SEMINARS

Our group seminar features talks on current research topics from the field of Quantum Information and Quantum Many-Body Physics, and in particular Tensor Networks, given both by group members and by external guest speakers.

If you are interested in receiving seminar announcements, please send an informal e-mail to schuch-office.quantum[at]univie.ac.at to be added to our mailing list.

› For info on past seminars click here.

2026 Summer Semester

During the 2026 summer semester, the seminar generally takes place on Thursday at 11:30 in the Erwin Schrödinger lecture hall (Boltzmanngasse 5, 5th floor). Occasionally, there might be additional seminars out of schedule as announced here.

Kshiti Sneh Rai (Leiden University)

Matrix product state approximations to quantum states of low energy-variance
Abstract: In this work, we show how to efficiently simulate pure quantum states in one dimensional systems that have both finite energy-density (i.e. in the bulk of the spectrum) and vanishingly small energy fluctuations. We analyze a tensor network algorithm that produces matrix product states whose energy-variance decreases as the bond dimension increases. Our results imply that variances as small as ∝ 1 / log N can be achieved with polynomial bond dimension. This establishes that there exist states with a very narrow support in the bulk of the spectrum that still have moderate entanglement entropy, in contrast with typical eigenstates that display a volume law. Our main technical tool is a Berry-Esseen theorem for spin systems, which strengthens central limit behavior for the energy distribution of product states.

(date/time/location: 03.03.2026, 11:30, Boltzmanngasse 5/Strudlhofgasse 4, ground floor, Kurt Gödel lecture room)

Shuhei Oyama (University of Vienna)

Generalized Cluster States and Strip 2-Algebras
Abstract: The importance of symmetry in understanding the properties of physical systems is well recognized. In particular, the presence of symmetry can impose strong constraints on the low-energy behavior of a system, as captured by Lieb-Schultz-Mattis (LSM)-type theorems. Furthermore, when a system supports nontrivial defect structures, considering how symmetry acts on those defects can also constrain the interfaces and defects allowed in the system.

In our previous work [1], by studying defects in 1+1-dimensional SPT phases protected by non-invertible symmetry, we established a bulk-boundary correspondence and classified parameterized families. In recent work, we generalized these results to 2+1 dimensions [2,3]. In particular, we constructed the 2+1-dimensional cluster state and its generalization, the generalized cluster state, and classified defects by studying representations of the symmetry acting on them, namely the strip 2-algebra.

In this seminar, I will first explain the PEPS representation of the ground state　of the (generalized) cluster state and the PEPO representation of the non-invertible symmetry, and show how to construct the partial action of the PEPO on the PEPS using action tensors. I will then use these action tensors to construct symmetry operators acting on defects, and study their representation theory. Finally, I will discuss an application of this framework: the formulation and proof of the bulk-boundary correspondence for generalized cluster states. If time permits, I will also comment on further physical consequences, including symmetry-enforced degeneracy and applications to the classification of parameterized families.

This talk is based on the following papers:
[1] arXiv:2408.15960
[2] arXiv:2601.08615
[3] arXiv:2601.08616

(date/time/location: 19.03.2026, 11:30, Boltzmanngasse 5/Strudlhofgasse 4, 5th floor, Erwin Schrödinger lecture hall)

Erickson Tjoa (Max Planck Institute of Quantum Optics)

Continuous matrix product operators for quantum fields
Abstract: Within tensor network theory, matrix product operators (MPOs) are a class of many-body operators acting on finite-dimensional lattice quantum systems that admit efficient classical description and do not create a large amount of entanglement. In this talk, we discuss an ansatz for continuous matrix product operator (cMPO) for quantum field theory. We show that (i) they admit a closed-form expression in terms of finite number of matrix valued functions without reference to any lattice parameter; (ii) they are obtained as a suitable continuum limit of matrix product operators; (iii) they preserve the entanglement area-law directly in the continuum. As an application, we use this ansatz to construct the continuum limit of matrix product unitaries beyond quantum cellular automata, and we conclude with some future directions.

(date/time/location: 26.03.2026, 11:30, Boltzmanngasse 5/Strudlhofgasse 4, 5th floor, Erwin Schrödinger lecture hall)

Alexander Jahn (Freie Universität Berlin)

Universal fault-tolerant logic with heterogeneous holographic codes
The study of holographic bulk-boundary dualities has led to the construction of novel quantum error correcting codes. Although these codes have shed new light on conceptual aspects of these dualities, they have widely been believed to lack a crucial feature of practical quantum error correction: The ability to support universal fault-tolerant quantum logic. In this work, we introduce a new class of holographic codes that realize this feature. These heterogeneous holographic codes are constructed by combining two seed codes in a tensor network on an alternating hyperbolic tiling. We show how this construction generalizes previous strategies for fault tolerance in tree-type concatenated codes, allowing one to implement non-Clifford gates fault-tolerantly on the holographic boundary. We also demonstrate that these codes allow for high erasure thresholds under a suitable heterogeneous combination of specific seed codes. Compared to previous concatenated codes, heterogeneous holographic codes achieve large overhead savings in physical qubits, e.g. a 21.8% reduction for a two-layer Steane/quantum Reed-Muller combination. Unlike standard concatenated codes, we establish that the new codes can encode more than a single logical qubit per code block by applying "black hole'' deformations with tunable rate and distance, while possessing fully addressable, universal fault-tolerant gate sets. Therefore, our work strengthens the case for the utility of holographic quantum codes for practical quantum computing.
[arXiv:2504.10386]

(date/time/location: 13.04.2026, 11:30, Boltzmanngasse 5/Strudlhofgasse 4, ground floor, Kurt Goedel lecture room)

Dimitris Saraidaris (Freie Universität Berlin)

Searching for emergent spacetime in spin glasses
Recent work on algebraic formulations of holographic dualities in terms of large N algebras has revealed a deep connection between the properties of the associated spectral functions and the emergence of a semiclassical spacetime. One of the main lessons is that, for a radial direction to emerge, the spectral function has to exhibit non-compact support. Furthermore, there exist conjectures upon a possible duality between complex gravitational configurations and glassy systems. The goal of this work is to combine these ideas by studying many-body quantum-mechanical systems and assess in which parameter regimes they could potentially be holographic. Thus, we compute the spectral functions of three many-body systems with quenched disorder, the SYK model, the p-spin model and the SU(M) Heisenberg chain in the large N limit and present results in different parameter regimes. Our main finding is that in the quantum spin glass phase of the SU(M) Heisenberg model, the spectral function develops an exponential tail, similar to the large q limit of SYK, while for the rest of the spin glass phase, compact support. In addition, we demonstrate the presence of an exponential tail in the spectral function for all cases without compact support and conformal symmetry.
[authors: Dimitris Saraidaris and Leo Shaposhnik]

(date/time/location: 13.04.2026, 11:30, Boltzmanngasse 5/Strudlhofgasse 4, ground floor, Kurt Goedel lecture room)

Leo Shaposhnik (Freie Universität Berlin)

Towards the continuum: Algebras in Infinite Tensor Networks and Holographic Codes
Abstract: The study of holographic dualities has led to intense interaction between the fields of high energy physics, quantum information theory and many-body physics. One of the outcomes has been an improved understanding of the quantum information theoretic structure of quantum gravitational systems. It was found that certain many-body systems can be described by a higher dimensional gravitational dual, as the number of degrees of local freedom diverges. This led to the idea, that the gravitational bulk description "emerges" from and describes the quantum information theoretic structure of the boundary state. This situation is similar to the thermodynamic limit of a lattice system, which can give rise to an effective description in terms of a continuum field theory, which in this language "emerges", in the limit of an infinitely dense lattice. These effective descriptions are often derived based on certain heuristics and matching of a subset of observables to continuum objects, but a more microscopic understanding "how" the continuum system arises is mostly left implicit. In parallel, tensor networks which implement quantum error correcting codes have led to model systems, holographic quantum error correction codes, that recover aspects of holographic dualities from a quantum information theoretic lens and provide microscopic models for the bulk-to-boundary encoding. It has been unclear however, to what degree these models are faithful representations of the actual mechanism underlying holographic dualities.
To gain further understanding of this relationship, we studied, if these models allow for an effective description in terms of a quantum field theory as the system becomes infinitely large or if there are fundamental limitations to the emergence of a relativistic continuum in these systems. In this talk I will describe the method we used to study this problem through an operator algebraic lens and how it led us to believe that "extensive magic" is a necessary ingredient to allow for a relativistic continuum to emerge.
This talk is based on arxiv.org/abs/2504.00096

(date/time/location: 16.04.2026, 11:30, Boltzmanngasse 5/Strudlhofgasse 4, 5th floor, Erwin Schrödinger lecture hall)

Lilith Zschetzsche (University of Vienna)

Sequential Data Loading without QRAM: A Review of Zhao et al. (2026)
Abstract: Standard quantum machine learning algorithms for massive classical datasets typically assume the availability of Quantum Random Access Memory (QRAM), an architecture that requires O(N) ancilla qubits. I will review a recent preprint [1] that proposes an alternative data-loading framework, which they term Quantum Oracle Sketching. Rather than requiring spatial storage of the dataset, the proposed algorithm processes a stochastic stream of classical data points sequentially. Through incremental phase rotations, the method allows an O(polylog N) quantum register to coherently approximate an oracle representing the underlying data distribution. I will outline the theoretical mechanism of this coherent phase accumulation and review the trade-offs it introduces.

[1] H. Zhao et al., "Exponential quantum advantage in processing massive classical data" (2026), arXiv preprint: arxiv.org/abs/2604.07639

(date/time/location: 23.04.2026, 11:30, Boltzmanngasse 5/Strudlhofgasse 4, 5th floor, Erwin Schrödinger lecture hall)

Julia Mathé (TU Wien)

Estimating entanglement monotones in spin systems using symmetries
Abstract: Entanglement in many-body systems is usually analyzed for pure ground states, but realistic systems are often mixed because of temperature, noise, or nonequilibrium dynamics. In such cases, even deciding whether a state is entangled can be difficult, let alone quantifying how much entanglement it contains. In this work, we study this problem for collective spin states by asking how far a given mixed state is from the set of fully separable states. This distance is quantified by the best separable approximation, which tells us how well the state can still be described by a classical-like mixture of unentangled particles. We derive a lower bound on this quantity from spin-squeezing inequalities built from standard collective observables, and an upper bound from an iterative algorithm that searches for the closest separable state while exploiting the symmetries of the system.

Applying these tools to thermal states of fully connected spin models, we show that this method can provide insightful quantitative information about mixed-state entanglement across different phases. In particular, we find that entanglement may appear at nonzero temperature even in regimes where the ground state itself is separable, highlighting that relevant quantum correlations can emerge beyond the usual ground-state picture. Our results also suggest that entanglement quantification may refine the usual phase diagram by revealing a finer structure within conventional phases, where states with different amounts of entanglement define distinct regimes.

Beyond providing practical methods for entanglement quantification in large systems, this approach opens the door to studying entanglement in more general settings, including out-of-equilibrium dynamics and less symmetric models such as spin chains.

(date/time/location: 30.04.2026, 11:30, Boltzmanngasse 5/Strudlhofgasse 4, 5th floor, Erwin Schrödinger lecture hall)

Lucas Winter (University of Vienna)

DysonNet: Constant-Time Local Updates for Neural Quantum States
Abstract: Neural quantum states offer a flexible variational ansatz for many-body wavefunctions, but their practical use is often limited by expensive local updates and by the difficulty of interpreting what the network has learned. In this talk, I will introduce DysonNet, a neural-network architecture designed to recover some of the contraction logic familiar from tensor networks. The key idea is to couple strictly local nonlinear features through global linear layers in a structure analogous to a truncated Dyson series. This gives local wavefunction updates a physical interpretation as scattering processes and makes it possible to precompute environment-like tensors.

Using a resummation algorithm called ABACUS, single-spin-flip updates can then be evaluated in constant time, independent of system size. I will discuss how this changes the scaling of variational Monte Carlo calculations and show benchmarks on long-range Ising and frustrated spin-chain models, where DysonNet retains state-of-the-art neural-network accuracy while removing the dominant local-update bottleneck.

(date/time/location: 07.05.2026, 11:30, Boltzmanngasse 5/Strudlhofgasse 4, 5th floor, Erwin Schrödinger lecture hall)

Alexander Wietek (Max Planck Institute for the Physics of Complex Systems)

Supersolidity and Stripes: Quantum Microscopy of the 2D Hubbard Model
Abstract: Imaging of microscopic particles allows fundamental insights into puzzling macroscopic physics. For quantum materials, scanning tunneling microscopy (STM) has played a pivotal role in unraveling the physics of high-temperature superconductors, while quantum gas microscopy with ultracold atoms allows sampling of highly entangled states of matter. In this talk, we present how, numerically, microscopy of quantum many-body systems is performed using minimally entangled typical thermal states, and demonstrate its relation to quantum gas microscopes.

Applied to the paradigmatic Hubbard model of high-temperature superconductors, the ground state stripe phase in the underdoped regime emerges from a heterogeneous charge background, with nanoscale charge clusters forming already within the pseudogap regime. In the superconducting case, the parent state features local pairing tightly bound to these charge clusters, with phase coherence across the full system emerging upon entering a supersolid ground state: a phase characterized by coexisting charge order and superconductivity. We relate these findings to STM and nuclear magnetic resonance (NMR) experiments on cuprate superconductors, corroborating a physical picture of nanoscale phase separation.

References:
A. Sinha, H. Karlsson, M. Ulaga, A. Wietek, arXiv:2603.20368 [cond-mat.str-el]
T. Chalopin et al., Proceedings of the National Academy of Sciences 123 (4), e2525539123 (2026)
A. Sinha, A. Wietek, Nat. Commun., 16, 10807 (2025)
A. Wietek, Y.-Y. He, S. R. White, A. Georges, E. M. Stoudenmire, Phys. Rev. X 11, 031007 (2021)

(date/time/location: 18.05.2026, 13:30, Boltzmanngasse 5/Strudlhofgasse 4, 3rd floor, Christian Doppler lecture hall)

Johannes Wladika (University of Vienna)

Exact Isometric Tensor Networks for 3D Stabilizer Hamiltonians
Abstract: I will present a practical method to construct exact, isometric tensor-network (TN) representations of three-dimensional , local stabilizer Hamiltonians, with an emphasis on fracton models, a family of models that host immobile excitations [1]. I construct local tensors constrained by virtual stabilizer symmetries and show how contracting a finite patch maps a symmetric virtual boundary space isometrically onto the physical ground space. The X‑Cube model [2] will be used to showcase the method in a fully worked example, highlighting which steps are generic and which are model dependent. With the exact TN obtained, I will discuss how the localization of virtual operators relates to the action of the operator on the physical space. In particular, this will reveal how certain virtual operators correspond to excitations of the X‑Cube model with restricted mobility.

References:
[1] R. Nandkishore, M. Hermele, arxiv.org/abs/1803.11196
[2] S. Vijay, J. Haah, and L. Fu, arxiv.org/abs/1603.04442

(date/time/location: 21.05.2026, 11:30, Boltzmanngasse 5/Strudlhofgasse 4, 5th floor, Erwin Schrödinger lecture hall)

Milán Rozmán (University of Vienna)

Representability of infinite-volume ground states with matrix product states
Abstract: Given a translation invariant normal Matrix Product State (MPS), it is well known that one can construct a finite-range Hamiltonian that has a unique ground state given by that MPS. After some formal adjustments, this statement can be proven to extend to the thermodynamic limit. In the late 1990s, Matsui has raised a related question [1]: under what conditions does the ground state of an infinite volume Hamiltonian have a non-approximate MPS representation? In this talk, I will give a pedagogical introduction to Matsui's theorem, primarily to the toolbox required to formulate and understand this question, including concepts like UHF algebras and infinite-volume dynamics.

[1] T. Matsui. A characterization of finitely correlated pure states. Infinite dimensional analysis and quantum probability 1, 647–661 (1998)

(date/time/location: 28.05.2026, 11:30, Boltzmanngasse 5/Strudlhofgasse 4, 5th floor, Erwin Schrödinger lecture hall)

Yuhan Liu (Max Planck Institute of Quantum Optics)

Exploring Mixed-state Quantum Phases via Analytical Tensor Network
Abstract: Understanding quantum phases of matter is a fundamental goal in physics. In this talk, I focus on mixed-state quantum phases within the framework of matrix product density operators (MPDOs). In particular, MPDOs that are renormalization fixed points (RFPs) are believed to characterize mixed-state phases of matter in one dimension, where nontrivial topological phases have already been shown to exist. I address the following questions: (1) Do MPDO RFPs admit a physical interpretation as the unique steady state (or as one of a minimally degenerate set of steady states) of a local frustration-free Lindbladian? (2) What is the mathematical framework for MPDO RFPs that incorporates anomalous phases? (3) How to go beyond the RFPs and rigorously establish phase equivalence? Addressing these questions provides unique tools to explore mixed-state quantum phases.

(date/time/location: 11.06.2026, 11:30, Boltzmanngasse 5/Strudlhofgasse 4, 5th floor, Erwin Schrödinger lecture hall)

Anna Francuz (University of Vienna)

Variational PEPS representations of the toric code in parallel magnetic fields
Abstract: The toric-code ground state admits two inequivalent PEPS representations that cannot be related by a gauge transformation. I will present both representations at zero field and highlight their differences using ZX-calculus. I will then investigate whether variational iPEPS optimization tends to favor one representation over the other, and whether this preference depends on whether the phase transition is driven by condensation of charges or fluxes.

(date/time/location: 18.06.2026, 11:30, Boltzmanngasse 5/Strudlhofgasse 4, 5th floor, Erwin Schrödinger lecture hall)

András Molnár (University of Vienna)

On the XXZ model
Abstract: The XXZ model is a prime example of an integrable model. Such models possess an extensive set of local conserved quantities that commute with the Hamiltonian. As a consequence, the Hamiltonian can be diagonalized exactly. In this talk, I will explain how both the symmetries and the integrable structure of the model can be described using tensor-network methods.

(date/time/location: 25.06.2026, 11:30, Boltzmanngasse 5/Strudlhofgasse 4, 5th floor, Erwin Schrödinger lecture hall)