Simulating matrix models with tensor networks
Enrico M. Brehm, Yibin Guo, Karl Jansen, and Enrico Rinaldi · 2024
Read Simulating matrix models with tensor networks by Enrico M. Brehm, Yibin Guo, Karl Jansen, and Enrico Rinaldi (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-04133.txt.
Introduction
Enrico Brehm, Yibin Guo, Karl Jansen, and Enrico Rinaldi test tensor networks as a tool to simulate matrix models, in work from 2024. Matrix models are quantum systems with no space built in. They give insight into gauge and gravity theories and, through holography, may describe quantum black holes.
Some ranges of these models cannot be solved by hand. The team builds ground states as matrix product states, a compact tensor form. They study traits such as how entanglement is spread in those states. The work checks whether tensor methods can open hard ranges of matrix models to study, much as they have for other quantum systems.
Catalog record
| CREATOR | Enrico M. Brehm, Yibin Guo, Karl Jansen, and Enrico Rinaldi |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 73 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2412-04133
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Simulating matrix models with tensor networks Enrico M. Brehm, Yibin Guo, Karl Jansen, and Enrico Rinaldi Abstract Matrix models, as quantum mechanical systems without explicit spatial dependence, provide valuable insights into higher-dimensional gauge and gravitational theories, especially within the framework of string theory, where they can describe quantum black holes via the holographic principle. Simulating these models allows for exploration of their kinematic and dynamic properties, particularly in parameter regimes that are analytically intractable. In this study, we examine the potential of tensor network techniques for such simulations. Specifically, we construct ground states as matrix product states and analyse features such as their entanglement structure. 1 Introduction Matrix models are quantum mechanical models whose degrees of freedom are organised in matrices. Although they do not directly depend on space, they can offer a way to study theories in higher dimensions, for example supersymmetric gauge theories. In particular, they can provide a non-perturbative formulation of certain quantum field theories and string theories, allowing for a more comprehensive understanding of these theories and their dynamics. See Ref. [ 1 ] for a recent review. The gauge/gravity duality [ 2 ] , which establishes a deep connection between problems in quantum and gravitational physics, has further motivated the study of matrix models. This duality maps certain strongly-coupled quantum field theories to weakly-curved gravitational theories in higher dimensions. Via this duality, the study of matrix models can also discover insights into gravity in higher dimensions and may, for example, reveal more about features of black holes. Simulating higher-dimensional quantum field theories and string theories poses significant challenges. Discretising these systems on a spatial lattice is a somewhat artificial construction, which may make it difficult to capture the full underlying physics. Leveraging dualities and constructions from string theory can offer a more physical realisation of higher-dimensional quantum field theories, potentially leading to more efficient and accurate simulations. Particularly, matrix model constructions are free from any need of discretising spacetime and may open a path to simulate aspects of higher-dimensional quantum field theories and string theory. The study of matrix models has a rich history, dating back to the early days of string theory and quantum gravity research. Notable examples include the IKKT model [ 3 ] , which provides a non-perturbative definition of type IIB superstring theory, and the BFSS model [ 4 , 5 ] , which describes M-theory in the infinite momentum frame and will be discussed in Sec. 2.1 . These models have massively advanced our understanding of string theory and quantum gravitational theories in general. The complexity of matrix models, particularly in the large- N limit, where N is the size of the matrices, necessitates the development of powerful computational techniques. Traditional methods, such as perturbative expansions in large N , often fall short in capturing the full non-perturbative dynamics of these systems. This has led to more and more interest in applying advanced numerical methods to study matrix models, ranging from Monte Carlo simulations (see, e.g., Ref. [ 6 , 7 , 8 , 9 ] ), to bootstrap methods (see, e.g.,Ref. [ 10 , 11 ] ), to machine learning algorithms (see, e.g., Ref. [ 12 , 13 , 14 ] ), and, recently, to quantum computing approaches (see, e.g., Ref. [ 15 , 13 ] ). In this work we want to focus on tensor network methods as a versatile framework for approximating and simulating matrix models. Tensor networks, originally developed in the context of condensed matter physics to study strongly correlated quantum systems [ 16 ] , have emerged as a promising tool for tackling the computational challenges in other areas of physics [ 17 ] . They are also a promising tool to tackle the challenges posed by matrix models: the ability of tensor networks to efficiently represent and manipulate high-dimensional quantum states makes them well-suited for exploring the large Hilbert spaces characteristic of matrix models at larger N . Our analysis aims to shed light on the strengths and limitations of tensor network method when applied to matrix models. We will examine aspects such as computational efficiency, scalability to larger matrix sizes, and the ability to capture specific physical effects and observables. By doing so, we hope to provide valuable insights into the most effective strategies for simulating matrix models, ultimately contributing to the understanding of quantum field theories, string theory, and the nature of quantum gravity. Our results provide a comprehensive overview of the convergence of tensor network methods as matrix models grow in size and complexity, both in terms of the size N and the number of matrices, as well as the truncation of the local Hilbert space. In practice, we provide a guide to the approximation error of these numerical methods as a function of the bond dimension in the entire complexity landscape of bosonic and fermionic matrix models. The remainder of the paper is organised as follows. In Sec. 2 , we provide an overview of two important matrix models, the BFSS and BMN models, and their connection to string theory. We then show how to simulate matrix models in the Hamiltonian formulation in Sec. 3 and discuss how tensor networks methods are applied to the task of finding the ground state. Our results across a range of models are presented in Sec. 4 , including an analysis of the entanglement structure of the ground states and of the computational efficiency of the tensor network simulations. The numerical results look very promising. Our results align with previous findings from simulations employing machine learning and quantum computing approaches, as e.g. presented in Ref. [ 12 , 13 , 14 ] and Ref. [ 15 , 13 ] . However, our analysis of the computational cost for finding ground states indicate a polynomial scaling with the number of bosonic degrees of freedom, enabling us to access matrix sizes and quantities beyond the reach of these methods. This demonstrates that tensor networks are a promising and scalable tool for simulating matrix models involving many large matrices. Finally, we conclude in Sec. 5 with a summary of our findings and an outlook on future research directions. More technical details are provided in the appendices. 2 Matrix models and the connection to String/M-Theory As teased already, matrix models play a crucial role in string theory, offering unique insights into non-perturbative regimes that are often challenging to explore through conventional methods. They serve as powerful tools for probing the fundamental structure of string theory and its various manifestations. In this section, we present a short overview of two important matrix models and their diverse contexts within string theory without assuming too much background in the latter. The rich duality structure in string theory allows these models to emerge in remarkably varying scenarios, showcasing their fundamental importance and versatility. By examining these models, we can gain a deeper understanding of the interconnected nature of different formulations of string theory and potentially uncover new ways for exploring quantum gravity. 2.1 The BFSS matrix model Among the first matrix models that appeared in the context of string theory was the so-called BFSS matrix model [ 4 , 5 ] . It is the effective description of the dynamics of interacting