Learnijoy
ResearchAI ResearchAI Engineering & DevTools

Neural Spline Flows Aid Dark Matter Search in CMS Data.

Hitesh Rasineni (VIT-AP University, Amaravati, India), Bhavishya Chebrolu (Mohan Babu University, Tirupati, India)· July 16, 2026 View original

Summary

This paper reports a search for dark matter produced with a leptonically decaying Z boson using CMS Run 2015D open data and Neural Spline Flows. The method models signal and background densities to set upper limits on signal-strength parameters for various dark matter mediators, though observed limits are weaker than expected due to background modeling discrepancies.

Scientists have conducted a search for dark matter (DM) produced in conjunction with a Z boson that decays into leptons, utilizing open data from the CMS experiment's 2015D run at the Large Hadron Collider. The analysis focused on events where a Z boson decays into either muons or electrons, alongside missing transverse energy indicative of dark matter. A key innovation in this search is the application of Neural Spline Flows (NSFs). Five independent NSFs were trained to model the densities of both Standard Model background processes and specific dark matter signal scenarios. This allowed for the construction of a per-event test statistic based on the log-likelihood ratio, providing sensitivity across the entire kinematic phase space without relying on arbitrary thresholds. While the search yielded upper limits on dark matter signal strengths, the observed limits were weaker than anticipated, primarily due to a persistent discrepancy in background modeling at high missing transverse energy, rather than evidence of a dark matter signal.

Why it matters

This research demonstrates a novel application of Neural Spline Flows in high-energy physics for dark matter searches, showcasing advanced AI techniques for complex data analysis in scientific discovery.

How to implement this in your domain

  1. 1Explore the use of Neural Spline Flows or other density estimation techniques for anomaly detection in complex datasets.
  2. 2Apply likelihood-ratio scoring methods to improve sensitivity in searches for rare events or signals.
  3. 3Investigate advanced machine learning techniques for background modeling in scientific or industrial data analysis.
  4. 4Collaborate with data scientists to leverage open scientific datasets for developing and testing new AI methodologies.

Who benefits

Scientific ResearchData ScienceAI/ML DevelopmentAerospaceDefense

Key takeaways

  • Neural Spline Flows are applied to model signal and background densities in dark matter searches.
  • The method uses a likelihood-ratio test statistic for broad kinematic sensitivity.
  • Upper limits on dark matter signal strengths were set using CMS open data.
  • Background modeling discrepancies impacted the observed limits.

Original post by Hitesh Rasineni (VIT-AP University, Amaravati, India), Bhavishya Chebrolu (Mohan Babu University, Tirupati, India)

"arXiv:2607.13771v1 Announce Type: new Abstract: We report a search for dark matter (DM) produced in association with a leptonically decaying \(Z\) boson at \(\sqrt{s}=13\) TeV using CMS Run 2015D open data corresponding to an integrated luminosity of \(2.32\,\mathrm{fb}^{-1}\) to…"

View on X

Originally posted by Hitesh Rasineni (VIT-AP University, Amaravati, India), Bhavishya Chebrolu (Mohan Babu University, Tirupati, India) on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses

More in AI Research

AI Engineering & DevToolsAI Research

NodeImport Improves Imbalanced Node Classification on Graphs

NodeImport is a new framework addressing class imbalance in graph node classification by assessing node importance to create a balanced meta-set for training. It dynamically filters valuable labeled, unlabeled, and synthetic nodes, outperforming existing baselines across various datasets and GNN architectures.

Nan Chen, Zemin Liu, Bryan Hooi, Bingsheng He, Jun Hu, Jia ChenJul 16, 2026
AI Engineering & DevToolsAI Research

Multiplex Graph Transformer Boosts Power Grid Model Generalization.

Researchers introduce MxGPS, a multiplex graph transformer designed to overcome "topology overfitting" in power grid problems. By jointly training on multiple tasks with a shared encoder, MxGPS achieves superior zero-shot generalization across unseen grid topologies, demonstrating high accuracy and low boundary violation rates with significantly fewer parameters.

Charilaos Papaioannou, Ioannis Tsantilas, Dimitris Giannakakos, Vasilis Michalakopoulos, Sotiris Pelekis, Vangelis Marinakis, Arsam Aryandoust, Antonello Monti, Ricardo J. Bessa, Perdo P. Vergara, Jochen Cremer, Elissaios SarmasJul 16, 2026
AI ResearchAI Engineering & DevTools

Algebraic Representability Explains Grokking's Limiting Behavior.

This research explores grokking in neural networks trained on modular arithmetic, focusing on networks with holomorphic monomial activations. It demonstrates that when the network's expressible function class collapses to a finite-dimensional algebraic variety, grokking disappears, and outcomes become binary: instant success or outright failure based on algebraic representability.

Chon-Fai Kam, Xavier Cadet, Miloud Bessafi, Frederic CadetJul 16, 2026