How do galaxy cluster dynamics and gravitational lensing constrain dark matter over modified gravity differently?

# Galaxy Clusters, Lensing, and the Dark Matter vs Modified Gravity Question

The core tension you're asking about is this: when we observe galaxy clusters, their gravitational behavior reveals "missing mass" that we can't see. Is that mass real (dark matter), or does gravity itself work differently than Einstein predicted? Lensing and dynamics give us two windows on this, but they answer slightly different questions.

The Degeneracy Problem

Here's the central difficulty: dark matter with new interactions and modified gravity can produce identical observational signatures [AG-2024.04-266]. When you measure how fast structures grow and how light bends around massive objects, both a modification to gravity's equations *and* a dark matter particle coupled to a new force field can fit the data equally well. This is called a degeneracy — like trying to solve an equation with two unknowns.

The good news is that this degeneracy isn't unbreakable. Gravitational redshift — the subtle change in light's color as it escapes a gravitational well — responds differently to true modified gravity versus dark forces [AG-2024.04-266]. Future surveys like the Square Kilometre Array should be able to use this to distinguish between generalized Brans-Dicke theories (modified gravity) and coupled quintessence (dark matter with extra interactions).

What Galaxy Clusters Themselves Reveal

Galaxy clusters are particularly useful testing grounds because their huge masses make gravitational effects strong and measurable. When physicists use strong and weak gravitational lensing on clusters, they can map both the overall mass distribution and small-scale substructure with exquisite precision.

For modified gravity: When tested against real cluster lensing data using disformal coupling (a specific kind of gravity modification), the characteristic coupling length comes out extremely small — around 100 kpc at best [AG-2024.12-570]. This is consistent with general relativity, placing tight constraints on how much gravity can deviate. Similarly, modified gravity theories like Hu-Sawicki *f(R)* and nDGP do predict stronger lensing effects than general relativity — larger Einstein radii and higher lensing probabilities — which should be testable with upcoming surveys [AG-2024.11-168].

For dark matter: Cluster lensing reveals something unexpected: the substructure inside clusters shows inner density profiles and radial distributions that disagree with cold dark matter (CDM) predictions [AG-2026.01-1153]. The cores of dark matter subhalos appear denser (inner slopes steeper than ∼2.5) and are distributed differently than CDM simulations expect. This hints that dark matter might be self-interacting rather than collisionless — a different dark matter model, not a modified gravity problem.

A Potential Unified Test

A clever new approach combines clustering (how matter clumps in space) with lensing (how geometry curves around that matter). In general relativity, the growth of density clumps and the growth of spacetime curvature are governed by the same equations. Modified gravity can break this connection [AG-2025.06-066]. Current data shows no deviation — the Weyl potential (geometry) tracks density growth to within 33% — but this becomes a powerful null test as surveys improve.

The MOND Counterexample

Finally, Modified Newtonian Dynamics (MOND) — perhaps the most radical alternative — fails catastrophically in clusters, leaving a large residual missing mass that cannot be explained [AG-2024.10-060]. This isn't a feature; it's a bug. It suggests that any viable modified gravity theory must still account for most of what we call dark matter in the cluster regime.

---

The bottom line: Clusters constrain modified gravity tightly (ruling out large deviations), reveal unexpected dark matter substructure incompatible with standard CDM, and offer multiple independent tests (lensing geometry vs growth rate, Einstein radii, time delays) to break degeneracies between competing models.