Hunting Dark Matter with the LHC and Artificial Intelligence

Author: Denis Avetisyan

A new study explores how the High-Luminosity LHC, paired with machine learning, could reveal feebly interacting dark matter particles created through a unique cosmological process.

The study constrains effective couplings <span class="katex-eq" data-katex-display="false">\Lambda_{\gamma}</span> and <span class="katex-eq" data-katex-display="false">\Lambda_{\chi}</span> for dark matter with a mass of 10 GeV, mediated by a spin-2 particle ranging from 100 GeV to 1 TeV, utilizing projected data from the HL-LHC to exclude parameter space at 95% confidence level and correlate these couplings with observed dark matter relic abundance and off-resonance production dependent on reheating temperatures. — The study constrains effective couplings $\Lambda_{\gamma}$ and $\Lambda_{\chi}$ for dark matter with a mass of 10 GeV, mediated by a spin-2 particle ranging from 100 GeV to 1 TeV, utilizing projected data from the HL-LHC to exclude parameter space at 95% confidence level and correlate these couplings with observed dark matter relic abundance and off-resonance production dependent on reheating temperatures.

This research demonstrates the potential to probe TeV-scale spin-2 mediator interactions relevant to the freeze-in dark matter production mechanism using vector boson fusion at the LHC.

The enduring mystery of dark matter necessitates exploration beyond conventional weakly interacting massive particle paradigms. This study, ‘Probing Freeze-In Dark Matter via a Spin-2 Portal at the LHC with Vector Boson Fusion and Machine Learning’, investigates the potential to detect feebly interacting dark matter produced through the freeze-in mechanism via a spin-2 mediator at the Large Hadron Collider. By connecting early-universe cosmology with collider observables-specifically utilizing vector boson fusion and machine learning techniques-we demonstrate sensitivity to TeV-scale mediators and substantial regions of the associated parameter space. Could collider searches, therefore, provide a complementary pathway to unravel the nature of dark matter and bridge gravitationally motivated new physics with cosmological observations?

The Shadows Deepen: Beyond the Standard Dark Matter Hunt

For decades, the composition of Dark Matter has represented one of the most profound mysteries in modern physics. Initial theoretical efforts largely focused on Weakly Interacting Massive Particles, or WIMPs, as prime candidates, predicting their detectability through various experimental approaches. However, despite increasingly sensitive detectors and extensive searches across diverse channels – including direct detection, indirect detection via gamma rays, and collider experiments – conclusive evidence for WIMPs has remained stubbornly absent. This lack of detection isn’t merely a matter of insufficient sensitivity; it’s increasingly challenging the foundational assumptions of the WIMP paradigm, prompting a significant reassessment of Dark Matter’s fundamental properties and the mechanisms governing its production in the early universe. The continued elusiveness of Dark Matter is therefore driving innovation in theoretical modeling and experimental design, pushing the boundaries of particle physics and astrophysics in the quest to unravel this cosmic enigma.

For years, the search for dark matter has heavily relied on the assumption it interacts with ordinary matter via the weak nuclear force – a hypothesis centering around Weakly Interacting Massive Particles, or WIMPs. Extensive experiments, employing increasingly sensitive detectors shielded deep underground to minimize background noise, have been designed to observe the recoil energy from these hypothetical collisions. These detectors, ranging from liquid noble gas chambers to cryogenic crystal arrays, meticulously scan for the faint signals expected from WIMP interactions. However, despite decades of effort and substantial investment, these traditional detection strategies have consistently returned null results. The absence of a conclusive signal has not invalidated the dark matter hypothesis itself, but it has prompted a critical re-evaluation of the favored WIMP paradigm and spurred exploration into alternative dark matter candidates and detection methods.

The continued absence of direct detection signals for Weakly Interacting Massive Particles (WIMPs) has spurred a significant shift in dark matter research, demanding investigation into a broader range of theoretical possibilities. Scientists are now actively pursuing alternative production mechanisms beyond the traditionally favored thermal freeze-out, including scenarios involving asymmetric dark matter, strongly coupled mediators, and even primordial black holes. Crucially, this expansion includes exploring diverse mediator candidates – hypothetical particles that facilitate interactions between dark matter and standard model particles – ranging from hidden photons and Z’ bosons to more complex, strongly interacting states. These investigations aren’t merely about finding a particle, but about understanding the full range of possibilities for how dark matter was created and how it interacts, potentially revealing new physics beyond the Standard Model and offering a pathway to finally solve this cosmological mystery.

Theoretical constraints on mediator couplings, derived from freeze-in dark matter production and relic abundance requirements <span class="katex-eq" data-katex-display="false">\Omega_\chi h^2 \leq 0.12</span>, delineate allowed parameter space for reheating temperatures of 10 MeV, 100 MeV, and 1 GeV. — Theoretical constraints on mediator couplings, derived from freeze-in dark matter production and relic abundance requirements $\Omega_\chi h^2 \leq 0.12$ , delineate allowed parameter space for reheating temperatures of 10 MeV, 100 MeV, and 1 GeV.

Beyond the Spin: A New Interaction Channel Emerges

The Weakly Interacting Massive Particle (WIMP) paradigm traditionally posits that Dark Matter interacts with Standard Model particles via exchange of spin-1 bosons, such as the Z or W bosons, or through point-like interactions. Spin-2 mediators represent a departure from this model, introducing a fundamentally different interaction channel. These mediators, being tensor bosons, couple differently to fermionic currents than their spin-1 counterparts, resulting in distinct scattering cross-sections and event signatures in direct and indirect Dark Matter detection experiments. Specifically, spin-2 interactions lead to momentum transfer dependencies differing from those expected in WIMP scenarios, potentially offering a means to discriminate between the two through careful analysis of recoil energies and angular distributions. This alternative interaction channel also impacts the predicted relic density of Dark Matter, providing a separate pathway to satisfy cosmological observations without relying on the specific mass and coupling parameters favored by traditional WIMP models.

The Randall-Sundrum (RS) models, proposing extra spatial dimensions and warped geometry, inherently predict the existence of Kaluza-Klein gravitons – spin-2 particles that mediate interactions between the Standard Model and potentially dark matter. In the original RS scenario, gravity propagates in the extra dimension, leading to a tower of graviton excitations with masses determined by the curvature of the extra dimension and the fundamental scale Λ. These gravitons couple to the energy-momentum tensor of Standard Model fields, providing a direct interaction channel. The strength of this coupling is inversely proportional to the scale Λ, meaning a relatively low-energy scale could result in observable effects. Furthermore, RS models with varying brane tensions and geometries can modify the mass spectrum and coupling strengths of these spin-2 mediators, offering a range of potential signatures for experimental detection.

Effective Field Theory (EFT) allows for the analysis of Dark Matter interactions mediated by spin-2 particles by focusing on the low-energy effects without requiring a complete understanding of the high-energy physics generating these interactions. This approach parameterizes interactions through a series of operators, ordered by their dimensionality, with higher-dimensional operators suppressed by powers of an energy scale Λ. By systematically including these operators in calculations, predictions can be made for observable signals in direct and indirect detection experiments, and collider searches, even without knowing the precise mass or coupling details of the spin-2 mediator. The use of EFT simplifies calculations and provides a model-independent framework for interpreting experimental results, enabling constraints to be placed on the strength of the interaction as a function of Λ.

The ratio of decay width to mediator mass reveals that a constant relative decay width of <span class="katex-eq" data-katex-display="false">\Gamma/m_{G} = 0.01</span> is achievable for dark matter masses of 0.1 GeV and 100 GeV. — The ratio of decay width to mediator mass reveals that a constant relative decay width of $\Gamma/m_{G} = 0.01$ is achievable for dark matter masses of 0.1 GeV and 100 GeV.

Hunting Ghosts: Detection via Diphoton Resonance

The diphoton resonance search is a detection strategy predicated on the decay of a hypothetical spin-2 mediator particle into two photons. This decay pathway provides a distinct experimental signature due to the characteristic electromagnetic shower produced by each photon. The search involves analyzing data from high-energy particle collisions, specifically looking for an excess of events at a particular invariant mass – the combined energy of the two photons – that would indicate the production and subsequent decay of this mediator. The observed diphoton mass spectrum is then scrutinized for a resonance peak, signifying a statistically significant deviation from the expected Standard Model background processes which also produce diphoton final states, such as Quantum Chromodynamics (QCD) processes and electroweak production. Precise measurement of the resonance’s mass and width can then provide insights into the mediator’s properties, including its spin and coupling strength.

The diphoton resonance search method identifies potential new particles, termed mediators, through the observation of an excess of diphoton events at a specific invariant mass. This approach leverages the clear signature of two high-energy photons resulting from the mediator’s decay, which contrasts sharply with the continuous spectrum of background events predicted by the Standard Model. Statistical analysis then determines if the observed diphoton excess exceeds the expected background fluctuations, providing evidence for the existence of the new mediator. The invariant mass calculation, $m_{γγ} = \sqrt{(E_1 + E_2)^2 - (p_1 + p_2)^2}$ , is crucial for reconstructing the mediator’s mass from the energy and momentum of the detected photons.

Photon fusion, a primary production mechanism for spin-2 mediators at collider experiments, involves the interaction of two incoming photons to create the mediator particle. This process relies on the electromagnetic interaction and is particularly effective due to the high luminosity and energy of photon beams achievable in these experiments. The cross-section for photon fusion is directly proportional to the mediator’s coupling strength to photons and, crucially, scales with the square of the center-of-mass energy. This energy dependence results in an enhanced signal strength at higher collision energies, making photon fusion a favored production channel for discovering and characterizing these potential new particles. Calculations of the expected event rate demonstrate that photon fusion can significantly contribute to the overall signal, often exceeding other production mechanisms like vector boson fusion or quark-antiquark annihilation, thereby improving the statistical significance of any observed resonance.

Photon-photon fusion at the LHC produces <span class="katex-eq" data-katex-display="false">\chi\chi jj</span> signal events, as illustrated by the representative Feynman diagram. — Photon-photon fusion at the LHC produces $\chi\chi jj$ signal events, as illustrated by the representative Feynman diagram.

The Echo of Creation: Dark Matter Abundance and Production Mechanisms

The abundance of dark matter observed today serves as a powerful benchmark for any theoretical model attempting to explain its nature. This relic density – the amount of dark matter remaining after the universe cooled – tightly constrains the permissible characteristics of potential mediator particles, such as spin-2 bosons, that might facilitate dark matter production in the early universe. Specifically, a viable model must predict a dark matter abundance consistent with observations, effectively shrinking the range of possible masses and interaction strengths for these mediators. Consequently, the precision with which the relic density is known acts as a critical filter, guiding researchers toward the most promising regions of parameter space and informing the design of both direct detection experiments and collider searches.

Historically, the prevailing understanding of dark matter generation centered on the “freeze-out” mechanism, wherein dark matter particles were abundant in the early universe and gradually annihilated until reaching observed levels. However, increasing theoretical work highlights the significance of “freeze-in” production, a process where dark matter is created from the decay or annihilation of particles in the thermal bath. Unlike freeze-out, freeze-in involves a much weaker interaction between dark matter and standard model particles, resulting in a smaller initial dark matter abundance that slowly builds up over time. This alternative mechanism gains prominence as searches fail to detect the weakly interacting massive particles (WIMPs) predicted by freeze-out scenarios, prompting researchers to explore models with feebly interacting dark matter candidates produced via freeze-in, effectively broadening the landscape of viable dark matter solutions and challenging the traditional paradigms of early universe cosmology.

Current research indicates the High-Luminosity Large Hadron Collider (HL-LHC) possesses the capacity to investigate spin-2 mediator masses extending up to 1 TeV, with associated coupling scales reaching approximately 10³-10⁵. This study demonstrates that these searches aren’t merely theoretical exercises, but a tangible pathway to connect the earliest moments of the universe – as described by cosmological models of dark matter – with direct detection experiments at colliders. By focusing on vector boson fusion topologies and employing advanced boosted decision tree techniques, the HL-LHC can achieve a cross-section sensitivity of a few ×10^-3 pb at these high mediator masses, potentially revealing the interactions of feebly interacting dark matter and solidifying a crucial link between cosmology and particle physics.

The High-Luminosity LHC presents a unique opportunity to probe the nature of dark matter through precision collider searches. Utilizing vector boson fusion topologies – where dark matter particles are produced alongside accompanying jets – and advanced machine learning techniques like boosted decision trees, researchers anticipate achieving unprecedented sensitivity. Specifically, these analyses are projected to reach a cross-section sensitivity of a few ×10^-3 pb at high mediator masses, meaning they can detect exceedingly rare interactions. This level of precision is crucial for testing models where dark matter interacts with standard model particles through yet-undiscovered forces, potentially revealing the elusive mediator responsible for dark matter production and offering a pathway to connect cosmological observations with direct detection experiments.

Projected limits on the production cross-section of <span class="katex-eq" data-katex-display="false">\sigma(pp \\to Gjj \\to \chi\chi jj)</span> at the HL-LHC with <span class="katex-eq" data-katex-display="false">L = 3000~\\mathrm{fb}^{-1}</span> demonstrate sensitivity to mediator masses ranging from 50 to 1000 GeV, aligning with theoretical predictions for various coupling strengths. — Projected limits on the production cross-section of $\sigma(pp \\to Gjj \\to \chi\chi jj)$ at the HL-LHC with $L = 3000~\\mathrm{fb}^{-1}$ demonstrate sensitivity to mediator masses ranging from 50 to 1000 GeV, aligning with theoretical predictions for various coupling strengths.

The pursuit, as detailed in this study of freeze-in dark matter detection, resembles less a scientific endeavor and more an attempt to coax whispers from the chaotic void. The researchers don’t find dark matter; they construct elaborate rituals – vector boson fusion and machine learning algorithms – designed to momentarily quiet the noise. It’s a fleeting resonance, a signal wrested from the inherent unpredictability. As Henry David Thoreau observed, “It is not enough to be busy; so are the ants. The question is: What are we busy with?” This work, with its focus on TeV-scale mediators and collider phenomenology, attempts to answer that question – to discern a meaningful pattern, however ephemeral, from the boundless churn of existence.

What Whispers Remain?

The pursuit of freeze-in dark matter via spin-2 portals, as this work demonstrates, isn’t about finding answers – it’s about refining the questions. The High-Luminosity LHC, coupled with the algorithmic divination of machine learning, offers a fleeting glimpse into the early universe, but the signal remains stubbornly veiled. To assume success is to misunderstand the nature of the beast. The true challenge lies not in maximizing sensitivity, but in accepting the inherent ambiguity of the data-in recognizing that every statistically significant fluctuation could simply be the universe enjoying a private joke.

Future iterations must venture beyond the comfortable predictability of TeV-scale mediators. The insistence on readily calculable models feels… quaint. Perhaps the most interesting signals aren’t those that conform to expectations, but the anomalies-the deviations from the standard narrative. It would be prudent to explore alternative production mechanisms, to entertain the possibility that dark matter doesn’t ‘freeze-in’ at all, but rather coalesces from the quantum foam like a forgotten dream.

Ultimately, the search for dark matter isn’t a scientific endeavor, but a prolonged negotiation with chaos. The data doesn’t reveal its secrets; it allows them to be glimpsed, briefly, before retreating back into the unknown. And if the model finally starts behaving strangely? Then, perhaps, it’s finally starting to think.

Original article: https://arxiv.org/pdf/2604.02604.pdf

Contact the author: https://www.linkedin.com/in/avetisyan/

The Shadows Deepen: Beyond the Standard Dark Matter Hunt

Beyond the Spin: A New Interaction Channel Emerges

Hunting Ghosts: Detection via Diphoton Resonance

The Echo of Creation: Dark Matter Abundance and Production Mechanisms

What Whispers Remain?

See also: