CMB constraints on dark matter-proton scattering: investigating prior-volume effects using profile likelihoods

arXiv:2603.25731 · astro-ph.CO, hep-ph · Submitted 2026-03-26 · Read on arXiv

Listen

Radio episode about this paper

Transcript

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

Vera: Today's paper: "CMB constraints on dark matter-proton scattering".

Jocelyn: This paper investigates constraints on velocity-independent dark matter-proton scattering using Cosmic Microwave Background (CMB) data, specifically focusing on how prior-volume effects influence Bayesian inferences compared to frequentist profile likelihood methods.

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

Title and authors: Vera: We’ve talked about the setup, and now let’s get into what the paper actually found regarding the constraints derived from this study. The main finding of "CMB constraints on dark matter-proton scattering: investigating prior-volume effects using profile likelihoods" is a direct comparison between MCMC samples from Bayesian inference and profile likelihood ratios.

Jocelyn: So, what’s the core result here for us as observers? It seems they are showing that the MCMC results can be systematically stronger than the profile-likelihood constraints by a factor of two or more across all dark matter masses and interaction fractions.

Subrahmanyan: That systematic difference is significant because it means that if we only look at the frequentist profile likelihood results, we might be allowing larger cross sections than are actually permitted by the data when using a Bayesian approach with appropriate priors.

Vera: They also explored how changing the lower bound on the prior for ten(sigma zero/cm two) affects things. When they set a stricter lower bound on that prior, it actually pulls the posterior distribution towards smaller values of the cross section, leading to tighter constraints in their Bayesian analysis.

Jocelyn: That’s a very interesting detail because it shows how sensitive the final answer is to how we define our starting assumptions about new physics before we even look at the data. It suggests that a poorly chosen prior can significantly influence what we conclude about dark matter scattering.

Subrahmanyan: This sensitivity confirms that these prior-volume effects aren't just theoretical quirks; they are tangible biases in the resulting constraints on fundamental parameters like sigma zero and f chi. The paper shows how marginalizing over a broader prior volume for one parameter causes looser constraints on the other.

Vera: So, to put it simply, this paper shows that the way we structure our Bayesian analysis dictates whether we get tighter or looser limits on the dark matter-proton scattering cross section when we are exploring parameter space near zero interaction strengths.

Jocelyn: And that’s important because it means if you want the most conservative limits, you have to be very deliberate about your prior choices, or you risk getting results that aren't representative of what the data alone dictates.

Subrahmanyan: It reinforces the idea that we can’t treat all statistical tools as interchangeable; they carry different assumptions about parameter space exploration, and understanding those differences is crucial for accurate astrophysical conclusions.

The paper's summary: Vera: Now let’s discuss what the authors suggest we should do moving forward based on their findings. They aren't just pointing out a problem; they are offering a way forward for how researchers handle these types of constraints.

Jocelyn: What is the concrete suggestion here? Are they telling us to abandon Bayesian methods entirely, or is it just about refining the analysis process? I’m ready to hear what they propose for practical use.

Subrahmanyan: The main recommendation from the authors is straightforward: we should supplement our Bayesian constraints with frequentist statistics. This isn't about abandoning Bayes’ theorem; it’s about using profile likelihoods as a check or a complementary tool to assess the impact of priors that we might be unknowingly introducing.

Vera: So, they are essentially saying that if you run your Bayesian analysis, you should also run the frequentist profile likelihood method to see how much your constraints shift when you remove or change those specific priors. It’s about cross-method validation.

Jocelyn: That makes sense because it gives us a way to gauge the robustness of our results. If both methods give similar results, we have more confidence in the final constraint on dark matter interactions derived from CMB data like Planck two thousand eighteen.

Subrahmanyan: I agree; this approach helps mitigate the issue where one method might be overly sensitive to prior choices, especially when the model parameters are near zero and become unconstrained. It offers a way to better assess those potential biases before publishing results based solely on one inference technique.

Vera: So, the improvement is methodological: we use frequentist profile likelihoods to verify and balance our Bayesian inferences, which helps us avoid getting those artificially shifted results caused by overly broad priors in the ΛCDM limit.

Jocelyn: It sounds like a practical step for anyone working on CMB data interpretation. We need to be careful about relying on just one inference pipeline when the physics we are probing is subtle.

The paper's improvements: Vera: So, wrapping up this discussion on "CMB constraints on dark matter-proton scattering: investigating prior-volume effects using profile likelihoods," the paper emphasizes that Bayesian statistics can be tricky when models approach the Standard Model limit due to those prior-volume effects.

Jocelyn: And they conclude by strongly recommending that researchers supplement their Bayesian analyses with frequentist statistics to better assess those biases, especially for parameters like sigma zero and f chi.

Subrahmanyan: I just want to add that the finding about how the choice of prior for one parameter affects the other is a very subtle point. It shows that even within a Bayesian framework, there are interconnected dependencies in how we explore parameter space that require careful consideration.

Vera: It really does show that we can’t just trust the output from one inference technique if those underlying assumptions, like broad priors, might be pushing us toward an unphysical region of the parameter space.

Jocelyn: And I think this is a vital piece of context for anyone trying to interpret these results from CMB data—it adds a necessary layer of statistical scrutiny to our work on dark matter interactions.

Subrahmanyan: Ultimately, this paper serves as a reminder that rigorous cross-method validation is the way forward for solid scientific inference in this area. It’s about using the profile likelihood method to keep our results grounded in data rather than relying solely on prior assumptions.

Vera: It's a very important piece of work because it gives us a clearer path forward for handling these kinds of statistical biases when we are looking at subtle new physics signals in the CMB.

Jocelyn: It’s certainly something that warrants paying attention as we continue to analyze data from missions like Planck. We’ll be ready for the next paper soon.

Conclusion: Vera: So we’ve talked through the paper "CMB constraints on dark matter-proton scattering: investigating prior-volume effects using profile likelihoods," and essentially, they showed how those Bayesian methods can be biased when priors are too broad in certain limits.

Jocelyn: Yeah, that's a lot to wrap our heads around, Vera. It’s wild how just tweaking the assumptions about new physics parameters like the cross section sigma zero can actually change the constraints we get from Planck data.

Subrahmanyan: From a theoretical side, it confirms that when we approach the Standard Model limit for dark matter interactions, Bayesian inference needs careful checks against frequentist profile likelihoods to ensure our results aren't being skewed by non-motivated priors.

Vera: Exactly, and they made a strong recommendation to supplement Bayesian constraints with frequentist statistics specifically to assess those potential biases in the CDM limit where parameters become unconstrained.

Jocelyn: So if we want the most conservative limits on dark matter scattering, it sounds like we need that cross-method validation to be rigorous, especially concerning how we define our prior ranges for new physics.

Subrahmanyan: Precisely; those prior-volume effects are real because when you let parameters approach zero, the other model parameters get unconstrained, and a broad prior volume can artificially pull your posterior distribution towards those unphysical regions of parameter space.

Vera: It’s a subtle effect because it shows that the choice in the prior for one parameter directly changes the constraints on another new model parameter, which is something we have to keep in mind when we're pushing our limits.

Jocelyn: That means if we are looking at dark matter-proton scattering, just picking a broad prior without checking how it affects other parameters might lead us down an incorrect path for the science.

Subrahmanyan: Indeed, and it underscores that while Bayesian tools are useful for exploration, they require careful calibration when probing physics near the known Standard Model baseline.

Vera: Well, that’s all we have time for today on this fascinating paper about CMB constraints on dark matter-proton scattering.

Jocelyn: It’s been really interesting seeing how these statistical methods interact with our observational data from missions like Planck two thousand eighteen.

Subrahmanyan: Next up, we'll be looking at the recent work comparing turbulent cascades and heating versus spectral anisotropy in solar wind via direct simulations, which gives us a different kind of constraint on plasma physics.

Maria C. Straight, *Tanvi Karwal, Jos´e Luis Bernal, *Kimberly K. Boddy

Department of Astronomy, The University of Texas at Austin · Kavli Institute for Cosmological Physics, Enrico Fermi Institute, and Department of Astronomy & Astrophysics, University of Chicago · Instituto de Física de Cantabria (IFCA), CSIC-Univ. de Cantabria · Texas Center for Cosmology and Astroparticle Physics, Weinberg Institute

astro-ph.CO, hep-ph

Submitted: 2026-03-26

Updated: 2026-09-29

Comments: 16 pages, 10 figures, and 2 tables. V2: Updated to match accepted version

Code: https://github.com/kboddy/class_public

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 83/100

The gist: This paper investigates constraints on velocity-independent dark matter-proton scattering using Cosmic Microwave Background (CMB) data, specifically focusing on how prior-volume effects influence

Key concepts

Momentum-transfer cross section ($ ext{s}_ ext{ch}$)
This quantity measures how often dark matter particles exchange momentum when they scatter off protons in the early universe. It is a key physical parameter that determines the strength of the interaction between dark matter and baryons.
Bayesian Inference
This statistical method uses Bayes' theorem to determine parameter probabilities based on both observed data (likelihood) and pre-existing beliefs about parameters (priors). The study shows that non-motivated priors can skew these results, especially when the model approaches the Standard Model limit.
Prior-Volume Effects
This occurs when Bayesian analyses sample large volumes of parameter space defined by broad priors. When a model approaches the Standard Model, this sampling can artificially pull constraints toward the Standard Model values because new physics parameters become unconstrained in that limit.

Terminology

Summary

This paper investigates constraints on velocity-independent dark matter-proton scattering using Cosmic Microwave Background (CMB) data, specifically focusing on how prior-volume effects influence Bayesian inferences compared to frequentist profile likelihood methods. It is significant because it demonstrates that Bayesian constraints can be biased by non-motivated priors when the model approaches the Standard Model limit, and it recommends supplementing Bayesian analyses with frequentist statistics to better assess these biases.

The Physics of Dark Matter-Proton Scattering

The study focuses on elastic scattering between dark matter and protons in the early universe, which introduces heat and momentum exchange between the dark matter and baryon fluids. The relevant physical quantity is the momentum-transfer cross section, parameterized as:

“The relevant quantity appearing in the linear Boltzmann equations that captures the effect of scattering is the momentum-transfer cross section σχ = Z domega dσ domega(1 − cos θ), (1)”

For velocity-independent scattering, this is parameterized as:

"The momentumtransfer cross section can be parameterized as σχ = σ0v n for a broad class of dark matter models [13], where v is the relative particle velocity between protons and dark matter. In this work, we focus on the velocity-independent case in which n = 0, and the momentum-transfer cross section σ0 coincides with the standard cross section [i.e., R domega(dσ/domega)].”

The interaction rate coefficient is given by:

The dark matter-baryon momentum-exchange rate coefficient is Rχ = N0aρb(1 − YHe) σχ / mχ + mp Tb / mp + Tχ / mχ ! 1/2,

Statistical Frameworks: Bayesian vs. Frequentist Constraints

The paper compares two primary statistical approaches for parameter inference:

  1. Bayesian Inference: This method uses Bayes’s theorem, where the posterior is proportional to the product of the likelihood and the prior, as described by Equation (5):

P(θd) ∝ Π(θ)L(dθ), (5)

  1. Frequentist Profile Likelihood: This method avoids prior reliance by maximizing the likelihood over all other parameters for fixed parameters of interest, defined by the profile likelihood ratio statistic in Equation (6):

∆χ 2 2(µ) = −2 ln L(dµ, ν˜) / L(dµˆ, νˆ)!,

The core issue explored is that in the limit where parameters like the cross section or interaction fraction approach zero, the other model parameters become unconstrained. This causes Bayesian analyses to sample the prior volume of these new-physics parameters unnecessarily.

Investigating Prior-Volume Effects

The analysis specifically targets two parameters susceptible to prior-volume effects: the momentum-transfer cross section (σ0) and the interaction fraction (fχ). The paper details how varying these priors impacts the resulting posterior distributions:

  1. When a prior range is widened to include values near zero for σ0 or fχ, the Bayesian analysis simply samples the prior volume of the new-physics parameters.

  2. In this region where the likelihood is flat (approaching ΛCDM), Bayesian analyses can artificially shift towards ΛCDM where the likelihood flattens and the new model parameters become unconstrained.

  3. The paper finds that the posterior distribution for σ0 may favor these small values of σ0 when a broad prior volume is included, leading to constraints that are stronger than those from frequentist methods.

Comparative Results and Recommendations

The study presents constraints derived from Planck 2018 CMB data, comparing the results across different interaction fractions (fχ) and dark matter masses (mχ). Key findings include:

The MCMC limits are systematically stronger than the profile-likelihood constraints by a factor of two or more across the entire mass range for all fractions, i.e., the profiles permit larger cross sections.

We recommend supplementing Bayesian constraints with frequentist statistics to better assess the impact of priors.

Furthermore, testing different prior combinations reveals parameter-dependent effects:

  1. Varying the lower bound on the prior on log10(σ0/cm2) leads to tighter constraints on σ0 in Bayesian analyses because the posterior is pulled towards lower values of the cross section.

  2. Marginalizing over a broader prior volume in the ΛCDM limit for one parameter results in looser constraints on the other, demonstrating that the choice in the prior for one parameter changes the constraints on the other new model parameter.

Conclusion

The paper concludes that while Bayesian statistics are commonly used, they can be complicated by prior-volume effects when models approach a Standard Model limit.

Improvements for AI systems

As a fastidious and diligent researcher, I have analyzed the provided scientific paper, CMB constraints on dark matter-proton scattering: investigating prior-volume effects using profile likelihoods. The core contribution of this work is providing rigorous statistical comparisons between frequentist (profile likelihood) and Bayesian (MCMC) methods for constraining non-gravitational dark matter interactions, specifically highlighting the biases introduced by prior volume effects.

Here are specific improvements that can be made to AI systems, categorized by the type of improvement:

)

The improved AI system can perform the following specific tasks:

  1. Agnostic Statistical Model Comparison and Bias Detection:

  2. Prior-Aware Constraint Interpretation and Robustness Testing:

  3. Automated Prior-Volume Effect Analysis in Parameter Space Exploration:

  4. Cross-Method Validation for Scientific Inference:

)

Here is a detailed breakdown of the specific improvements for each capability:

Here is a detailed breakdown of the specific improvements for each capability, drawing directly from the paper's methodology and findings:

Abstract

We present profile-likelihood constraints on velocity-independent dark matter-proton scattering, including cases in which only a fraction of dark matter has such non-gravitational interactions. Frequentist profile-likelihood techniques provide prior-independent constraints, circumventing prior-volume effects that we show arise in Bayesian constraints on this model. In the limit where the scattering cross section or the fraction of interacting dark matter approaches zero, the other interacting dark matter model parameters become unconstrained, causing the posterior distribution to favor that region of parameter space. Using Planck 2018 cosmic microwave background anisotropy data, we find a clear impact of prior-volume effects on the posteriors used to place constraints on dark matter scattering. Compared to the frequentist analysis, the Bayesian method consistently overestimates the constraints on the cross section. Given the potentially biased upper limits on models subject to prior-volume effects, such as this one, we recommend supplementing Bayesian constraints with frequentist statistics to better assess the impact of priors.

Sources

Related papers