06.10.2026 Sebastian Paeckel (LMU Munich)
Location: HIT E 41.1, Time: 12:00
Unlocking Low-Energy Many-Body Spectra with Dynamic Krylov Frames
The recent years have seen enormous experimental progress in designing and controlling strongly correlated quantum materials. At the same time, this progress poses new challenges for theoretical and numerical methods. A particularly pressing problem is the efficient computation of low-energy spectra in strongly correlated systems beyond one spatial dimension.
In this talk, I will introduce dynamic non-unitary Krylov frames as a new framework for addressing this challenge. Building on the recent success of complex-time methods, dynamic non-unitary Krylov frames provide a unifying theoretical framework and a foundation for developing highly efficient numerical approaches to many-body spectral problems.
As an example, I will consider two-hole bound states in the two-dimensional t-J model. I will show how exploiting the structure of dynamic non-unitary Krylov frames enables resolutions at low energies that are inaccessible with conventional approaches. These results demonstrate how the framework can bridge the gap between effective microscopic theories and experimentally relevant energy scales, opening new perspectives for understanding pairing mechanisms in unconventional superconductors.
13.10.2026 Thea Budde (ETH Zurich)
Location: HIT E 41.1, Time: 12:00
Ergodicity breaking from unconventional lattice symmetry
Statistical physics rests on the assumption that many-body systems uniformly explore all available states. The model is only constrained by its symmetries, which are sufficient to predict its long-time behavior. This raises the question of what counts as a symmetry. Traditional notions of symmetry rely on quantities with a straightforward analog in relativistic continuum theories. However, there are broader classes of conserved quantities on the lattice for which this is not the case, such as subsystem symmetries and partial isometries. These can explain why several models seem to break ergodicity and fail to thermalize as was expected when only traditional symmetries are considered. In this talk, I will present examples of models with quantum many-body scars and Hilbert space fragmentation, whose behavior was previously classified as ergodicity breaking, but can instead be explained by unconventional symmetries. These results raise two questions: does all ergodicity breaking originate in unconventional conserved quantities, and what would be a useful classification of lattice conserved quantities in this context?
20.10.2026 Matija Medvidović (ETH Zurich)
Location: HIT E 41.1, Time: 12:00
Interacting electrons through the lens of effective potentials
Density functional theory (DFT) is one of the most widely used approaches to interacting electrons in molecules and materials. Its central construction replaces the interacting system with independent electrons moving in an effective potential that reproduces the ground-state density. In practice, the quality of this description depends on approximations to exchange and correlation. Understanding where these approximations succeed or fail requires looking beyond total energies. In this talk, I will introduce the Kohn–Sham construction and discuss what we can learn from the effective potential itself. I will then describe two routes to computing it: optimizing a local potential to minimize an approximate energy functional, or reconstructing one from an independently computed electronic density. Both routes face numerical difficulties that have limited their practical use. In recent work, we address these problems using “multipole splats”, an expressive nonlinear representation of the potential. Optimizing the potential turns DFT into a Hamiltonian learning problem. I will show how this approach reveals spatial patterns of exchange-correlation errors as molecular bonds break, and how the potential’s long-range tail determines whether it supports the correct spectrum. Together, these examples provide a natural real-space view of the physics encoded in familiar DFT approximations, using modern machine learning methods.
27.10.2026 Tim Heightman (ICFO)
Location: HIT E 41.1, Time: 12:00
Foundation Neural Quantum States
Foundation neural quantum states are founded in the idea that a single neural model should be able to serve multiple Hamiltonians. Thus far for spin-based systems, foundation models have fixed an interaction topology (say a spin-chain Ising model) and vary within the parametric space of the Hamiltonian, sometimes for a changing system size. In our recent work, we show it is possible to have one model train over hundreds of thousands of different Hamiltonians, changing the interaction topology (e.g. from 1D to 2D to arbitrary non-local interactions), number of bodies, and the weighting of interactions. Our results indicate favorable scaling laws in pretraining, in which we used contemporary techniques from large language models and deep reinforcement learning on systems up to 64 qubits. Our deep reinforcement learning algorithm learns the entanglement geometry of systems in pretraining, finding it to generalize on systems with known entangling structures that it never saw in training. On a held-out generalisation dataset, we test our model on unseen systems, unseen scales, and fine-tune our model on system sizes of 2000 qubits and beyond.
To do this, we formulate qubit quantum ground-state learning as manifold variational optimization over centrally odd scalar functions on SU(2)^N. This replaces explicit Hilbert-space vector amplitudes with manifold functions on which the Hamiltonian acts through Lie derivatives, evaluated by custom automatic differentiation primitives. We prove that the resulting variational principle on this manifold preserves the usual spin-1/2 sector’s ground-state upper bound using the Peter-Weyl theorem and justify the choice of such a representation with a no-go theorem for pure state foundation NQS. We then pre-train our foundation model on a dataset of hundreds of thousands of different Hamiltonian systems, varying the connection topology, system size, interaction types and strengths, bringing together a century of many-body literature. Code for our method, as well as the model weights are available and open-source.
03.11.2026 Ilaria Maccari (ETH Zurich)
Location: HIT E 41.1, Time: 12:00
Unconventional electron condensates: from pairs to quadruplets
Over the last few years, experimental results have revealed signatures of unconventional electron condensates that go beyond the standard Cooper pair paradigm [1,2]. In multicomponent and unconventional superconductors, the proliferation of topological phase excitations —such as vortices, skyrmions, and domain walls— can indeed destroy the phase coherence among Cooper pairs while preserving coherence among pairs of Cooper pairs. The resulting beyond-mean-field phases are characterized by a four-electron order parameter and accompanied by spontaneous symmetry breaking. In the first part of this talk, I will briefly review the topological mechanisms and phenomenological Ginzburg–Landau models that support the emergence of these electron-quadruplet condensates [3-5].
In the second part, I will focus on bridging these phenomenological descriptions with underlying, material-specific microscopic ingredients. In systems lacking multidimensional irreducible representations, time-reversal symmetry breaking (TRSB) typically arises from phase frustration among multiple superconducting gaps residing on different Fermi surfaces; however, the precise microscopic triggers remain poorly understood. Starting from a microscopic two-orbital model, we identify the key normal-state properties—specifically band filling, spin–orbit coupling, and orthorhombic distortions—that promote phase frustration potentially driving the onset of TRSB superconductivity [6]. We apply this framework to d-electron materials and establish a realistic microscopic mechanism for the tunable TRSB relevant for iron-based superconductors.
References:
[1] Grinenko et al., Nature Physics 17, 1254-1259 (2021).
[2] I. Shipulin et al., Nature Communications 14, 6734 (2023).
[3] I. Maccari, E. Babaev, Phys. Rev. B 105 (21), 214520 (2022).
[4] I. Maccari, J. Carlström, E. Babaev, Phys. Rev. B 107 (6), 064501 (2023).
[5] I. Maccari, E. Babaev, J. Carlström, Physical Review B 113 (1), 014501 (2026).
[6] I. Maccari, A. Ramires, arXiv:2510.16843 (2025).
17.11.2026 Wenxiang Hu (Paul Scherrer Institute and ETH Zurich)
Location: HIT E 41.1, Time: 12:00
Ultrafast phase-locked free-electron laser pulses for X-ray quantum optics
The invention and development of optical laser mode-locking have revolutionized various fields, including spectroscopy, quantum optics and the study of ultrafast dynamics. Conversely, free-electron lasers (FELs), as accelerator-based photon sources, have the advantage of a widely tunable output frequency and a high brilliance. However, they typically suffer from limited coherence.
In this presentation, I will describe the recent progress in generating ultrafast phase-locked pulses using FELs, covering soft [1,2] to hard [3] X-ray regimes. The key challenges, temporal coherence and waveform control, are addressed by different techniques to manipulate the electron beam including laser-based seeding and electron-beam shaping. These novel modes will create new opportunities for coherent X-ray spectroscopies and X-ray quantum optics.
[1] E. Prat et al., Phys. Rev. Lett. 133, 205001 (2024).
[2] W. Hu et al., Phys. Rev. Lett. 135, 265001 (2025); Physics Today 79 (3), 8-10 (2026).
[3] W. Hu et al., arXiv:2508.00455.