KSZ Velocity Reconstruction with ACT and DESI-LS using a Tomographic QML Power Spectrum Estimator

Introduction to the show: ident: Astrophysics Radio. Generated commentary on the latest astrophysics papers.

Vera: Today's paper: "KSZ Velocity Reconstruction with ACT and DESI-LS using a Tomographic QML Power Spectrum Estimator".

Jocelyn: The gist We perform kinetic Sunyaev-Zel’dovich (kSZ) velocity reconstruction on data from ACT DR6 and DESI-LS DR9 to estimate the cross-power between kSZ velocity reconstruction and galaxy density,

Vera: First, who's behind it and why it matters.

Title and authors: Vera: So let’s talk about the paper "KSZ Velocity Reconstruction with ACT and DESI-LS using a Tomographic QML Power Spectrum Estimator." We want to explain what this title actually means for someone listening who isn't an astronomer.

Jocelyn: Basically, they are taking temperature data from ACT and galaxy data from DESI and combining them in a clever statistical way to figure out the speed of the gas and how it relates to where the galaxies are.

Subrahmanyan: The "Tomographic QML Power Spectrum Estimator" part is just describing a very specific, advanced mathematical tool they developed for that combination. It’s their unique recipe for getting the numbers out of all that data.

Vera: So, it’s not just measuring one thing; it’s using a technique designed to handle both the temperature fluctuations and the galaxy distributions simultaneously in a tomographic way, which is important because we have different kinds of data here.

Jocelyn: It means they are looking at how these two different things influence each other across the sky, which is exactly what you need when you want to measure that cross-correlation between gas velocity and galaxy density.

Subrahmanyan: If you think about it simply, they’ve created a sophisticated way to look at the data that lets them disentangle the signal from the noise in a way that isn't just standard cross-correlation.

Vera: It’s about taking complex, multi-layered observations and using this specific QML approach to extract a velocity field measurement.

Jocelyn: So when you hear it, think of it as building a bridge between the observed light and the actual motion of the stuff in between.

Subrahmanyan: And that bridge needs to be strong enough to let us see how much gas is actually moving relative to where the galaxies are clustered.

The paper's summary: Vera: Now we look at the core summary of "KSZ Velocity Reconstruction with ACT and DESI-LS using a Tomographic QML Power Spectrum Estimator." They explain what they actually did in terms of their pipeline.

Jocelyn: They start by setting up the whole framework, detailing how they relate the true radial velocity field to the observed galaxy overdensity through a linear model, vtrue r equals M gobs.

Subrahmanyan: That linear model is crucial because it’s how they get from what you see in galaxies to what the actual motion should be, setting up the relationship between position and velocity.

Vera: Then they build this joint probability distribution P(v true r, δobs g) using a large covariant matrix C defined by C equal to Cvrvr Cvrg Cgg, which is the core statistical engine for their reconstruction.

Jocelyn: That matrix structure is the statistical foundation; it’s what allows them to combine the galaxy information and the velocity information together correctly.

Subrahmanyan: They then achieve the actual reconstruction of the radial velocity field by minimizing an expression called Eq. eighty-eight with respect to vtrue r, which yields vtrue equals Cvrg(Cgg)−one g obs. That’s how they solve for the true velocity field using the galaxy overdensity maps.

Vera: They also build a linear estimator for the velocity field from galaxies by minimizing the variance of (vˆi,g − vi)(ˆvi,g − vi) T with respect to Mij, which results in M equal to Cvg(Cgg)−one. That’s the tool they use to get that final galaxy velocity estimate.

Jocelyn: This pixel-space estimator is quite compatible with sky masks in arbitrary shapes, which is a practical point for real data analysis on the sky. They also create coadded velocity maps by a linear combination of individual velocity maps from the same redshift bins and different frequency channels.

Subrahmanyan: They do this to get these coadded velocity maps, valpha coadd equals Avalpha90 + Bvalpha150, and they determine the noise covariance Nvv coadd in pixel space using Eq. fourteen <ref:2506.21684#pg2>.

Vera: Then they apply the quadratic maximum likelihood QML power spectrum estimator to the partial-sky velocity and galaxy overdensity data to reconstruct the full-sky velocity-galaxy cross-power C vgalpha gβl and compare them to fiducial calculations.

Jocelyn: It’s like they’re stitching together a high-resolution map of the temperature fluctuations with a map of galaxy positions to see how they interact across the sky, and the methodology is quite involved, but it seems like it’s giving them the precision they need for this kind of measurement.

Subrahmanyan: And that precision is what matters when you start talking about things like velocity bias, which is essentially how much the galaxy moves compared to the underlying gravitational pull from dark matter. They estimate this bias to be around zero point four, which is a specific number we can use for modeling.

The paper's improvements: Vera: Now let’s talk about the suggestions they make for improving this work in "KSZ Velocity Reconstruction with ACT and DESI-LS using a Tomographic QML Power Spectrum Estimator." They propose several specific tweaks to make the method even better.

Jocelyn: One big idea is implementing a "novel quadratic maximum likelihood QML power spectrum estimator implementation in red-shift binned spherical coordinates" to get an SNR that almost doubles the signal-to-noise compared to the more standard pseudo-Cl approach.

Subrahmanyan: That level of precision, if they actually achieve it, would allow them to perform kSZ velocity reconstruction with significantly higher precision, reaching an SNR of eleven point seven sigma and providing a powerful tool for kSZ cosmology <ref:2506.21684#pg1>.

Vera: Another key suggestion is developing that linear estimator for the velocity field from galaxies using the construction that minimizes variance by solving for the optimal matrix M, allowing them to reconstruct large-scale velocity fields directly from galaxy over-density maps before they compare them with their kSZ results using some correlation coefficient R l.

Jocelyn: That sounds like they are building a dual approach, so they aren't just relying on one reconstruction method; they’re building a comparison tool between the kSZ derived velocity and this galaxy-derived velocity.

Subrahmanyan: That comparison is important because it allows them to look at how biased the galaxy velocities are relative to the true underlying motions, which helps constrain the physics of structure formation in those halos.

Vera: They also introduce a correction factor for when they calculate the Fisher matrix for parameter estimation; they suggest using an expression that involves b inv, rvv, and Nvv. That lets them robustly estimate the SNR even when there’s some mismodeling happening in the electron-galaxy cross-power.

Jocelyn: It shows they're thinking about how to keep their error estimates honest, acknowledging that the noise environment can affect how much signal you actually get out of a measurement.

Subrahmanyan: That level of self-correction is necessary when you’re trying to extract cosmological parameters from noisy astrophysical data; it means the analysis is built to handle uncertainties in the input models.

Vera: They also suggest extending this whole pipeline for future work to include cosmological parameter inference, specifically looking at primordial non-Gaussianity, or fNL. That makes sense because their framework is already pretty good at mapping out these velocity fields.

Jocelyn: So they aren't just stopping at finding a signal; they’re showing that the architecture they built can be used to test even deeper cosmological physics than just the basic kSZ effect.

Subrahmanyan: And if they can successfully apply this framework to constrain parameters like fNL, it opens up a whole new door for testing models of inflation and primordial fluctuations using these large-scale structure observations.

Conclusion: Vera: So we’ve just gone through how they used ACT and DESI data to find that eleven point seven sigma detection in the paper "KSZ Velocity Reconstruction with ACT and DESI-LS using a Tomographic QML Power Spectrum Estimator." We’re wrapping up the discussion on this paper.

Jocelyn: And the big win here is that they built a method that actually works well for this cross-correlation, showing a detection at an eleven point seven sigma level. That's pretty high when you think about how tricky it is to measure those tiny bulk motions in the universe.

Subrahmanyan: From my side, what this really means is that we're getting a way to connect the gas dynamics—the movement of electrons causing the kSZ signal—directly to the large-scale distribution of matter traced by galaxies. It’s a direct probe into how dark matter halos are actually moving around.

Vera: Exactly, and they use this tomographic QML power spectrum estimator, which is what makes it so robust, to do this cross-correlation analysis. They're not just looking at one piece of data; they're combining different frequency channels from ACT and DESI.

Jocelyn: It really shows how these advanced statistical methods can help us move beyond just detecting the kSZ effect in isolation and start measuring things like velocity bias about zero point four with this kind of analysis.

Subrahmanyan: This paper’s result, combined with their proposed improvements, makes it a powerful tool for kSZ cosmology because it allows us to test models of how feedback operates in massive halos.

Vera: It certainly gives us a solid foundation for using this kind of reconstruction method as we look at the next set of experiments and data sets coming down the pipeline.

Jocelyn: We're ready to see what this means for future constraints on cosmological models as we keep pushing these detection limits higher.

Subrahmanyan: And I think this work, with its focus on robust estimators and comparison between reconstruction methods, provides a very useful blueprint for how we can tackle more complex velocity field mapping problems across different observational datasets.

Vera: We’re done with this paper and ready to look at what else is out there in the arXiv right now.

Jocelyn: And we’re ready for whatever comes next on our show.