Simultaneous Discovery of Conservation Laws and Hidden Particles with Smith Matrix Decomposition

Oliver SchulteIJCAI 2009 · OpenReview

Abstract

Particle physics experiments, like the Large Hadron Collider in Geneva, can generate thousands of data points listing detected particle reactions. An important learning task is to analyze the reaction data for evidence of conserved quantities and hidden particles. This task involves latent structure in two ways: first, hypothesizing hidden quantities whose conservation determines which reactions occur, and second, hypothesizing the presence of hidden particles. We model this problem in the classic linear algebra framework of automated scientific discovery due to Valdes-Perez, Zytkow and Simon, where both reaction data and conservation laws are represented as matrices. We introduce a new criterion for selecting a matrix model for reaction data: find hidden particles and conserved quantities that rule out as many interactions among the nonhidden particles as possible. A polynomial-time algorithm for optimizing this criterion is based on the new theorem that hidden particles are required if and only if the Smith Normal Form of the reaction matrix R contains entries other than 0 or 1. To our knowledge this is the first application of Smith matrix decomposition to a problem in AI. Using data from particle accelerators, we compare our algorithm to the main model of particles in physics, known as the Standard Model: our algorithm discovers conservation laws that are equivalent to those in the Standard Model, and indicates the presence of a hidden particle (the electron antineutrino) in accordance with the Standard Model.

Automated Scientific Discovery in Particle Physics

In particle physics, experiments produce massive amounts of data regarding which particles interact and how. Traditionally, physicists analyze these reactions to infer two primary components of physical reality: conservation laws (rules like the conservation of electric charge) and hidden particles (entities like neutrinos that are not directly detected but are required to satisfy those laws).

This paper introduces the Latent Nullspace Algorithm (LNA), an automated framework designed to look at a list of observed reactions and determine the most restrictive set of laws and the minimum number of hidden entities required to explain why other, unobserved reactions do not occur. By framing the problem through linear algebra, the researchers represent reactions as vectors and conservation laws as a matrix $Q$ such that $QR^T = 0$.

The Latent Nullspace Algorithm (LNA)

The core innovation of the LNA is the application of Smith Matrix Decomposition (or Smith Normal Form, SNF) to solve a problem in AI and physics. The algorithm is guided by a "maximally strict" criterion, inspired by Murray Gell-Mann's "totalitarian principle": anything not prohibited by a conservation law must happen. Therefore, the best model is the one that forbids as many unobserved reactions as possible.

The paper proves a novel theorem: a hidden particle is mathematically required if and only if the Smith Normal Form of the reaction matrix contains entries other than 0 or 1. If the SNF contains an entry greater than 1, the algorithm iteratively adds a "hidden" dimension—a new particle—to "fix" the matrix until the model reaches maximal strictness. This polynomial-time algorithm ($O(n^7)$) effectively shifts "forbidden" reactions out of the reachable integer span of observed reactions.

Recovering the Standard Model

The algorithm was evaluated using data from the Review of Particle Physics, encompassing 193 particles and 205 established reactions. Remarkably, the LNA successfully reconstructed the fundamental conservation laws used in physics: electric charge, baryon number, and the three lepton numbers (electron, muon, and tau).

Table 1: Standard Model quantum numbers for various particles, including Charge, Baryon number, and Lepton numbers (Tau, Electron, Muon).

As shown in the table above, the algorithm discovers a set of laws mathematically equivalent to the Standard Model's CBEMT (Charge, Baryon, Electron, Muon, Tau) numbers. This demonstrates that these physical constants are not just theoretical constructs but algebraic necessities for a model that perfectly separates possible reactions from impossible ones using integer-based laws.

Predicting Hidden Particles

In a "blind" test where the electron antineutrino was removed from the dataset, the LNA correctly flagged that the resulting model was no longer "maximally strict." It successfully indicated the mathematical necessity of a hidden particle to restore consistency, mirroring the historical discovery of the neutrino. Despite the theoretical complexity of the algorithm, an optimized implementation processed the full dataset of 205 reactions in approximately three minutes on standard hardware.

Conclusion and Future Directions

While the LNA currently assumes noise-free input data, it represents a significant step toward automated scientific discovery. It proves that the existence of certain particles and laws can be derived directly from reaction data through matrix decomposition. Future work aims to adapt the framework to handle the probabilistic and noisy data generated by the Large Hadron Collider, potentially allowing the algorithm to search for "New Physics" beyond the current Standard Model.

Summary updated

Abstract

Particle physics experiments, like the Large Hadron Collider in Geneva, can generate thousands of data points listing detected particle reactions. An important learning task is to analyze the reaction data for evidence of conserved quantities and hidden particles. This task involves latent structure in two ways: first, hypothesizing hidden quantities whose conservation determines which reactions occur, and second, hypothesizing the presence of hidden particles. We model this problem in the classic linear algebra framework of automated scientific discovery due to Valdes-Perez, Zytkow and Simon, where both reaction data and conservation laws are represented as matrices. We introduce a new criterion for selecting a matrix model for reaction data: find hidden particles and conserved quantities that rule out as many interactions among the nonhidden particles as possible. A polynomial-time algorithm for optimizing this criterion is based on the new theorem that hidden particles are required if and only if the Smith Normal Form of the reaction matrix R contains entries other than 0 or 1. To our knowledge this is the first application of Smith matrix decomposition to a problem in AI. Using data from particle accelerators, we compare our algorithm to the main model of particles in physics, known as the Standard Model: our algorithm discovers conservation laws that are equivalent to those in the Standard Model, and indicates the presence of a hidden particle (the electron antineutrino) in accordance with the Standard Model.
BibTeX
@inproceedings{schulte2009simultaneous,
  author = {Oliver Schulte},
  title = {{Simultaneous Discovery of Conservation Laws and Hidden Particles with Smith Matrix Decomposition}},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year = {2009},
  url = {https://openreview.net/forum?id=SkEGg4GObr},
}
Figure