Modified Newtonian

By Apiary Science Team

Introduction

For more than eight decades astronomers have been confronted with a puzzling discrepancy: stars and gas in the outer reaches of galaxies orbit far faster than the visible mass can justify. The standard answer—​that the Universe is permeated by an invisible, non‑baryonic component called dark matter—​has become the cornerstone of modern cosmology. Yet, despite massive underground detectors, satellite experiments, and collider searches, a dark‑matter particle has remained elusive.

Enter Modified Newtonian Dynamics (MOND), a bold alternative proposed by physicist Mordehai Milgrom in 1983. Rather than adding unseen mass, MOND tweaks the law of gravity (or inertia) at extremely low accelerations, roughly \(a_0 \approx 1.2 \times 10^{-10}\,\text{m s}^{-2}\). With only one new constant, MOND reproduces the flat rotation curves of spiral galaxies, the tight Tully–Fisher relation, and a host of other galactic phenomenology that otherwise require fine‑tuned dark‑matter halos.

Why does this matter for Apiary and the broader mission of conservation? The same scientific rigor that drives the hunt for dark matter also underpins our efforts to understand pollinator ecosystems and to design self‑governing AI agents that can adapt to complex, data‑rich environments. In both cases, we are asking: When should we modify the rules we think we know, and when should we look for hidden actors that we have yet to detect? This article walks through MOND’s origins, its successes, its challenges, and the current frontier where astrophysics, particle physics, and computational intelligence intersect.

1. From Galactic Puzzles to a New Law of Gravity

1.1 The Rotation‑Curve Crisis

The first hint that something was amiss came from Vera Rubin’s pioneering work in the 1970s. By measuring the Doppler shift of emission lines from ionized gas in spiral galaxies, Rubin and collaborators found that rotation velocities remain roughly constant out to several optical radii, contrary to the Keplerian decline expected from the luminous mass distribution.

A typical massive spiral such as NGC 3198 (distance ≈ 14 Mpc) shows a rotation speed of \(v \approx 150\;\text{km s}^{-1}\) at radii beyond 15 kpc, where the enclosed stellar mass should only support \(v \approx 80\;\text{km s}^{-1}\) if Newtonian dynamics held. The discrepancy implies an extra gravitational pull equal to roughly three times the visible mass.

1.2 Milgrom’s Insight

Milgrom asked a simple question: What if Newton’s second law, \(F = ma\), changes when the acceleration falls below a certain threshold? He proposed

\[ \mu\!\left(\frac{a}{a_0}\right)a = a_{\text{N}} , \]

where \(a_{\text{N}}\) is the Newtonian acceleration, \(\mu(x)\) is an interpolation function with \(\mu(x) \to 1\) for \(x \gg 1\) (high accelerations) and \(\mu(x) \to x\) for \(x \ll 1\) (low accelerations), and \(a_0\) is the universal constant introduced above. In the deep‑MOND regime (\(a \ll a_0\)), the law reduces to

\[ a^2 = a_0 a_{\text{N}} . \]

Applying this to a circular orbit (\(a = v^2 / r\)) yields

\[ v^4 = G M a_0 , \]

which directly predicts the baryonic Tully–Fisher relation: the asymptotic rotation speed \(v\) depends only on the total baryonic mass \(M\) of the galaxy, not on its radius or dark‑matter halo. Observationally, the relation holds over five orders of magnitude in mass, from dwarf spheroidals (\(M \sim 10^7\,M_\odot\)) to giant spirals (\(M \sim 10^{11}\,M_\odot\)), with a scatter of less than 0.1 dex—remarkably tight for a theory with a single free parameter.

1.3 Early Reception

MOND was initially dismissed by many as an ad‑hoc fix. However, its predictive power—for instance, correctly anticipating the rotation curves of low‑surface‑brightness (LSB) galaxies before they were measured—earned a modest following. The community split into three camps: (1) Dark‑matter proponents, (2) MOND advocates, and (3) Hybrid skeptics who look for a deeper theory that could encompass both ideas.

2. Empirical Successes at Galactic Scales

2.1 Low‑Surface‑Brightness Galaxies

LSB galaxies have central surface brightnesses fainter than 23 mag arcsec\(^{-2}\) in the B band, meaning their internal accelerations are well below \(a_0\). MOND predicts that their rotation curves should rise slowly and then flatten at velocities dictated solely by their baryonic mass. Observations of over 200 LSB systems (e.g., the SPARC database) confirm this: the measured curves match MOND predictions to within 10 % without tweaking halo parameters.

2.2 Dwarf Spheroidals and the “Mass‑Discrepancy–Acceleration Relation”

Dwarf spheroidal galaxies (dSphs) orbiting the Milky Way, such as Sculptor and Fornax, have velocity dispersions of only a few km s\(^{-1}\) but appear to be dominated by dark matter in the standard paradigm (mass‑to‑light ratios \(> 100\)). In MOND, the external field effect (EFE)—a subtle consequence of the theory’s non‑linearity—explains why some dSphs show larger apparent mass discrepancies: the Milky Way’s gravitational field raises the effective internal acceleration, pushing the system out of the deep‑MOND regime. Detailed modeling reproduces the observed dispersion profiles with no dark matter and a single \(a_0\) value.

2.3 The Radial Acceleration Relation (RAR)

A more recent synthesis, the Radial Acceleration Relation, plots the observed acceleration \(g_{\text{obs}}\) (derived from rotation curves) against the Newtonian acceleration \(g_{\text{bar}}\) expected from the baryonic mass distribution. Across 4,600 data points from 153 galaxies, the relation follows a tight curve that is essentially the MOND interpolation function. The scatter is comparable to measurement uncertainties, suggesting a universal underlying law rather than the stochastic assembly of dark‑matter halos.

3. Relativistic Extensions: From TeVeS to Emergent Gravity

3.1 The Need for a Relativistic Theory

Newtonian MOND works well for non‑relativistic dynamics, but it cannot predict gravitational lensing, cosmic microwave background (CMB) anisotropies, or the propagation of gravitational waves. To be a viable alternative to dark matter, MOND must be embedded in a fully relativistic framework.

3.2 Tensor‑Vector‑Scalar (TeVeS) Gravity

In 2004, Jacob Bekenstein introduced TeVeS, a covariant theory that adds a scalar field \(\phi\) and a unit timelike vector field \(U_\mu\) to Einstein’s metric tensor \(g_{\mu\nu}\). The scalar field supplies the MONDian modification at low accelerations, while the vector field ensures the theory respects Lorentz invariance and provides the correct lensing strength. TeVeS predicts lensing angles identical to General Relativity (GR) with dark matter, provided the scalar field contributes the extra “mass”.

Tests of TeVeS in galaxy clusters have been mixed: while it reproduces lensing maps of many clusters, the Bullet Cluster (1E 0657‑558) remains a challenge (see Section 5). Nonetheless, TeVeS opened the door for bimetric and Einstein‑Aether theories that continue to be explored.

3.3 Emergent Gravity and Entropic Ideas

Erik Verlinde (2016) proposed Emergent Gravity, an entropic approach where gravity arises from microscopic information associated with spacetime. In this picture, the apparent dark‑matter effect emerges as an additional “elastic” response of spacetime to baryonic mass, yielding a MOND‑like acceleration law. The model reproduces the RAR on galactic scales but struggles with cluster‑scale lensing.

These relativistic extensions illustrate a broader trend: MOND may be a low‑energy limit of a deeper, perhaps quantum‑gravitational, theory. The search for such a theory mirrors attempts in particle physics to unify forces, and it inspires AI‑driven symbolic regression techniques that seek compact equations from astronomical data sets.

4. The Dark Matter Counter‑Argument: Evidence from Large‑Scale Structure

4.1 Cosmic Microwave Background

The Planck satellite’s measurements of the CMB power spectrum (ℓ ≈ 2–2500) are exquisitely fit by the ΛCDM model, which includes ~26 % cold dark matter. The positions of the acoustic peaks, especially the third peak, depend sensitively on the amount of non‑baryonic matter. MOND‑based cosmologies can mimic some aspects of the spectrum by adjusting the primordial power spectrum or invoking massive neutrinos, but a clean fit comparable to ΛCDM remains elusive.

4.2 Galaxy Cluster Mass Profiles

Galaxy clusters, the most massive bound structures (M ≈ 10\(^{14-15}\,M_\odot\)), exhibit X‑ray emitting hot gas with temperatures up to 10 keV. Hydrostatic equilibrium calculations require a gravitational potential far deeper than the visible galaxies and intracluster medium can provide. In the standard view, dark matter accounts for ~85 % of the cluster mass. MOND, even with TeVeS, typically underestimates the required mass by a factor of 2–3 unless additional unseen components (e.g., sterile neutrinos) are introduced.

4.3 Weak Lensing Surveys

Surveys such as DES (Dark Energy Survey) and KiDS (Kilo‑Degree Survey) map the weak lensing shear of millions of background galaxies. The statistical shear signal correlates with the large‑scale distribution of matter, and analyses consistently infer a matter density parameter \(\Omega_m \approx 0.31\), compatible with ΛCDM. While MOND can reproduce the shear‑mass relation on galaxy scales, extending it to the cosmic web requires extra fields or particles, blurring the line between “modified gravity” and “dark sector”.

5. The Bullet Cluster and Other “Smoking‑Gun” Tests

5.1 Anatomy of the Bullet Cluster

The Bullet Cluster (z ≈ 0.296) consists of two colliding galaxy clusters. X‑ray observations (Chandra) reveal hot gas (the dominant baryonic mass) lagging behind the collisionless galaxies, while gravitational lensing maps (HST, VLT) show mass peaks aligned with the galaxies, not the gas. In the dark‑matter picture, this separation is interpreted as dark matter passing through the collision unaffected, while the gas experiences ram pressure.

5.2 MOND’s Explanation Attempts

Proponents argue that the lensing signal could arise from a phantom dark‑matter distribution generated by the MOND scalar field. However, detailed simulations (e.g., Clowe et al. 2006; Angus et al. 2007) indicate that the required phantom density would need to be twice the observed baryonic mass, exceeding the theoretical limits of TeVeS without violating other constraints.

5.3 Beyond the Bullet: Abell 3827 and the “Dark Matter Offset”

In 2018, observations of the galaxy cluster Abell 3827 suggested a ~1.6 kpc offset between the dark‑matter halo and the central galaxy, interpreted as potential self‑interactions of dark matter. MOND models with an EFE can produce modest offsets, but the magnitude and direction of the observed shift remain difficult to reconcile without invoking additional unseen particles.

These high‑profile cases have cemented the Bullet Cluster as a critical benchmark. While not a definitive falsification of MOND, they illustrate the theory’s difficulty in reproducing phenomena where dark matter’s collisionless nature is directly inferred.

6. Bridging to Bee Conservation: Scaling Laws and Collective Behavior

6.1 The Analogy of Modified Rules

Bees, like stars in a galaxy, operate under simple local rules that give rise to complex global patterns. In dense apiaries, forager bees adjust their flight paths based on the local nectar gradient, a rule reminiscent of MOND’s acceleration threshold: when the nectar signal falls below a certain level, bees switch from a directed to a random search. This parallels how MOND modifies dynamics only when the gravitational acceleration drops below \(a_0\).

6.2 Data‑Driven Modeling

Apiary’s platform uses self‑governing AI agents to simulate colony dynamics. These agents incorporate feedback loops—e.g., brood temperature regulation, pheromone diffusion, and foraging allocation—that adapt when environmental parameters cross critical thresholds. The same mathematical tools (e.g., differential equations with non‑linear interpolation functions) that describe MOND’s transition from Newtonian to modified regimes are employed in the agents’ decision‑making modules.

6.3 Cross‑Pollination of Techniques

Astronomers now routinely apply machine‑learning surrogates to accelerate N‑body simulations of dark matter. Conversely, bee‑conservation researchers are adopting Bayesian hierarchical models to disentangle colony health signals from weather noise. The shared challenge—extracting a universal law from noisy, multi‑scale data—creates fertile ground for collaboration. A future AI-agents project could train an algorithm on both galaxy rotation curves and foraging trajectories, seeking a unified interpolation function that captures threshold behavior across domains.

7. Computational Frontiers: Simulating MOND with AI

7.1 N‑Body Codes for Modified Gravity

Standard cosmological simulations (e.g., Illustris, EAGLE) rely on Newtonian gravity plus dark‑matter particles. MOND requires solving a non‑linear Poisson equation:

\[ \nabla \cdot \bigl[\mu(|\nabla \Phi|/a_0)\, \nabla \Phi \bigr] = 4\pi G \rho . \]

Unlike the linear Poisson equation, this demands iterative solvers (e.g., multigrid, conjugate‑gradient) at each timestep. The Phantom of RAMSES code (Lüghausen et al. 2015) implements this approach, achieving sub‑percent accuracy for galaxy‑scale simulations on modern GPUs.

7.2 AI‑Accelerated Solvers

Recent work leverages physics‑informed neural networks (PINNs) to approximate the MOND potential directly from the density field. By training on a suite of analytically solvable cases (e.g., isolated Plummer spheres), the network learns the mapping \(\rho \mapsto \Phi\) and can generalize to complex merger scenarios. Early benchmarks show a 5–10× speedup over traditional multigrid methods, enabling the exploration of large parameter spaces (e.g., varying the interpolation function) that were previously prohibitive.

7.3 Symbolic Regression for the Interpolation Function

The functional form of \(\mu(x)\) is not uniquely fixed. While the original “simple” form \(\mu(x) = x/(1+x)\) works well, other proposals (e.g., the “exponential” \(\mu(x) = 1 - e^{-x}\)) offer subtle differences in galaxy outskirts. Using genetic programming and the PySR library, researchers have performed symbolic regression on the SPARC data set, uncovering a family of functions that minimize chi‑square while preserving the asymptotic limits. The resulting expressions can be directly inserted into simulation pipelines, allowing a data‑driven refinement of MOND’s phenomenology.

8. Dark Matter vs. Modified Gravity: A Pragmatic Comparison

The table underscores that both frameworks excel in different regimes. Dark matter shines in the early Universe and large-scale structure, while MOND delivers uncanny precision for individual galaxies. A hybrid view—where a small population of dark particles coexists with a modified gravity sector—has been proposed, but it also inherits the complexities of both sides.

9. Future Directions: Experiments, Theory, and Interdisciplinary Synergy

9.1 Laboratory Tests of Low‑Acceleration Gravity

Experiments such as the Eöt-Wash torsion balance have probed Newton’s inverse‑square law down to 55 µm, placing constraints on Yukawa‑type deviations. However, MOND’s effects manifest at accelerations far below Earth’s surface gravity, making direct laboratory verification challenging. Proposed space‑based missions (e.g., STEP, MICROSCOPE) could test the weak equivalence principle at the \(10^{-15}\) level, potentially revealing MOND‑type violations in the free‑fall regime.

9.2 Next‑Generation Surveys

The Vera C. Rubin Observatory (LSST) will deliver deep, multi‑band imaging of billions of galaxies, dramatically improving measurements of the RAR across redshift. Simultaneously, the Euclid mission’s spectroscopic data will refine constraints on the growth rate of structure, a key discriminator between ΛCDM and modified gravity.

9.3 AI‑Enhanced Model Selection

Bayesian model comparison, powered by nested sampling (e.g., PolyChord), can quantitatively weigh MOND against ΛCDM using combined data sets (CMB, lensing, galaxy kinematics). Integrating deep generative models to produce synthetic sky maps under each theory will help assess systematic biases, a technique already adopted in the Cosmic Frontier community.

9.4 Cross‑Disciplinary Workshops

Apiary plans a series of “Gravity and Bees” workshops, bringing together astrophysicists, entomologists, and AI researchers. Topics will include: (1) threshold dynamics in biological collectives, (2) scalable solvers for non‑linear field equations, and (3) ethical considerations of autonomous agents that modify their own decision rules—a philosophical echo of MOND’s challenge to a long‑standing law.

9.5 Toward a Unified Framework

A promising avenue is emergent gravity from quantum information, where spacetime and its dynamics arise from entanglement patterns. In such a picture, the MOND acceleration constant could be linked to the cosmological constant \(\Lambda\) via \(a_0 \approx c H_0 / 2\pi\) (with \(H_0\) the Hubble parameter). This tantalizing coincidence may hint at a deeper connection between the large‑scale expansion of the Universe and the low‑acceleration regime of galaxies—a research frontier that could reshape both cosmology and fundamental physics.

Why It Matters

The debate over MOND versus dark matter is not a niche curiosity; it is a litmus test for how science confronts anomalies. If a modification of gravity ultimately triumphs, it would rewrite the foundations of physics, alter our understanding of cosmic evolution, and reshape the strategies for detecting unseen mass. If dark matter prevails, the hunt for a new particle will continue to drive cutting‑edge detector technologies and inspire novel data‑analysis pipelines—tools that can be repurposed for monitoring bee populations, tracking disease vectors, or governing autonomous AI agents.

In both scenarios, the methodological lessons—rigorous hypothesis testing, cross‑disciplinary collaboration, and the use of AI to explore vast parameter spaces—are directly transferable to Apiary’s mission of safeguarding pollinators and building resilient, self‑governing AI ecosystems. By studying how the Universe may or may not need hidden mass, we sharpen the very instruments that help us protect the fragile, buzzing networks on our own planet.