How much dark matter really matters?

arXiv:2605.12593 · gr-qc, astro-ph.CO · Submitted 2026-05-12 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: General Relativity and Quantum Cosmology. The day in general relativity and quantum cosmology.

Kai: We're looking at the paper: "How much dark matter really matters?".

Mira: Here's the gist of the paper. Strong gravitational lensing is an important tool to trace dark matter via the light deflection it causes as part of cosmic structures The standard reconstructions fit a pre-defined mass-density model to the data, but eliminating these models reveals that observations directly constrain local properties of light-deflecting masses.

Kai: Modelling light deflections by massive cosmic structures Dark matter was introduced in our concordance cosmology to solve the missingmass problems in the cosmic microwave background and in gravitationally bound structures,

6: It has to provide 85% of the total cosmic matter content.

Mira: Strong gravitational lensing is thus an important tool to trace dark matter via the light deflection it causes as part of cosmic structures It only relies on General Relativity to reconstruct the light-deflecting mass distributions, also called “gravitational lenses”, without the need for hydrostatic or dynamic equilibria like other methods.

Kai: Separating data-based evidence from model assumptions To answer this question, we need to eliminate the lens model from the reconstruction process and determine the information about the light-deflectors that is directly contained in the observables of the multiple images The strong-lensing formalism is constructed by Fermat’s principle and correlates local differences in the overall light-deflection between the light paths.

Paper summary: Kai: So we’re diving into the specifics of "How much dark matter really matters?". The paper sets up the problem by saying that dark matter has to make up eighty-five percent of all cosmic matter content, and strong gravitational lensing is our main tool because it uses general relativity to reconstruct those mass distributions without needing any assumptions about hydrostatic or dynamic equilibria.

Mira: But the paper immediately points out a flaw in the standard approach, showing how fitting a pre-defined mass-density model to that lensing data results in an overestimate of the dark matter contents within those light-deflecting masses.

Lev: That’s the core tension here, isn't it? They’re saying that because we rely on these models to reconstruct things, we might be putting in too much dark matter than what is actually there, so they want to eliminate that model entirely to see the direct information in the observables.

Kai: They construct this strong-lensing formalism using Fermat's principle, which links arrival-time differences between light pulses deflected into different images to differences in their two dimensional light-deflecting gravitational potential and their geometric paths, which is what they call the "2d gravitational potential" and "geometric paths" <ref:2605.12593#pg3,light pulses deflected into different>.

Mira: And by measuring these arrival-time differences for time-varying sources like supernovae or quasars, they can constrain the average mass density inside the area enclosed by those multiple images. This gives us a way to put a hard constraint on that local density, which is what they call Tobs.

Lev: But the paper also flags a major issue right away: we usually only get about one hundred multiple images per galaxy cluster and only three per individual galaxy, which means the observables are typically sparse. Plus, there’s this assumption that additional light-deflecting masses between us and the light-emitting sources are mostly concentrated.

Kai: And to get past those assumptions, they focus on how the local lens properties at an image position x are a scaling factor due to the local two dimensional mass density, which is written as one minus kappa of x, and a shearing matrix that captures the local distortion of the source galaxy's surface brightness profile.

Mira: They then equate this to T based on observables with something like one minus kappa of two times A of x squared times A of one at x one to get f sub twelve equal to one minus kappa of x one over one minus kappa of x two, plus g sub one at x one, g sub two at x one, and g sub two at x two.

Lev: That’s where the math gets dense. It's trying to extract these local properties directly from what we measure—the arrival times and image positions—to get those terms that describe the local environment around a specific image position.

Kai: And the implication is that by determining the extent of this region of approximately constant mass density, they can infer the local smoothness scale of dark matter, which directly probes how smooth all matter along those lines of sight really is.

Mira: This connects back to what we discussed earlier: if we can determine that smoothness scale, it’s a direct probe of the smoothness of all matter along these specific lines of sight, and they find a candidate for a dark matter clump without any stellar counterpart near galaxy M94.

Lev: So, to summarize this part: they show that by removing the model and focusing on what the observables directly constrain about local properties, we can get information about the smoothness scale of dark matter without relying on big models. That sets up the next step for them, which is going beyond just strong lensing to three dee reconstruction.

Conclusion: Kai: So looking at "How much dark matter really matters?", the paper is essentially arguing that we need to be much more careful about the model choices we make when reconstructing cosmic structures through strong gravitational lensing, because those standard models seem to overestimate the dark matter content.

Mira: And it’s a call for us to move away from fitting those pre-defined mass models and instead use the observables directly to constrain the local properties of light-deflecting masses, which is a much more data-driven approach.

Lev: For me, what’s significant is that it pushes us toward realizing that our understanding of dark matter distribution might be limited by these modeling choices, even when we have high-quality lensing data. It suggests that the missing piece isn't necessarily something entirely new but rather a missing local curvature perturbation in the gravitational field.

Kai: And to wrap up on what this paper means for us: it’s about shifting our focus from just mapping out structures to understanding the underlying physics of those local distortions, which is crucial as we look for ways to detect dark matter directly.

Mira: It suggests that future astrophysics advances more efficiently when we leave knowledge gaps for future data to fill rather than holding onto a set of model assumptions that have been hard to fully test or support for decades.

Lev: And the final thought they bring up is whether the way space curves, maybe through some local curvature perturbation, is being incorporated correctly in our standard quasi-Newtonian strong-lensing formalism, which ties back into how we build things on real hardware.

Dr. Jenny Wagner

Academia Sinica Institute of Astronomy and Astrophysics

gr-qc, astro-ph.CO

Submitted: 2026-05-12

Updated: 2026-11-09

Comments: Honorable mention in the Gravity Research Foundation Essay Contest (published in IJMP D)

DOI: 10.1142/S0218271826410099

License: http://creativecommons.org/licenses/by-nc-nd/4.0/

Importance score: 92/100

The gist: Strong gravitational lensing is an important tool to trace dark matter via the light deflection it causes as part of cosmic structures The standard reconstructions fit a pre-defined mass-density…

Key concepts

Strong Gravitational Lensing
This technique uses the bending of light by massive cosmic structures, like galaxy clusters, as a tool to map out dark matter. It relies on General Relativity to reconstruct the mass distributions causing these deflections, providing information about how much mass is present along specific lines of sight.
Lens Model Elimination
The paper focuses on removing pre-defined mass models from the reconstruction process. By doing this, researchers can determine what information about the light-deflectors is directly contained in the observable multiple images, rather than relying on assumptions about their shape or distribution.
Local Lens Properties
These are specific measurements derived from observables that describe the mass density and distortion in the immediate vicinity of a multiple image position. By relating these local properties to observables, researchers can uniquely extract parameters like the ratio of mass densities and shear components at that point.
Smoothness Scale
Determining the extent of a region with approximately constant mass density allows researchers to infer the local smoothness scale of dark matter. This probes how smoothly matter is distributed along specific lines of sight, potentially revealing features like small dark-matter clumps.

Terminology

Summary

Strong gravitational lensing is an important tool to trace dark matter via the light deflection it causes as part of cosmic structures The standard reconstructions fit a pre-defined mass-density model to the data, but eliminating these models reveals that observations directly constrain local properties of light-deflecting masses. p. 1

Where the paper says it

Strong gravitational lensing is thus an important tool to trace dark matter via the light deflection it causes as part of cosmic structures. It only relies on General Relativity to reconstruct the light-deflecting mass distributions, also called “gravitational lenses”, without the need for hydrostatic or dynamic equilibria like other methods.

Modelling light deflections by massive cosmic structures

Dark matter was introduced in our concordance cosmology to solve the missingmass problems in the cosmic microwave background and in gravitationally bound structures, [6] It has to provide 85% of the total cosmic matter content. Strong gravitational lensing is thus an important tool to trace dark matter via the light deflection it causes as part of cosmic structures It only relies on General Relativity to reconstruct the light-deflecting mass distributions, also called “gravitational lenses”, without the need for hydrostatic or dynamic equilibria like other methods. p. 1

Where the paper says it

Strong gravitational lensing is thus an important tool to trace dark matter via the light deflection it causes as part of cosmic structures. It only relies on General Relativity to reconstruct the light-deflecting mass distributions, also called “gravitational lenses”, without the need for hydrostatic or dynamic equilibria like other methods.

Separating data-based evidence from model assumptions

To answer this question, we need to eliminate the lens model from the reconstruction process and determine the information about the light-deflectors that is directly contained in the observables of the multiple images The strong-lensing formalism is constructed by Fermat’s principle and correlates local differences in the overall light-deflection between the light paths. Being able to measure arrival-time differences for time-varying sources like supernovae or quasars, the average mass density inside the area enclosed by the multiple images is constrained. p. 2

Where the paper says it

1, left bottom). Being able to measure arrival-time differences for time-varying sources like supernovae or quasars, the average mass density inside the area enclosed by the multiple images is constrained.

Local lens properties as data-based information

The strong-lensing formalism starts with the fact that the 2d lensing mass density is approximately constant in the vicinity of a multiple image position x on the sky. The local lens properties at an image position x are an overall scaling due to the local 2d mass density, 1−κ(x), and a shearing matrix that captures the local distortion of the surface brightness profile of the source galaxy, the “reduced shear” due to masses close to the light path. p. 4

Where the paper says it

It can be limited to an area even smaller than a multiple image or extended beyond the images, as detailed below. The local lens properties at an image position x are an overall scaling due to the local 2d mass density, 1−κ(x), and a shearing matrix that captures the local distortion of the surface brightness profile of the source galaxy, the “reduced shear” due to masses close to the light path.

Equating T based on observables, Tobs, with T= A−1 (x2)A(x1) of local lens properties, we can uniquely extract f12 ≡ (1 − κ(x1))/(1 − κ(x2)), g1(x1), g2(x1), and g2(x2). p. 4
Where the paper says it

2 (right) for an example. Equating T based on observables, Tobs, with T= A−1 (x2)A(x1) of local lens properties, we can uniquely extract f12 ≡ (1 − κ(x1))/(1 − κ(x2)), g1(x1), g2(x1), g1(x2) and g2(x2).

Local lens properties constrain dark-matter properties

Determining the extent of the region of approximately constant mass density with our approach allows us to infer the local smoothness scale of dark matter. This criterion directly probes the smoothness of all matter along these lines of sight. p. 5

Where the paper says it

Local lens properties constrain dark-matter properties Determining the extent of the region of approximately constant mass density with our approach allows us to infer the local smoothness scale of dark matter. As our approach only analyses the areas of the multiple images and around them, this criterion directly probes the smoothness of all matter along these lines of sight.

A first candidate for a dark-matter clump without a stellar counterpart was found close to galaxy M94, [2]. p. 5
Where the paper says it

Both smoothness scales are lower bounds, as there is no sign of dark-matter clumps disrupting the continuity of the lensing. A first candidate for a dark-matter clump without a stellar counterpart was found close to galaxy M94, [2].

Such objects are often put at the borders of lens-model-based reconstructions, like in Abell 3827. p. 6
Where the paper says it

So it still seems unlikely that dark clumps without any luminous counterparts found in lens-model-based massdensity maps are real. Such objects are often put at the borders of lens-model reconstructions, like in Abell 3827 (see Fig.

Extending the mass-density reconstruction to include weakly lensing masses at larger distances can reveal if the strong shear of the first matching can have an astrophysical origin in the lens plane

A 3d reconstruction will also clarify if the cluster-size dark-matter halo currently used in the lens-model-based reconstructions is really necessary, as claimed in [4]. This scenario can also explain the slight relative rotations between images 1 to 3 of the same source also observed in Abell 3827. p. 7

Where the paper says it

Opening up for a full 3d lens reconstruction instead of the effective 2d approach allows for relative rotations between multiple images beyond the local lens properties in Eq. This scenario can also explain the slight relative rotations between images 1 to 3 of the same source also observed in Abell 3827.

Conclusion

As detailed in [21, 22, 24, 25] a deeper and clear understanding of all degeneracies occurring in the strong-lensing formalism can be gained from the derivation of this information. Most importantly, having reached the data-driven era, astrophysics advances more efficiently by leaving knowledge gaps to be filled with future data than holding on to a bag of model assumptions whose predictions have been difficult to clearly falsify or support for decades. If the sun curves space to double the deflection angle of background stars compared to the Newtonian formulation, the local curvature of galaxies and clusters may not be incorporated correctly in the quasi-Newtonian strong-lensing formalism. p. 9

Where the paper says it

One example is the wavelength independence of gravitational lensing, another one is the scale-freeness of Einsteinian and Newtonian gravity that should lead to scale-free local lens properties, unless some modified gravitational law holds. Observations from multiple filter bands are available, for instance from the Hubble or James Webb Space Telescopes, so we investigated two different strong-lensing examples on galaxy-cluster scale for potential wavelength dependencies, [13].

Maybe, the missing mass eluding a direct detection so far is actually a missing local curvature perturbation. p. 7
Where the paper says it

While this conclusion might be sufficient, it is by far not necessary, as supported by our findings and the arguments given above. Therefore, a fresh look into a 3d mass reconstruction of Abell 3827 will reveal where additional mass to the already observed one is actually required and can potentially be provided by luminous structures not yet taken into account in the reconstruction of the strong-lensing effects.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Robust dark matter constraint derivation: The improved AI can directly connect observables in strong-lensing configurations with the light-deflecting local lens properties that are necessary to create these lensing effects, yielding the maximum information contained in the data. This allows the system to bypass model assumptions when inferring dark matter content, focusing instead on data-based constraints at observable positions.

  2. Degeneracy resolution via local properties: The system can uniquely extract f12 ≡ (1 − κ(x1))/(1 − κ(x2)), g1(x1), g2(x1), and g2(x2) by equating the linear transformation T based on observables with T set up from local lens properties. This capability enables it to determine local lens properties which are free of the degeneracies occurring in the lens-modelbased approaches.

  3. Substructure detection refinement: The AI can identify potential dark matter clumps by analyzing the smoothness scale of dark matter between multiple images, probing the smoothness of all matter along these lines of sight. This refined method helps distinguish real substructures from artifacts by checking if features are consistent across different image matchings.

  4. Artifact identification in source reconstruction: The system can identify potential errors by constructing the local lensing approach to avoid the introduction of dark-matter artefacts in the lens, as well as one can identify artefacts in the source reconstruction for noisy surface brightness profiles of the multiple images. This allows for a data-driven pixelisation of sources and fixing resolution limits based on observable distances.

  5. Fundamental principle testing: The improved system can test fundamental principles like the wavelength independence of gravitational lensing by comparing local lens properties across different filter bands, confirming they coincided within the error bounds of about 10%. It can also compare properties across different mass scales to test the scale-freeness of Einsteinian and Newtonian gravity.

Abstract

Strong gravitational lensing is a key probe to trace dark matter. It assumes that mass curves spacetime so that light from a background source is deflected on its way to the observer. If dark matter contributes the major part to a massive cosmic structure, reconstructing the latter from strong-lensing observables allows us to infer characteristics of dark matter. Standard reconstructions fit a pre-defined mass-density model to the data. In this essay, I show how these mass models over-estimate the dark-matter contents of light-deflecting masses. Eliminating these models from the reconstruction reveals that observations directly constrain local properties of light-deflecting masses. How much dark matter is really needed in strong-gravitational-lensing effects and how much do we make up by our model choices?

Sources

Related papers