Entanglement across scales: Quantics tensor trains as a natural framework for renormalization

Stefan Rohshap, Jheng-Wei Li, Alena Lorenz, Serap Hasil, Karsten Held, Anna Kauch, Markus Wallerberger · 2025

Read Entanglement across scales: Quantics tensor trains as a natural framework for renormalization by Stefan Rohshap, Jheng-Wei Li, Alena Lorenz, Serap Hasil, Karsten Held, Anna Kauch, Markus Wallerberger (2025) 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-2507-19069.txt.

Introduction

Stefan Rohshap and co-authors published a 2025 physics paper linking two ways of thinking about how information is shared across scales. Entanglement usually describes ties between particles or places. The authors focus instead on ties between length scales: how much the small-scale and large-scale parts of a system know about each other. That question had drawn far less study.

A computing technique called quantics tensor trains compresses hard calculations by splitting them across scales. The team shows this technique is, in exact terms, a form of renormalization, the classic method of zooming out step by step. At each zoom step, the number of new couplings produced matches exactly a key size measure in the tensor method. They prove this on a simple chain model of electrons hopping between sites. The result gives the computing tool a firm physical meaning, and may cut costs in theory work as well.

Catalog record

CREATORStefan Rohshap, Jheng-Wei Li, Alena Lorenz, Serap Hasil, Karsten Held, Anna Kauch, Markus Wallerberger
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE135 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-19069

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Entanglement across scales: Quantics tensor trains as a natural framework for renormalization Preprint: APS/123-QED Stefan Rohshap, Jheng-Wei Li, Alena Lorenz, Serap Hasil, Karsten Held, Anna Kauch, Markus Wallerberger Abstract Understanding entanglement remains one of the most intriguing problems in physics. While particle and site entanglement have been studied extensively, the investigation of length or energy scale entanglement — quantifying the information exchange between different length scales — has received far less attention. Here, we identify the quantics tensor train (QTT) technique, a matrix product state-inspired approach for overcoming computational bottlenecks in resource-intensive numerical calculations, as a renormalization group method by analytically expressing an exact cyclic reduction-based real-space renormalization scheme in QTT language, which serves as a natural formalism for the method. In doing so, we precisely match the QTT bond dimension — a measure of length scale entanglement — to the number of rescaled couplings generated in each coarse-graining renormalization step. While QTTs have so far been applied almost exclusively to numerical problems in physics, our analytical calculations demonstrate that they are also powerful tools for mitigating computational costs in semi-analytical treatments. We present our results for the one-dimensional tight-binding model with 𝐧 \mathbf{n} -th-nearest-neighbor hopping, where the 𝟐 ​ 𝐧 \mathbf{2n} rescaled couplings generated in the renormalization procedure precisely match the QTT bond dimension of the one-particle Green’s function. I Introduction Romanesco broccoli and the neuron-based brain architecture Smith et al. 2021 are two striking examples of naturally occurring fractals, exhibiting patterns that repeat across multiple scales. The scale invariance inherent in such structures directly connects to the concept of the renormalization group (RG) in physics, which leverages coarse-graining and rescaling to study complex systems. Since the early spin-blocking ideas Kadanoff 1966 , the formulation of real-space RG methods Wilson 1975 has laid the foundation for powerful techniques in quantum field theory, including the numerical renormalization group Bulla et al. 2008 ; Weichselbaum et al. 2009 ; Kugler et al. 2021 ; Ritz et al. 2025 , functional renormalization group Wetterich 1993 ; Kugler and von Delft 2018 ; Hille et al. 2020 ; Bippus et al. 2025a , and density matrix renormalization group (DMRG) White 1992 ; White 1993 ; Schollwöck 2005 approaches. Although the matrix product state (MPS) Baxter 1968 representation was originally developed independently, its integration into DMRG allowed the method to become the state-of-the-art for obtaining low-energy physics in one-dimensional lattice systems. The success of DMRG for treating one-dimensional short-ranged correlated systems relies on a bound on the strength of quantum entanglement dictated by the correlation length via the entropic area law Cho 2018 , which ensures efficient representation of ground states as MPSs with low bond dimensions. However, near criticality, entanglement can grow large. This problem was addressed by the development of the multiscale entanglement renormalization ansatz (MERA) Vidal 2007 ; Vidal 2008 ; Crosswhite et al. 2008 . MERA implements a quantum circuit with logarithmic depth that effectively coarse-grains the system, renormalizing entanglement organized in layers corresponding to different length scales. Beyond its role in quantum criticality Pfeifer et al. 2009 and holography Swingle 2012 , entanglement renormalization has also been linked to wavelet transforms Evenbly and White 2016 ; Haegeman et al. 2018 . Notably, the development of wavelets Mallat 1989 ; Daubechies 1990 was influenced by RG concepts from the outset, and the scope of this technique was only recently extended to efficiently compress correlation functions in a numerical study Moghadas et al. 2024 . In a parallel development, the compressibility of correlators was studied in the same model in Ref. Rohshap et al. 2025a , following a different approach with the quantics tensor train (QTT) Oseledets 2009 ; Oseledets 2011 ; Khoromskij 2011 ; Dolgov et al. 2012 ; Khoromskij 2018 technique. Besides studying the compressibility of correlators with QTTs, also involved two-particle calculations, namely the parquet equations, were performed within the QTT framework. Given that QTTs were identified as algebraic wavelet transforms with adaptively determined filters Oseledets and Tyrtyshnikov 2011 , it is unsurprising that both approaches allow the tackling of similar problems. This connection suggests a deeper link between QTTs and MERA, as both exploit the separation of exponentially different length or energy scales. Figure 1: QTT decomposition with separation of exponentially different length scales as a natural framework of real-space RG methods. QTTs have rapidly emerged as practical tools across various fields. The method has proven useful not only in the study of turbulence Gourianov et al. 2022 ; Peddinti et al. 2024 ; Kornev et al. 2023 ; Hölscher et al. 2025 ; Gourianov et al. 2025 , plasma physics Ye and Loureiro 2022 and quantum chemistry Jolly et al. 2025 , but especially in quantum many-body physics Shinaoka et al. 2023 ; Erpenbeck et al. 2023 ; Ritter et al. 2024 ; Ishida et al. 2024 ; Murray et al. 2024 ; Takahashi et al. 2025 ; Eckstein 2024 ; Otero Fumega et al. 2024 ; Frankenbach et al. 2025 ; Niedermeier et al. 2025 . In this context, the QTT approach enables the treatment of highly resolved grids by approximating input data with only a logarithmic number of parameters, offering the potential to outperform state-of-the-art sparse modeling approaches Eckhardt et al. 2020 ; Astretsov et al. 2020 ; Wentzell et al. 2020 ; Krien et al. 2020 ; Krien et al. 2021 or compact representations in a suitable basis Wallerberger et al. 2021 ; Shinaoka et al. 2020 ; Shinaoka et al. 2018 ; Kiese et al. 2025 . The success of the QTT method depends on the amount of length scale entanglement in the system, which characterizes the flow of information between different scales and, thus, the separability of those length scales. Despite its importance, this notion of entanglement, distinct from site or particle entanglement Li and Waintal 2024 ; Bellomia et al. 2024 ; Bippus et al. 2025b ; Bippus et al. 2025c , has received little attention. Hitherto only in a recent work has time scale entanglement been shown to be a system-inherent property that peaks at phase transitions, enabling QTT-based diagnostics of critical points Rohshap et al. 2025b . In this paper, we bridge the gap between length scale entanglement, QTTs, and RG by analytically demonstrating that the QTT framework naturally represents a coarse-graining real-space renormalization procedure Bini and Meini 1999 for the n -th-nearest-neighbor tight-binding model (see Fig. 1 for comparison). Leveraging mathematical methods for analyzing the QTT bond dimension of Toeplitz matrices Zamarashkin et al. 2009 ; Oseledets et al. 2011 ; Kazeev and Khoromskij 2012 ; Kazeev et al. 2013 ; Vysotsky and Rakhuba 2023 , we recast this RG scheme into QTT language in an exact manner, showing that the QTT bond dimension of the single-particle Green’s function precisely matches the number of rescaled couplings generated in the RG procedure. While QTTs have proven powerful in compressing data, their approximation accuracy relative to the achieved compression rate is generally not known a priori in numerical treatments. Here, we derive analytical error bounds for the approximate QTT construction of the one-particle Green’s function, specifically for infinitely decaying couplings, as a function of the chosen maximum bond dimension. Finally, we complement our analytical results with detailed numerical studies,

Read the complete text