Split and Conquer : Extracting More Information from Galaxy Surveys and CMB by Aritra Kumar Gon

Split and Conquer : Extracting More Information from Galaxy Surveys and CMB by Aritra Kumar Gon

🎙 Aritra Kumar Gon 👥 74K 📅 September 10, 2026 ⏱ 30 min 👁 0 📄 original study 🧭 2026-09-10
Available in: English (current) Français

Keywords

density split clusteringredshift space distortionspeculiar velocitykSZ effectcosmological parameters

Summary

This talk by Aritra Kumar Gon, presented at the ICTS workshop ‘The Concurrence of Mega-Surveys’, introduces two methods to extract more cosmological information from large datasets. The first part focuses on density split clustering (DSC), which involves assigning each halo a local density estimate and splitting the sample into density bins. The speaker motivates this by noting that the peculiar velocity field probes structure growth and is better measured with large-scale correlations, but traditional two-point statistics discard small-scale information due to model limitations. DSC retains more information by using the density PDF and cross-correlating density-selected samples. However, previous DSC models were built in real space, requiring reconstruction and ignoring pair non-conservation. The speaker presents a new model that works entirely in redshift space: instead of random query points, they use halo centers, solving pair conservation. They model the density PDF via a log-normal transformation, derive density-split biases, and propose a generalized halo model for the monopole and quadrupole power spectra. The model fits simulations well up to moderate k. Fisher forecasts show that combining DSC bins with global clustering yields tighter constraints on (Ωm, σ8) due to different degeneracy directions, provided nuisance parameters are controlled. The second part applies a similar splitting idea to the pairwise kinetic Sunyaev-Zel’dovich (kSZ) signal. Since kSZ is extremely weak, statistical estimators like pairwise kSZ are used, but they average over all pairs, losing information about environment-dependent velocities. By splitting the density field and computing pairwise velocities in each bin, the speaker shows higher signal-to-noise and sensitivity to outflow and inflow regimes, despite larger variance. This suggests improved constraints on cosmological parameters. The talk concludes with ongoing work to incorporate kSZ directly and account for additional noise sources.

283 words

Critical Evaluation

Value of the Information & Strength of the Argument

The value of this talk is substantial: it addresses a key limitation in cosmological analysis—the loss of small-scale information—and proposes a novel, practical solution in density split clustering that works in redshift space without reconstruction. The argumentation is rigorous: the speaker carefully motivates each step, from the construction of the density PDF to the modeling of power spectra, and validates the model against simulations. The Fisher forecast demonstrates concrete gains when combining density-split bins with standard clustering, though the speaker honestly notes that nuisance parameters are currently treated as free or fixed. The kSZ part is similarly well-argued, showing that splitting by density can enhance sensitivity to different velocity regimes, even if the final constraints are not yet presented. The talk is technically deep and shows strong command of the subject, though some steps (like the one-island term in the halo model) are fitting functions without full physical understanding, which is acknowledged.

Scientific Rigor, Source Quality, Title Accuracy

The talk does not cite specific papers, but it references prior work on density split clustering and previous kSZ observations (e.g., 2012, 2016, and recent results) without explicit references. The description provides a link to the workshop program, which contains relevant context but not direct citations. The methodology appears based on standard cosmological perturbation theory and simulation techniques, and the speaker clearly states where approximations are made. The title perfectly matches the content: ‘Split and Conquer’ captures the splitting of density fields, and the dual application to galaxy surveys and CMB is accurately presented. There are no obvious discrepancies between the title and the material covered. Overall, while the lack of explicit source citations diminishes the verifiability, the rigor of the analysis and the credibility of the institution (ICTS) support a high reliability score.

301 words

Title / Content Match

The title accurately reflects the content: the talk presents a splitting technique to extract more information from galaxy surveys and CMB, applied to density-split clustering and pairwise kSZ.

Quality & Reliability

8/10

The talk presents original research with validated models and Fisher forecasts, but it is a conference presentation and some parts are preliminary (kSZ application). The methodology is sound and the speaker is expert, though no external sources are cited.

Key Moments

Cited Sources

Contribution & Novelties

The main contributions are: (1) a fully redshift-space density split clustering model that avoids reconstruction and solves pair conservation by using halo centers; (2) a physical model for the density PDF and bias, validated with simulations; (3) a demonstration that combining density-split bins with standard clustering improves Fisher constraints on cosmological parameters due to rotated degeneracies; (4) an extension of the splitting principle to pairwise kSZ velocities, showing enhanced signal-to-noise despite larger variance. These advances promise to unlock more information from current and future surveys like DESI, Euclid, and Rubin.

Pour aller plus loin :

153 words

Radar Profile

The radar profile shows high scores across information quantity, quality, and technical level, reflecting a dense and advanced presentation. Fiabilité is slightly lower due to lack of explicit citations and preliminary results in the kSZ part, but still solid.

Reliability 7/10