Multi-tracers, multi-surveys: a joint Fisher analysis of DESI+PFS

arXiv:2604.25171 · astro-ph.CO · Submitted 2026-04-28 · Read on arXiv

Listen

Transcript

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

Vera: We're looking at the paper: "Multi-tracers, multi-surveys: a joint Fisher analysis of DESI+PFS".

Jocelyn: Here's the gist of the paper. The gist The proposed framework is a multi-tracer Fisher analysis that combines internal DESI cross-correlations with an external PFS×DESI overlap channel to provide model-independent constraints on cosmological parameters by marginalizing over nuisance parameters from data alone.

Vera: How it works The core methodology involves a volume-partitioned joint Fisher analysis that combines the internal DESI channel, which uses LRG × ELG × QSO cross-spectra over the full footprint, with the PFS×DESI overlap channel in the 1,200 deg2 intersection.

Jocelyn: This construction ensures that "the volume partition reflects tracer availability" and that "no information is double-counted" because Fourier modes in disjoint volumes are uncorrelated under the Gaussian-covariance approximation.

Vera: The final cosmological constraints are derived by combining the per-z-bin Fishers across all redshift bins with shared cosmological parameters (fσ8, Mν, omegam) and per-tracer, per-z-bin nuisance parameters.

Title and authors: Vera: We’re talking about "Multi-tracers, multi-surveys: a joint Fisher analysis of DESI+PFS" here, and what they’re demonstrating is that by using cross-spectra between different galaxy populations—like LRG mixed with ELG and QSO in the DESI part—and adding that overlap channel from the PFS survey, they can actually pin down those nuisance parameters much better than just looking at one tracer.

Jocelyn: Right. The paper really hammers home how this helps us handle those nasty systematic uncertainties, like bias and stochastic amplitudes, which usually require heavy reliance on simulation-based priors for things like neutrino masses.

Subrahmanyan: They propose using cross-spectra between galaxy populations to calibrate those EFT bias and stochastic parameters directly from the data itself, which is a big step toward model independence.

Vera: And they achieve this by partitioning the volume into distinct regions, so you get an internal DESI channel and then an external PFS times DESI overlap channel that doesn't double-count any information because those volumes are disjoint.

Jocelyn: The payoff is in the numbers too. They forecast across a fourteen thousand square degree area, and they show that this combined approach improves constraints on things like the growth rate of structure, f sigma8 by thirty-three percent with just the internal DESI channel alone <ref:2604.25171#pg2>.

Subrahmanyan: And then adding that PFS times DESI overlap channel brings those improvements up further—nine to twenty-four percent gains on sigma(M nu) depending on what you look at.

Vera: The authors point out that the biggest jump, twenty-four percent, is for the neutrino mass constraint because it picks up extra information where the PFS data adds new cross-spectra in lower redshift bins less than one point six.

Jocelyn: That’s a big deal for structure formation studies. If these data-driven constraints are right, it means we might not need to rely on HOD mocks as much when testing simulation predictions for how dark matter halos actually form.

Subrahmanyan: And they show that the calibration of the linear bias parameter b one sigma eight is actually the main driver here, tightening that prior from a flat value down to sigma approximately zero point one three, which helps break the degeneracy between b one sigma eight and f sigma8 <ref:2604.25171#pg2>.

Vera: So what this means for us is that we get an independent handle on whether those shifts in sigma eight coming from simulation priors actually reflect new physics or just issues with how we model the halos themselves.

Jocelyn: It really sets up a strong foundation for using these multi-tracer Fisher analyses as a way to check the consistency between what’s happening in simulations and what we’re seeing in the real sky.

Subrahmanyan: And they mention that this method is generalizable; you can partition any spectroscopic survey footprint into disjoint regions and build a multi-tracer Fisher from active tracers for each part, which means it could work for lots of different surveys.

The paper's summary: Vera: We just talked about how they set up this joint Fisher analysis combining DESI and PFS data to get constraints on cosmology, right?

Jocelyn: Yeah, and what they’re showing us is that by using these cross-spectra between different galaxy populations—like LRG mixed with ELG and QSO in the DESI part—and adding that overlap channel from the PFS survey, they can actually pin down those nuisance parameters much better than just looking at one tracer.

Subrahmanyan: They propose using cross-spectra between galaxy populations to calibrate those EFT bias and stochastic parameters directly from the data itself.

Vera: They’re doing this by partitioning the volume into distinct regions, so you get an internal DESI channel and then an external PFS times DESI overlap channel that doesn't double-count any information because those volumes are disjoint.

Jocelyn: The big payoff is in the numbers. They forecast across a fourteen thousand square degree area, and they show that this combined approach improves constraints on things like the growth rate of structure, f sigma8 by thirty-three percent with just the internal DESI channel alone <ref:2604.25171#pg2>.

Subrahmanyan: That’s what I mean. And adding that PFS times DESI overlap channel brings those improvements up further—nine to twenty-four percent gains on sigma(M nu) depending on what you look at.

Vera: They also showed that calibrating the linear bias parameter b one sigma eight is actually a major driver here, tightening that prior from a flat value down to sigma approximately zero point one three, which helps break the degeneracy between b one sigma eight and f sigma8 <ref:2604.25171#pg2>.

Jocelyn: So what this means for us is that we get an independent handle on whether those shifts in sigma eight coming from simulation priors actually reflect new physics or just issues with how we model the halos themselves <ref:2604.25171#pg2>.

Subrahmanyan: The next step is seeing if these results, derived from this model-independent framework, align with the predictions coming out of the next generation of large-scale structure simulations.

The paper's improvements: Vera: We just talked about how they set up this joint Fisher analysis combining DESI and PFS data to get constraints on cosmology, right?

Subrahmanyan: They propose using cross-spectra between galaxy populations to calibrate those EFT bias and stochastic parameters directly from the data itself.

Subrahmanyan: And then adding that PFS times DESI overlap channel brings those improvements up further—nine to twenty-four percent gains on sigma(M nu) depending on what you look at.

Vera: The authors highlight that the biggest jump, twenty-four percent, is for the neutrino mass constraint because it picks up extra information where the PFS data adds new cross-spectra in lower redshift bins less than one point six.

Jocelyn: They also showed that calibrating the linear bias parameter b one sigma eight is actually a major driver here, tightening that prior from a flat value down to sigma approximately zero point one three, which helps break the degeneracy between b one sigma eight and f sigma8 <ref:2604.25171#pg2>.

Subrahmanyan: So what this means for us is that we get an independent handle on whether those shifts in sigma eight coming from simulation priors actually reflect new physics or just issues with how we model the halos themselves.

Vera: It really sets up a strong foundation for using these multi-tracer Fisher analyses as a way to check the consistency between what’s happening in simulations and what’s happening in the real sky.

Jocelyn: It’s a powerful way to get superior constraints on cosmological parameters like M nu and f sigma8 when you combine internal survey data with external overlapping datasets <ref:2604.25171#pg2>.

Conclusion: Vera: So we've seen how they set up this joint Fisher analysis combining DESI and PFS data to get constraints on cosmology, right?

Subrahmanyan: They propose using cross-spectra between galaxy populations to calibrate those EFT bias and stochastic parameters directly from the data itself.

Subrahmanyan: And then adding that PFS times DESI overlap channel brings those improvements up further—nine to twenty-four percent gains on sigma(M nu) depending on what you look at.

Vera: The paper confirms consistency with numerical convergence tests and analytical limit checks, so the mathematical framework holding up under scrutiny is solid.

Jocelyn: It’s a powerful way to get superior constraints on cosmological parameters like M nu and f sigma8 when you combine internal survey data with external overlapping datasets.

Vera: So that’s our take on "Multi-tracers, multi-surveys: a joint Fisher analysis of DESI+PFS."

Jocelyn: It really shows how using cross-spectra between different galaxy populations can actually pin down those nuisance parameters much better than just looking at one tracer.

Subrahmanyan: They show that this framework provides superior constraints on cosmological parameters like M nu and f sigma8 when you combine internal survey data with external overlapping datasets.

Vera: I think the most important part is using observed cross-spectra to calibrate the bias parameters directly, instead of just relying on HOD mocks for those shifts.

Jocelyn: That gives us a way to check if those predictions from simulations are consistent with what we actually see in the cross-correlations across different tracers.

Subrahmanyan: It’s a powerful way to get superior constraints on cosmological parameters like M nu and f sigma8 when you combine internal survey data with external overlapping datasets.

DESI collaboration · Subaru Prime Focus Spectrograph

astro-ph.CO

Submitted: 2026-04-28

Updated: 2026-10-22

Comments: v3: 21+7 pages; revise model and code; fix bugs; update the main analyses and key results; add more analyses of prior+model sensitivity and information content. Code is available at: https://github.com/MinhMPA/PFSxDESI-multi-fish

Code: https://github.com/MinhMPA/PFSxDESI-multi-fish

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 90/100

The gist: The gist The proposed framework is a multi-tracer Fisher analysis that combines internal DESI cross-correlations with an external PFS×DESI overlap channel to provide model-independent constraints on

Key concepts

Fisher Analysis: A statistical method used to estimate the precision of cosmological parameters derived from observational data. It constructs a Fisher matrix that quantifies how well different parameters can be measured, allowing researchers to determine the expected uncertainty in cosmological values.
Nuisance Parameters: These are systematic uncertainties inherent in galaxy power spectrum analyses, such as bias coefficients and stochastic amplitudes, which complicate parameter estimation. The paper addresses these by using cross-spectra between independent tracers to calibrate and marginalize over them, reducing their impact on the final cosmological constraints.
Volume Partitioned Joint Fisher: This methodology divides the total observational volume into disjoint regions. A joint Fisher analysis is then performed across these partitions, ensuring that information from different volumes is combined without double-counting, which allows for a comprehensive constraint on cosmological parameters.
Model-Independent Constraints: These are cosmological constraints derived directly from observed cross-spectra between galaxy populations, rather than relying solely on theoretical models like Halo Occupation Distribution (HOD) mocks. This provides an independent check on simulation-based priors and helps distinguish between new physics and systematic errors.

Terminology

Calibration — 2× in this paper · explained in 9 episodes
Dark Energy Spectroscopic Instrument — 1× in this paper
Degeneracy — 1× in this paper · explained in 4 episodes
N-body simulations — 1× in this paper · explained in 2 episodes
Prime Focus Spectrograph — 1× in this paper

Summary

The gist The proposed framework is a multi-tracer Fisher analysis that combines internal DESI cross-correlations with an external PFS×DESI overlap channel to provide model-independent constraints on cosmological parameters by marginalizing over nuisance parameters from data alone.

How it works

The core methodology involves a volume-partitioned joint Fisher analysis that combines the internal DESI channel, which uses LRG × ELG × QSO cross-spectra over the full footprint, with the PFS×DESI overlap channel in the 1,200 deg2 intersection. This construction ensures that the volume partition reflects tracer availability and that no information is double-counted because Fourier modes in disjoint volumes are uncorrelated under the Gaussian-covariance approximation. The final cosmological constraints are derived by combining the per-z-bin Fishers across all redshift bins with shared cosmological parameters (fσ8, Mν, omegam) and per-tracer, per-z-bin nuisance parameters.

Nuisance Parameter Marginalization

A dominant systematic cost in full-shape galaxy power spectrum analyses is the marginalization over roughly 12 effective-field-theory (EFT) nuisance parameters per tracer per redshift bin which are summarized by bias coefficients, counterterms, and stochastic amplitudes. The paper proposes two channels to relax this prior:

The internal-DESI channel uses cross-spectra between independent populations to isolate bias and stochastic parameters from cross-spectra between independent populations. This exploits cosmic-variance cancellation in the bias ratios which is developed by Seljak [44] and McDonald & Seljak [41].
The PFS × DESI overlap channel adds cross-spectra from the PFS-ELG sample, contributing tracers with the lowest b1 in the joint analysis.

Cosmological Forecasts and Gains

The analysis forecasts across the 14,000 deg2 DESI footprint at kmax = 0.20 h Mpc−1. The internal-DESI channel alone improves constraints by improving σ(fσ8) by 33%, σ(Mν) by 80%, and σ(omegam) by 49% over a single-tracer broad-prior baseline. Adding the PFS × DESI overlap channel further tightens these constraints by 9%, 24%, and 9% respectively after marginalizing over residual cross-population stochasticity. The largest fractional gain is on σ(Mν) (24%) precisely because "the marginalized Mν constraint inherits the residual b1σ8 posterior width, and that posterior contracts most where PFS adds new cross-spectra (z < 1.6)".

Parameter Drivers and Robustness

The dominant driver for improvement is the calibration of the linear bias b1σ8, which is tightened from a flat prior to σ ≈ 0.13, breaking the b1σ8–fσ8 degeneracy. The analysis demonstrates that the cross-spectra between two tracers in the same volume see the same density field, so their amplitude ratio determines the bias ratio bX1/bY1 cosmic-variance-free in the high-density limit. Robustness tests show that constraints are insensitive to both [the FoG velocity ratio and cross-stochasticity] at the sub-percent level when sweeping over physically plausible ranges. The paper concludes that the data-driven channel serves as a model-independent companion to simulation-based priors, offering an independent handle on whether SBP-driven σ8 shifts reflect new physics or HOD systematics.

Conclusion

The joint analysis successfully combines internal DESI cross-correlations with the PFS×DESI overlap channel to provide superior constraints on cosmological parameters compared to single-tracer analyses. The framework is generalizable to any number of overlapping spectroscopic surveys by partitioning the volume into disjoint regions, each carrying a multi-tracer Fisher built from active tracers. This approach provides a model-independent check on simulation-based priors by constraining parameters directly from observed cross-spectra rather than HOD mocks. The method is validated against numerical convergence tests and analytical limit checks, confirming the consistency of the assembled Fisher matrix. All marginalized constraints obey the ordering σDESI+PFS ≤ σDESI-only for all cosmological parameters.

--- Page 1 ---

IPMU26-0019

Prepared for submission to JCAP

Multi-tracers, multi-surveys: a joint

Fisher analysis of DESI+PFS

Nhat-Minh Nguyen (Nguyễn Nhật Minh) a,b,c

aCenter for Data-Driven Discovery, Kavli IPMU (WPI), UTIAS, The University of Tokyo,

Kashiwa, Chiba 277-8583, Japan

bKavli IPMU (WPI), UTIAS, The University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba

277-8583, Japan

Institute For Interdisciplinary Research in Science and Education, ICISE, Quy Nhon, 55121, Vietnam

E-mail: nhat.minh.nguyen@ipmu.jp

Abstract. Marginalizing over roughly 12 effective-field-theory (EFT) nuisance parameters

per tracer per redshift bin is a dominant systematic cost in full-shape galaxy power spectrum

analyses. Simulation-based priors (SBP) tighten these parameters but rely on N-body simulations and halo-occupation

distribution (HOD) models. We propose a multi-tracer Fisher analysis as a model-independent alternative: cross-spectra between galaxy populations calibrate EFT bias and stochastic parameters from data alone, through two channels—within a survey and across overlapping surveys—combined in a volume-partitioned joint Fisher. We forecast across the 14,000 deg2 Dark Energy Spectroscopic Instrument (DESI) footprint,

including the ∼ 1,200 deg2 Prime Focus Spectrograph (PFS) overlap at z ∈ [0.6, 1.6] with

up to 4 tracers (PFS-ELG, DESI-ELG, DESI-LRG, DESI-QSO). The internal-DESI channel (LRG, ELG, and QSO over the full footprint) provides most of the gain,.

--- Page 2 ---

Contents

1 Introduction 1

2 Joint multi-tracer, multi-survey Fisher analysis 3

2.1 Marginalization cost 3

2.2 Cross-power constraints on nuisance parameters 3

2.3 Asymmetric scale cuts 5

2.4 Joint Fisher across the partitioned footprint 5

2.5 Comparison with simulation-based priors 5

--- Page 3 ---

1 Introduction

Full-shape analyses of the galaxy power spectrum, built on the effective field theory of largescale structure (EFTofLSS) [1–4], are now standard in spectroscopic-survey clustering. From their first applications on the Sloan Digital Sky Survey BOSS galaxy samples [e.g., 5, 6], full-shape EFT analyses have improved large-scale structure constraints on, for example, neutrino masses [7–9], primordial non-Gaussianity [10–12], early dark energy [13–15], clustering dark energy [16], interacting dark energy [17]. The DESI collaboration applied this framework to the DESI DR1 fullshape clustering analyses, reporting evidence for dynamical dark energy [21, 22]. Independent reanalyses by Chudaykin, Ivanov & Philcox traced the same signal through a sequence of selfcontained studies: ΛCDM constraints from the joint power spectrum and bispectrum [23], dark energy and neutrino masses [9], and primordial non-Gaussianity [10–12]. Combining DR1 with CMB data, Ivanov et al. [24] reported a robust preference for the normal neutrino mass hierarchy over the inverted hierarchy, and Chudaykin et al. [25] folded simulation-based priors into the same pipeline. Sharpening σ(Mν) from galaxy clustering is therefore a direct goal of next-generation full-shape analyses, since Mν is currently the most prior-sensitive parameter in the EFT pipeline.

At next-to-leading (i.e. one-loop) order, the EFT framework parameterizes our ignorance of small-scale physics through roughly 12 nuisance parameters per tracer per redshift bin [26, 27]: bias coefficients, counterterms, and stochastic amplitudes (Tab. 2).

Improvements for AI systems

Calibration of bias parameters via cross-spectra allows AI to perform model independent checks on simulation based priors by using observed cross-spectra between galaxy populations rather than relying solely on HOD mocks for constraining b1sigma8 shifts.
The system can use a volume partitioned joint Fisher where the per bin Fisher decomposes into Foverlap and Fnonoverlap, ensuring that no information is double-counted while combining internal DESI channel constraints with PFS cross spectrum information.
AI systems can be trained to utilize the dominant driver of improvement, which is calibration of the linear bias b1sigma8, by using cross-spectra between tracers in the same volume to determine the bias ratio bX1/bY1 cosmicvariance-free in the high-density limit.
The system can implement a robust assessment of systematic uncertainty by comparing data driven constraints with simulation based priors, serving as an independent handle on whether SBP driven sigma8 shifts reflect new physics or HOD systematics.
AI can dynamically adapt the analysis by using an asymmetric scale cut based on the FoG velocity ratio rsigma v to determine kPFSmax, and then focus per redshift bin nuisance constraints where they are most efficient in reducing uncertainty.

Multi-tracers, multi-surveys: a joint Fisher analysis of DESI+PFS

Listen

Transcript

Key concepts

Terminology

Summary

Improvements for AI systems

Sources

Related papers