A_{DM} \approx 9\%$, higher than in temperature but within error bars and systematics-dominated uncertainty. The T-E axes show a mutual alignment at the $\sim 5\%$ level.

21-cm and galaxy redshift surveys are emerging as tools to probe HPA at intermediate and small scales. The Square Kilometre Array (SKA) and DESI are forecasted to detect HPA-type dipoles at $A > 0.04$ at $>2\sigma$ significance, assuming a true cosmic signal of this magnitude (Li et al., 2019, Zhai et al., 2017). These measurements will bridge the scale gap between CMB and large-scale structure, critically testing the scale dependence of $A(k)$.

6. Current Status and Prospects

Consensus has solidified around the existence of a statistically significant ($\sim 3\sigma$) hemispherical power asymmetry in the CMB temperature sky at large angular scales, with all currently viable models requiring a rapid falloff in its amplitude at small scales, in accordance with quasar and high-$\ell$ CMB bounds. The anomaly remains unexplained within the minimal ΛCDM or single-field inflation, motivating models involving super-horizon modulations, multi-field dynamics, topological defects, or pre-inflationary inhomogeneities (Yang et al., 2016, Gandhi et al., 29 Sep 2025, Wang et al., 2015, Kohri et al., 2013, McDonald, 2013).

Polarization measurements with improved sensitivity and control of systematics (e.g., Simons Observatory, LiteBIRD, CMB-S4) and future high-precision large-scale structure data will allow decisive tests of HPA-related models. Detection of a consistent, scale-dependent dipole in E-mode polarization and matter clustering, aligned with the temperature HPA axis, would strongly support a primordial origin (Gimeno-Amo et al., 2023, Li et al., 2019). Conversely, absence of polarization or LSS anisotropies would argue for residual systematics or a statistical fluke.

7. Summary Table: Key Findings from Recent HPA Studies

Reference	Data/Method	Main Finding
(Sanyal et al., 2024)	Planck PR4 LVE	HPA confined to large scales; $A \sim 0.02$–0.03; $\gtrsim 3\sigma$
(Sanyal et al., 20 Jan 2026)	WMAP, Planck LVE	$A_0 \sim 0.09$, $n \sim 0.31$; robust to frequency, channel
(Gimeno-Amo et al., 2023)	Planck PR4, PR3	E-mode HPA amplitude $\sim 9\%$; $p = 2.8\%$; T-E axis alignment
(Aluri et al., 2017)	Planck 2015 Pol.	E-mode HPA $\sim 2.6$–$3.9\%$ at $p < 0.001$; axis near kinetic dip.
(Yang et al., 2016)	Defect+inflation	Topological defect + inflation explains scale-dependent HPA
(Wang et al., 2015)	Multi-speed $\delta N$	HPA via varying primordial sound-speed; scale-dependent $A(k)$
(Flender et al., 2013)	High-$\ell$ cap power	$A(k) < 0.0045$ for $k \gtrsim 0.04$ Mpc$^{-1}$; null small-scale HPA

Continued investigation of HPA offers a powerful probe of pre-inflationary physics, cosmic topology, and the validity of the cosmological principle. Its resolution will have profound implications for our understanding of the initial conditions and symmetries of the Universe.