Clustering Dipole in Observational Cosmology

• Clustering dipole is defined as the flux-weighted vector sum of galaxy observations, serving as a proxy for the Local Group’s gravitational acceleration.
• The methodology accounts for survey windows and effective distances, highlighting how flux-limited selection influences the apparent convergence.
• Analyses using the clustering dipole help constrain key cosmological parameters like β and Ωₘ by comparing observed growth with ΛCDM predictions.

In observational cosmology, the clustering dipole is the dipole moment of the angular distribution of galaxies in an all-sky catalog, usually constructed as a flux-weighted vector sum over sources. Because both the gravitational acceleration induced by surrounding matter and the observed flux from a galaxy scale as $r^{-2}$, the flux dipole provides an observational proxy for the gravitational acceleration of the Local Group (LG) generated by the nearby large-scale structure. In the 2MASS literature, the subject became a focal test of whether the LG acceleration inferred from galaxy light converges within the survey depth, and of how the inferred acceleration can be compared to the CMB dipole velocity to constrain $\beta \equiv \Omega_m^{0.55}/b$ (Chodorowski et al., 2010).

1. Definition and linear-theory framework

For a galaxy of luminosity $L_i$ at distance $r_i$, the observed flux is

$S_i=\frac{L_i}{4\pi r_i^2}.$

The clustering dipole is then formed from the discrete sum of flux vectors,

$\mathbf{D}=\sum_i S_i\,\hat{\mathbf r}_i,$

or, in the velocity-scaled normalization used in the 2MASS analyses,

$$\tilde{\mathbf g}\equiv \frac{H_0}{j}\sum_i S_i\,\hat{\mathbf r}_i, \qquad \mathbf v=\beta\,\tilde{\mathbf g},$$

where $j$ is the luminosity density in the survey band and $\beta=f(\Omega_m)/b \simeq \Omega_m^{0.55}/b$ (Bilicki et al., 2011). In this construction, the flux-weighted dipole is a proxy for the LG gravitational acceleration up to the mean mass-to-light ratio and linear bias.

The velocity used for comparison is the LG peculiar velocity inferred from the CMB dipole. In the convention adopted in the 2MASS studies, this is $v_{\rm CMB}=622\pm35\ {\rm km\,s^{-1}}$ toward $\beta \equiv \Omega_m^{0.55}/b$0 (Bilicki et al., 2011). The central inference problem is therefore not simply measuring a dipole amplitude, but relating a filtered estimate of $\beta \equiv \Omega_m^{0.55}/b$1 to a measured $\beta \equiv \Omega_m^{0.55}/b$2 in a way that is consistent with linear theory, survey selection, and cosmic variance.

2. Observational window and the meaning of convergence

The convergence question is inseparable from the survey window. In a distance-limited or volume-limited redshift survey, the natural real-space window is a top-hat,

$\beta \equiv \Omega_m^{0.55}/b$3

where all matter within radius $\beta \equiv \Omega_m^{0.55}/b$4 is weighted equally up to geometry. In a flux-limited angular catalog such as the 2MASS XSC, the relevant window is the flux-weighted selection function,

$\beta \equiv \Omega_m^{0.55}/b$5

which decays smoothly with distance and does not correspond to a sharp radius cut (Bilicki et al., 2011).

This distinction propagates directly into theoretical predictions, because filtered velocities and accelerations depend on the power spectrum weighted by the survey window. In Fourier space, the relevant filtered moment is schematically

$\beta \equiv \Omega_m^{0.55}/b$6

with $\beta \equiv \Omega_m^{0.55}/b$7 determined by the adopted observational window (Chodorowski et al., 2010). The 2MASS flux-weighted window passes less large-scale power than a top-hat of the same nominal depth, so the predicted conditional dipole growth is slower. This is the technical reason the 2MASS dipole can appear not to converge even when the observations remain compatible with $\beta \equiv \Omega_m^{0.55}/b$8CDM.

A recurring misconception in the literature is to identify a nominal distance limit with the actual scales contributing to the dipole estimator. The 2MASS analyses showed that a flux-limited angular catalog gives significant weight to numerous faint galaxies at large effective distances, and that clusters contribute through many faint members that are assigned to larger effective distances than their true radial positions when no redshifts are available (Chodorowski et al., 2010). In that sense, convergence is an estimator-dependent statement, not a purely geometric one.

3. Measurement with the 2MASS Extended Source Catalog

The principal empirical case study is the 2MASS Extended Source Catalog (XSC), a near-infrared all-sky survey in the $\beta \equiv \Omega_m^{0.55}/b$9, $L_i$0, and $L_i$1 bands. The XSC contains $L_i$2 million extended sources, $L_i$3 of them galaxies, and is complete to $L_i$4 mag ($L_i$5 mJy) for resolved diameters $L_i$6–$L_i$7 arcsec (Bilicki et al., 2011). Near-infrared photometry was used because it traces old stars and is minimally affected by Galactic extinction.

In the 2MASS dipole analyses, fluxes were corrected for extinction, the Zone of Avoidance was masked, and the masked region was filled by cloning adjacent latitude strips. Bright large galaxies missing from the XSC, including Maffei 1/2 and Circinus, were added from the 2MASS Large Galaxy Atlas with tailored extinction corrections. Galactic contamination was reduced with the color cut $L_i$8, and the extinction correction used SFD maps with $L_i$9 (Bilicki et al., 2011).

The dipole was accumulated cumulatively by increasing the limiting magnitude $r_i$0, equivalently lowering the flux threshold $r_i$1. To present the result on an approximately linear depth axis, the survey used an effective-distance estimator derived from the $r_i$2-band luminosity function,

$r_i$3

so that the cumulative dipole growth could be plotted against effective depth rather than against raw source counts (Bilicki et al., 2011). This reparameterization was methodologically important, because plotting the dipole against the number of galaxies can make the amplitude appear to stabilize even when the effective survey depth is still increasing.

4. The 2MASS convergence controversy

The central empirical finding of the 2MASS studies is that the clustering dipole does not converge before the XSC completeness limit. When the cumulative dipole amplitude is plotted against effective distance, it continues to grow up to $r_i$4 mag, corresponding to an effective depth of about $r_i$5 in the 2010 analysis and $r_i$6 Mpc in the 2011 presentation (Chodorowski et al., 2010). The 2011 paper further noted that the Galactic $r_i$7-component continues to grow beyond $r_i$8 Mpc, whereas the $r_i$9 and $S_i=\frac{L_i}{4\pi r_i^2}.$0 components are nearly flat, indicating anisotropic contributions from intermediate scales (Bilicki et al., 2011).

This non-convergence did not, however, imply a failure of $S_i=\frac{L_i}{4\pi r_i^2}.$1CDM. Using the $S_i=\frac{L_i}{4\pi r_i^2}.$2CDM linear matter power spectrum with WMAP-constrained parameters and the conditional-growth formalism of Juszkiewicz et al. and Lahav et al., the 2MASS analyses found that the observed growth lies within the $S_i=\frac{L_i}{4\pi r_i^2}.$3 confidence band of the theoretical prediction once the correct 2MASS flux-limited window is used (Chodorowski et al., 2010). By contrast, if the same observations are compared with a top-hat prediction, the theoretical curve grows substantially faster and converges much earlier, producing an apparent disagreement.

The phrase “only apparent” became central to the interpretation. In the 2MASS window, growth beyond $S_i=\frac{L_i}{4\pi r_i^2}.$4 reflects the weighting of many faint galaxies at large effective distances rather than genuinely new distant structures dominating the LG acceleration (Bilicki et al., 2011). The same analyses also emphasized that at $S_i=\frac{L_i}{4\pi r_i^2}.$5, the mean depth of 2MRS, the true dipole is expected to have reached only $S_i=\frac{L_i}{4\pi r_i^2}.$6 of its final value in $S_i=\frac{L_i}{4\pi r_i^2}.$7CDM, so claims of convergence at $S_i=\frac{L_i}{4\pi r_i^2}.$8 are strongly window-dependent (Bilicki et al., 2011).

Directional information was secondary in these papers, but not absent. The 2011 2MASS study reported a typical misalignment of $S_i=\frac{L_i}{4\pi r_i^2}.$9 between the 2MASS clustering dipole and the CMB dipole direction for $\mathbf{D}=\sum_i S_i\,\hat{\mathbf r}_i,$0 Mpc, with a minimum $\mathbf{D}=\sum_i S_i\,\hat{\mathbf r}_i,$1 near $\mathbf{D}=\sum_i S_i\,\hat{\mathbf r}_i,$2 Mpc; bright nearby galaxies such as Maffei 1/2 and Circinus change the misalignment by $\mathbf{D}=\sum_i S_i\,\hat{\mathbf r}_i,$3, illustrating the sensitivity of the direction to a small number of high-flux sources (Bilicki et al., 2011).

5. Determination of $\mathbf{D}=\sum_i S_i\,\hat{\mathbf r}_i,$4 and $\mathbf{D}=\sum_i S_i\,\hat{\mathbf r}_i,$5

Once the observed growth is shown to be consistent with the conditional $\mathbf{D}=\sum_i S_i\,\hat{\mathbf r}_i,$6CDM prediction for the correct survey window, the dipole becomes a dynamical estimator of $\mathbf{D}=\sum_i S_i\,\hat{\mathbf r}_i,$7. In the 2011 2MASS analysis, fitting the observed $\mathbf{D}=\sum_i S_i\,\hat{\mathbf r}_i,$8 to the conditional velocity curve yielded

$\mathbf{D}=\sum_i S_i\,\hat{\mathbf r}_i,$9

and, with $$\tilde{\mathbf g}\equiv \frac{H_0}{j}\sum_i S_i\,\hat{\mathbf r}_i, \qquad \mathbf v=\beta\,\tilde{\mathbf g},$$0, implied

$$\tilde{\mathbf g}\equiv \frac{H_0}{j}\sum_i S_i\,\hat{\mathbf r}_i, \qquad \mathbf v=\beta\,\tilde{\mathbf g},$$1

(Bilicki et al., 2011). The earlier 2010 treatment quoted $$\tilde{\mathbf g}\equiv \frac{H_0}{j}\sum_i S_i\,\hat{\mathbf r}_i, \qquad \mathbf v=\beta\,\tilde{\mathbf g},$$2 and a rough estimate $$\tilde{\mathbf g}\equiv \frac{H_0}{j}\sum_i S_i\,\hat{\mathbf r}_i, \qquad \mathbf v=\beta\,\tilde{\mathbf g},$$3 (Chodorowski et al., 2010).

A later analysis presented two distinct estimators using the same 2MASS XSC data. The first was the growth-curve method, which compared the cumulative dipole $$\tilde{\mathbf g}\equiv \frac{H_0}{j}\sum_i S_i\,\hat{\mathbf r}_i, \qquad \mathbf v=\beta\,\tilde{\mathbf g},$$4 to the $$\tilde{\mathbf g}\equiv \frac{H_0}{j}\sum_i S_i\,\hat{\mathbf r}_i, \qquad \mathbf v=\beta\,\tilde{\mathbf g},$$5CDM prediction with the proper 2MASS window and again obtained $$\tilde{\mathbf g}\equiv \frac{H_0}{j}\sum_i S_i\,\hat{\mathbf r}_i, \qquad \mathbf v=\beta\,\tilde{\mathbf g},$$6. The second was a maximum-likelihood method that optimized the measurement window by excluding the brightest nearby galaxies, thereby suppressing nonlinear contributions and shot noise, and incorporated the nonlinear power spectrum of the velocity divergence and the gravity–velocity coherence function calibrated with $$\tilde{\mathbf g}\equiv \frac{H_0}{j}\sum_i S_i\,\hat{\mathbf r}_i, \qquad \mathbf v=\beta\,\tilde{\mathbf g},$$7CDM N-body simulations. That method gave a preliminary

$$\tilde{\mathbf g}\equiv \frac{H_0}{j}\sum_i S_i\,\hat{\mathbf r}_i, \qquad \mathbf v=\beta\,\tilde{\mathbf g},$$8

described there as the tightest clustering-dipole-based estimate to date (Bilicki et al., 2012).

The significance of these results is methodological as much as numerical. In the 2MASS framework, $$\tilde{\mathbf g}\equiv \frac{H_0}{j}\sum_i S_i\,\hat{\mathbf r}_i, \qquad \mathbf v=\beta\,\tilde{\mathbf g},$$9 is not inferred from a single asymptotic dipole amplitude, but from the agreement between the observed growth of a flux-weighted acceleration proxy and a conditional theoretical growth curve filtered by the actual survey window. The numerical stability of $j$0 near $j$1 across several analyses is therefore tied to the robustness of the window treatment rather than to a claim of early geometric convergence.

6. Systematics, later applications, and broader dipole literature

The most explicit 2MASS-specific systematic studied in this context is the Local Void. If the Zone of Avoidance is randomly filled, the absence of real galaxies in the intersection of the Local Void with the mask creates a spurious acceleration component. Under the simplest assumption of a spherical, completely empty Local Void, the induced spurious velocity is about $j$2; the resulting change in the misalignment angle is smaller than $j$3, and the fractional change in the deduced $j$4 is about $j$5. For an elongated empty Local Void, the corresponding numbers rise to about $j$6 and about $j$7, so the effect is systematic but minor relative to the larger uncertainty budget (Bilicki et al., 2010).

Later dipole studies extended the terminology beyond the 2MASS flux-dipole problem to number-count dipoles in deeper catalogs, where the measured dipole is decomposed into kinematic, clustering, and shot-noise components. In AllWISE galaxies, progressively suppressing low-redshift structure caused the dipole direction to converge to within $j$8–$j$9 of the CMB dipole, and subtraction of a modeled residual clustering dipole left $\beta=f(\Omega_m)/b \simeq \Omega_m^{0.55}/b$0, corresponding to a velocity of $\beta=f(\Omega_m)/b \simeq \Omega_m^{0.55}/b$1, consistent with the CMB (Rameez et al., 2017). In NVSS, a $\beta=f(\Omega_m)/b \simeq \Omega_m^{0.55}/b$2CDM model including kinematic, clustering, and shot-noise contributions found most previous dipole measurements consistent with the CMB at better than $\beta=f(\Omega_m)/b \simeq \Omega_m^{0.55}/b$3 (Cheng et al., 2023). By contrast, in the CatWISE2020 quasar catalog, a reassessment including clustering dipole, shot noise, and survey-mask mode coupling reduced the reported excess from $\beta=f(\Omega_m)/b \simeq \Omega_m^{0.55}/b$4 to $\beta=f(\Omega_m)/b \simeq \Omega_m^{0.55}/b$5 without a clustering dipole, $\beta=f(\Omega_m)/b \simeq \Omega_m^{0.55}/b$6 with a randomly oriented clustering dipole, and $\beta=f(\Omega_m)/b \simeq \Omega_m^{0.55}/b$7 when the clustering dipole is aligned with the kinematic direction, but did not remove the anomaly (Bashir et al., 2 Nov 2025).

These later applications do not replace the original 2MASS usage; they broaden it. In the 2MASS program, the clustering dipole is primarily a flux-weighted proxy for LG acceleration and a tool for estimating $\beta=f(\Omega_m)/b \simeq \Omega_m^{0.55}/b$8. In deeper number-count surveys, the same phrase often denotes the intrinsic large-scale-structure contribution to the observed dipole, which must be modeled and subtracted to isolate the Ellis–Baldwin kinematic signal. The common lesson is the same in both settings: dipole measurements are only interpretable when the survey window, mask-induced mode coupling, shot noise, and local structure are treated explicitly rather than implicitly.