Cosmic radio flashes can untangle dark matter from galactic winds

Combining weak gravitational lensing with fast radio bursts strips out gas interference to reveal pristine dark matter clustering across megaparsec scales, according to a new astrophysics preprint.

When light from distant galaxies travels across billions of light-years to reach terrestrial telescopes, its path is subtly bent by the gravity of intervening matter.¹ This gentle distortion, known as cosmic shear, allows astronomers to reconstruct maps of invisible dark matter across the cosmos.¹ Yet the visible universe repeatedly complicates this picture. As supermassive black holes erupt and giant stars explode inside galaxies, they blow vast clouds of ordinary gas outward into intergalactic space.¹ This chaotic dispersal of ordinary matter reshapes the surrounding gravitational landscape on scales of millions of light-years, creating a noisy backdrop that obscures the subtle signature of dark matter itself.

Cosmologists call this confounding disruption baryonic feedback.¹² Because ordinary matter, or baryons, comprises roughly five percent of the total cosmic energy density while cold dark matter makes up roughly 25 percent, the gravitational pull of dark matter dictates how the largest cosmic structures form and grow.¹ However, on physical scales smaller than a few megaparsecs, the explosive energy released by active galactic nuclei and supernovae shoves gas around vigorously.¹ This pushes gas out of dense galactic centres and into diffuse halos, fundamentally altering the matter distribution at small spatial scales.

To measure the underlying properties of dark matter accurately, researchers must first understand how light, matter, and gas interact along the line of sight. When light leaves a distant galaxy, it travels across cosmic distances where it encounters gravitational pull from both dark matter halos and diffuse intergalactic gas. That gravitational tug warps the shapes of background galaxies in a predictable statistical pattern. At the same time, astrophysical processes inside galaxies blast hot gas into the space between them, altering the local density of matter. If scientists cannot separate the gravitational effect of this displaced gas from the gravitational pull of dark matter, their calculations of cosmological parameters become biased. Measuring the distribution of that displaced gas independently allows researchers to subtract its influence and isolate the pure dark matter signal.

A study posted on the arXiv preprint server by Calvin Leung of the University of California, Berkeley, and colleagues across the United States and the Netherlands introduces a mathematical technique to solve this dilemma.³¹ The researchers propose a method called baryon nulling, which combines weak gravitational lensing maps with dispersion measurements from fast radio bursts to remove the uncertain effects of gas physics without relying on complex, speculative feedback models.¹³

How does gas movement distort dark matter measurements?

Astrophysical feedback alters the total matter power spectrum by flattening the density peaks of gas that would otherwise track dark matter.¹ Weak gravitational lensing measures the total matter distribution directly, meaning that cosmic shear surveys capture the combined gravitational tug of dark matter and ordinary matter along the line of sight. In computer simulations that include only dark matter, gravity pulls mass into compact, dense halos. When hydrodynamical processes such as gas cooling, supermassive black hole winds, and supernova shockwaves are included, the gas heats up and expands outward into surrounding space.¹ This displacement reduces the concentration of matter on spatial scales corresponding to wavenumbers larger than roughly 1 inverse megaparsec, suppressing the matter power spectrum by 10 to 20 percent compared to a cosmos made purely of dark matter.¹

Because different models of galaxy evolution predict widely varying amounts of gas expulsion, cosmological surveys have struggled to interpret small-scale cosmic shear data. In past surveys such as the Kilo-Degree Survey, Subaru Hyper Suprime-Cam, and the Dark Energy Survey, researchers often discarded data from small angular scales altogether to avoid these systematic errors. Alternatively, scientists have attempted forward-modelling the gas distribution by introducing free parameters that describe how feedback behaves. That approach introduces nuisance parameters into statistical inferences and creates the risk that researchers might choose an incorrect physical model for feedback.

Fast radio bursts offer an independent probe that bypasses this modelling dilemma.¹ These millisecond-duration pulses of radio waves originate in distant galaxies and traverse the ionized intergalactic medium before reaching detectors on Earth.¹⁴ As the radio waves pass through free electrons in cosmic gas, different frequencies travel at slightly different speeds, spreading the pulse out in time. This delay, quantified as the dispersion measure, provides a direct line-of-sight measurement of the total column density of ionized electrons between the source and the observer.¹²

Why can fast radio bursts cancel the gas signal?

Fast radio burst dispersion measures track the cosmic electron distribution, which directly reflects the overall distribution of ordinary matter on cosmological scales.¹ Because stars and neutral gas account for less than two percent of the cosmic baryon inventory, the distribution of free electrons traces the spatial distribution of total baryons with high fidelity across physical scales relevant to weak lensing. Cross-correlating cosmic shear maps with dispersion measure catalogs measures the statistical overlap between total matter and intergalactic gas, yielding a quantity known as the matter-baryon cross-power spectrum.¹

The mathematical strategy designed by Leung and collaborators takes advantage of the relative proportions of matter in the universe. In the standard cosmological framework, the total matter density fluctuation consists of the cold dark matter fluctuation weighted by its cosmic mass fraction, plus the baryon fluctuation weighted by the baryonic mass fraction.¹ When cosmologists compute the total matter auto-power spectrum, the result contains three distinct terms: a dominant dark matter auto-power term, a cross-power term between dark matter and baryons, and a smaller baryon auto-power term.¹ The matter-baryon cross-power term contains the largest fraction of the baryonic contamination signal.¹

By multiplying the cross-correlation between cosmic shear and fast radio burst dispersion measures by a factor proportional to the cosmic baryon fraction and subtracting it from the total cosmic shear power spectrum, the baryon nulling method algebraically cancels the cross-power term.¹ The residual signal leaves behind the pure dark matter power spectrum, accompanied only by a negligible second-order baryonic auto-power term that accounts for roughly four percent of the amplitude.¹ This subtraction removes the primary pathway through which astrophysical gas physics corrupts the dark matter measurement.

Explaining the physical intuition behind this cancellation, Calvin said in response to questions from Primary that "Weak lensing distorts the shapes of distant galaxies, and gives us a way to sum up all the mass along the line of sight to those distant galaxies. Fast radio bursts allow us to measure all the gas along the line of sight, so if you can subtract those two, what’s left is predominantly the imprint of dark matter. This is what makes fast radio bursts uniquely powerful."C

Even though a small residual baryonic component remains uncancelled, Calvin noted that its influence on cosmological inference is suppressed.C "We probably still need to forward-model the baryons to some extent," Calvin told Primary. "What’s important is that because the baryonic feedback is a second order effect, you’re no longer in this uncomfortable regime where the (baryon) tail is potentially wagging the (cosmology) dog."C

What do cosmological simulations show about the nulling technique?

Applying the baryon nulling procedure to hydrodynamical simulations reduces the sensitivity of cosmic shear power spectra to baryonic feedback by nearly an order of magnitude.¹³ Leung and his colleagues tested the technique using the FLAMINGO suite of hydrodynamical simulations, which models cosmic structure across a volume of 1 cubic gigaparsec.¹ The FLAMINGO suite includes eight distinct variations of baryonic feedback, encompassing differing efficiencies of stellar feedback as well as thermal and jet-driven feedback from active galactic nuclei. These variations span a factor of two range in the median gas fraction of massive galaxy clusters, covering the full envelope of feedback behaviors permitted by modern observations.¹

Across these disparate simulation models, the raw total matter power spectra varied by roughly 10 percent at a wavenumber of 1 inverse megaparsec and a redshift of 0.5.¹ Once the researchers applied their baryon nulling formula, the spread in power spectra across all eight feedback variants shrank dramatically, demonstrating that the nulled observable is largely insensitive to the specific astrophysical mechanism that pushed the gas around. The method achieves this stability because it measures the gas distribution directly from data rather than predicting it through theoretical equations.

The cancellation held across every simulation condition examined by the researchers.C Calvin said that "There were no extreme outliers, which was exciting! Even the most extreme feedback scenario allowed by existing data, which goes by the catchy name of “fgas-8sigma”, behaved no differently from the rest."C

To connect three-dimensional physical power spectra to two-dimensional astronomical observations on the sky, the authors developed a mathematical re-mixing matrix.¹ Because cosmic shear and fast radio burst dispersion measures accumulate along the line of sight under slightly different geometric weighting kernels, combining them directly would otherwise distort the angular clustering signal. The re-mixing matrix inverts the projection kernel for dispersion measures and remaps it onto the lensing kernel. The team demonstrated that this matrix depends primarily on the geometric window functions of the surveys rather than on complex gas astrophysics, ensuring that the nulling process remains model-independent.¹

What are the boundaries of the method?

The results presented in the study are theoretical demonstrations based on idealized hydrodynamical simulations and mathematical approximations. The calculations assume that the redshift distributions of background galaxies and detected fast radio bursts follow identical, simple top-hat profiles, whereas real-world sky surveys contend with complex, uneven redshift distributions that must be calibrated carefully. Furthermore, the projection kernel inversion relies on the assumption that the growth rate of matter fluctuations matches the growth rate of electron fluctuations over the surveyed redshift intervals.¹ The preprint reporting these findings has been posted to the arXiv repository and has not yet completed formal peer review.³⁶

Calvin acknowledged that the current numerical test makes simplifying assumptions about the cosmic baryon census.C "One major caveat so far is that at present we’ve done a simplified test where we just focus on the ionized gas," Calvin said. "We ignore neutral atoms, stars, planets, and all the pretty parts of the universe that we like to look at. But it turns out that for cosmology, ignoring all those parts of the “baryon budget” only makes a big difference on length scales that are considered by cosmologists to be “very small” (i.e. single galaxies)."C

Addressing the assumption of identical growth rates, Calvin explained that the method depends only on relative differences.C "The evolution of electrons over cosmic time will certainly change differently from that of total matter," Calvin said. "How it does so is somewhat of an open question, and it’s never been measured. In the baryon nulling formalism, you end up being sensitive to only the difference between electrons and total matter. So the calibration of redshift evolution effects becomes relative and not absolute."C

Real-world measurements must also contend with observational noise and contamination from host galaxies. A fast radio burst dispersion measure includes contributions from the host galaxy and the Milky Way interstellar medium in addition to the intergalactic gas along the cosmic line of sight. Independent research published in The Open Journal of Astrophysics by Robert Reischke and Steffen Hagstotz demonstrated the first observational measurement of baryonic feedback from roughly 100 localized fast radio bursts, while emphasizing that precise modeling of host and Galactic dispersion measures is essential to avoid systematic errors. Cosmological surveys will require large statistical samples of thousands of localized radio bursts to ensure that local dispersion noise cancels out cleanly.

The upcoming generation of wide-field astronomical instruments is well positioned to supply the necessary data. Deep optical and near-infrared imaging surveys, including the Vera C. Rubin Observatory Legacy Survey of Space and Time, the Euclid space telescope, and the Nancy Grace Roman Space Telescope, will map billions of galaxy shapes across vast swathes of the sky.¹ Simultaneously, next-generation radio observatories such as the Canadian Hydrogen Observatory and Radio-transient Detector and the Deep Synoptic Array 2000 are projected to discover tens of thousands of fast radio bursts every year.¹⁴

Calvin highlighted the scale of data expected from these future radio arrays.C "Only if the redshift accuracy is worse than about 8% does it limit your ability to extract a cross correlation signal," Calvin said. "The crossover point depends on how good the weak lensing data looks, but it’ll probably happen in the ~10,000 FRB regime. Currently I lead an exciting project - the CHIME Outriggers - which has a few hundred FRBs. I am used to treating each of those sources like a unique child. But in the era of the DSA-2000 and CHORD, we will detect ~10,000 FRBs per year!"C

By cross-correlating these expansive datasets, cosmologists can deploy baryon nulling to extract dark matter clustering on megaparsec scales without discarding data or relying on arbitrary feedback priors, unlocking tighter constraints on the nature of dark matter and the sum of neutrino masses.C

This piece was prepared from the arXiv preprint and public records together with answers from Calvin to six questions from the Primary news team, completed September 2026.

Save this story

The record

56 sentences trace to 5 sources and 1 contributor.

All 6Documents 2Records 0Reporting 3People 1

1. 1 CL Calvin Leung, Josh Borrow, Kiyoshi W. Masui, Shion Andrew, Kai-Feng Chen, Joop Schaye, Matthieu Schaller Paper · arxiv.org · 1 Sep 2026 Nulling baryonic feedback in weak lensing surveys using cross-correlations with fast radio bursts Preprint · may not have been peer reviewed See the source
2. 2 A astro.theoj.org Third-party reporting · astro.theoj.org · 29 Jun 2026 A first measurement of baryonic feedback with Fast Radio Bursts | Published in The Open Journal of Astrophysics See the source
3. 3 A arxiv.org Paper · arxiv.org Nulling baryonic feedback in weak lensing surveys using cross-correlations with fast radio bursts Preprint · may not have been peer reviewed See the source
4. 4 K kipac.stanford.edu Third-party reporting · kipac.stanford.edu Prospects for Solving Baryonic Feedback in Weak Lensing using FRBs See the source
5. 5 C Calvin Contributor · interview 4 Sep 2026 · in the article since 15 Oct 2026 Contribution — Calvin 16 statements added to this article
6. 6 L leungcalvin.github.io Third-party reporting · leungcalvin.github.io Calvin Leung See the source

Also consulted: 4 pages across 4 sites, none cited.

Article history

1. Published 15 Oct 2026, 17:07
   Assembled by the Primary desk from 5 sources · 1 contributor · 56 cited sentences
2. C Calvin 16 statements added 4 Sep 2026 42 days before publication

   1. Go to the sentence: By multiplying the cross-correlation between cosmic shear and fast radio burst dispersion measures by a factor proportional to the cosmic baryon fraction and subtracting it from the total cosmic shear power spectrum, the baryon nulling method algebraically cancels the cross-power term.↑
   2. Go to the sentence: The residual signal leaves behind the pure dark matter power spectrum, accompanied only by a negligible second-order baryonic auto-power term that accounts for roughly four percent of the amplitude.↑
   3. Go to the sentence: Explaining the physical intuition behind this cancellation, Calvin said in response to questions from Primary that "Weak lensing distorts the shapes of distant galaxies, and gives us a way to sum up all the mass along the line of sight to those distant galaxies. Fast radio bursts allow us to measure all the gas along the line of sight, so if you can subtract those two, what’s left is predominantly the imprint of dark matter. This is what makes fast radio bursts uniquely powerful."↑
   4. Go to the sentence: Even though a small residual baryonic component remains uncancelled, Calvin noted that its influence on cosmological inference is suppressed.↑
   5. Go to the sentence: "We probably still need to forward-model the baryons to some extent," Calvin told Primary. "What’s important is that because the baryonic feedback is a second order effect, you’re no longer in this uncomfortable regime where the (baryon) tail is potentially wagging the (cosmology) dog."↑
   6. Go to the sentence: These variations span a factor of two range in the median gas fraction of massive galaxy clusters, covering the full envelope of feedback behaviors permitted by modern observations.↑
   7. Go to the sentence: The cancellation held across every simulation condition examined by the researchers.↑
   8. Go to the sentence: Calvin said that "There were no extreme outliers, which was exciting! Even the most extreme feedback scenario allowed by existing data, which goes by the catchy name of “fgas-8sigma”, behaved no differently from the rest."↑
   9. Go to the sentence: Furthermore, the projection kernel inversion relies on the assumption that the growth rate of matter fluctuations matches the growth rate of electron fluctuations over the surveyed redshift intervals.↑
   10. Go to the sentence: Calvin acknowledged that the current numerical test makes simplifying assumptions about the cosmic baryon census.↑
   11. Go to the sentence: "One major caveat so far is that at present we’ve done a simplified test where we just focus on the ionized gas," Calvin said. "We ignore neutral atoms, stars, planets, and all the pretty parts of the universe that we like to look at. But it turns out that for cosmology, ignoring all those parts of the “baryon budget” only makes a big difference on length scales that are considered by cosmologists to be “very small” (i.e. single galaxies)."↑
   12. Go to the sentence: Addressing the assumption of identical growth rates, Calvin explained that the method depends only on relative differences.↑
   13. Go to the sentence: "The evolution of electrons over cosmic time will certainly change differently from that of total matter," Calvin said. "How it does so is somewhat of an open question, and it’s never been measured. In the baryon nulling formalism, you end up being sensitive to only the difference between electrons and total matter. So the calibration of redshift evolution effects becomes relative and not absolute."↑
   14. Go to the sentence: Calvin highlighted the scale of data expected from these future radio arrays.↑
   15. Go to the sentence: "Only if the redshift accuracy is worse than about 8% does it limit your ability to extract a cross correlation signal," Calvin said. "The crossover point depends on how good the weak lensing data looks, but it’ll probably happen in the ~10,000 FRB regime. Currently I lead an exciting project - the CHIME Outriggers - which has a few hundred FRBs. I am used to treating each of those sources like a unique child. But in the era of the DSA-2000 and CHORD, we will detect ~10,000 FRBs per year!"↑
   16. Go to the sentence: By cross-correlating these expansive datasets, cosmologists can deploy baryon nulling to extract dark matter clustering on megaparsec scales without discarding data or relying on arbitrary feedback priors, unlocking tighter constraints on the nature of dark matter and the sum of neutrino masses.↑