arXiv subjects

Stefan Wolf

Publications and source records attributed to Stefan Wolf.

At least 19 recordsLinked to original sources

arXiv2026

Thermodynamic-limit dispersion relations on trapped-ion quantum hardware

We run a numerical linked-cluster expansion with a quantum algorithm (NLCE+QA), computing ground-state energies and one quasi-particle dispersions in the thermodynamic limit using a 20-qubit trapped-ion quantum processing unit (QPU). The NLCE+QA framework extracts thermodynamic-limit properties from small-cluster calculations, making it naturally suited for near-term quantum devices. Projector-based block-diagonalization schemes such as projective cluster-additive transformation (PCAT) are essential to NLCE+QA, and they involve matrix inversion and square root operations that amplify measurement noise. A central question is therefore whether current hardware can provide expectation values that are accurate enough to withstand non-linear classical post-processing. We explore this challenge for the transverse-field Ising model (TFIM) in one dimension, on a ladder geometry, as well as in a longitudinal field in one dimension. For the quantum algorithm, we consider adiabatic state preparation (ASP), as well as a variational quantum eigensolver (VQE) trained on a classical device. The final expectation values are obtained from the QPU, using a novel alternative to the Hadamard test that we name the CX-test. We explore the regimes currently attainable on quantum devices and comment on the improvements needed for quantum computers to achieve results beyond classical reach.

quant-ph↗

arXiv2026

Shot-noise reduction for lattice Hamiltonians

Efficiently estimating energy expectation values of quantum lattice systems on quantum computers is a crucial subroutine for various quantum algorithms, which can lead to significant overhead due to the high measurement shot numbers required. We introduce a measurement strategy tailored to quantum lattice systems and (noisy) energy eigenstates. It is based on a geometric partitioning of the Hamiltonian into local patches, and performing the measurements in the eigenbases of those patches. The resulting energy estimator has a smaller variance than the ones of Pauli grouping schemes, which leads to a reduction of the total number of shots. We provide rigorous guarantees for this variance improvement for energy eigentstates, also in the presence of depolarizing noise. As one can choose the subsystem size, one can ensure that measurement circuits remain within implementable depths. In numerical experiments, we demonstrate the shot count reduction for various 2D lattice models, including the transverse field XY and Ising models, as well as the Fermi-Hubbard model. We find sampling improvements of several orders of magnitude already for plaquettes of two by two qubits, where the required readout circuits remain very moderate in depth.

quant-ph↗

arXiv2026

Variational Time Evolution Compression for Solving Impurity Models on Quantum Hardware

Dynamical mean-field theory (DMFT) is a useful tool to analyze models of strongly correlated fermions like the Hubbard model. In DMFT, the lattice of the model is replaced by a single impurity site embedded in an effective bath. The resulting single impurity Anderson model (SIAM) can then be solved self-consistently with a quantum-classical hybrid algorithm. This procedure involves repeatedly preparing the ground state on a quantum computer and evolving it in time to measure the Greens function. We here develop an approximation of the time evolution operator for this setting by training a Hamiltonian variational ansatz. The parameters of the ansatz are obtained via a variational quantum algorithm that utilizes a small number of time steps, given by the Suzuki-Trotter expansion of the time evolution operator, to guide the evolution of the parameters. The resulting circuit has a fixed depth for the time evolution depending on the size of the bath and is significantly shallower than a comparable Suzuki-Trotter expansion.

quant-ph↗

arXiv2025

Measuring the Effect of Background on Classification and Feature Importance in Deep Learning for AV Perception

Common approaches to explainable AI (XAI) for deep learning focus on analyzing the importance of input features on the classification task in a given model: saliency methods like SHAP and GradCAM are used to measure the impact of spatial regions of the input image on the classification result. Combined with ground truth information about the location of the object in the input image (e.g., a binary mask), it is determined whether object pixels had a high impact on the classification result, or whether the classification focused on background pixels. The former is considered to be a sign of a healthy classifier, whereas the latter is assumed to suggest overfitting on spurious correlations. A major challenge, however, is that these intuitive interpretations are difficult to test quantitatively, and hence the output of such explanations lacks an explanation itself. One particular reason is that correlations in real-world data are difficult to avoid, and whether they are spurious or legitimate is debatable. Synthetic data in turn can facilitate to actively enable or disable correlations where desired but often lack a sufficient quantification of realism and stochastic properties. [...] Therefore, we systematically generate six synthetic datasets for the task of traffic sign recognition, which differ only in their degree of camera variation and background correlation [...] to quantify the isolated influence of background correlation, different levels of camera variation, and considered traffic sign shapes on the classification performance, as well as background feature importance. [...] Results include a quantification of when and how much background features gain importance to support the classification task based on changes in the training domain [...]. Download: synset.de/datasets/synset-signset-ger/background-effect

cs.CV↗

arXiv2025

Synset Signset Germany: a Synthetic Dataset for German Traffic Sign Recognition

In this paper, we present a synthesis pipeline and dataset for training / testing data in the task of traffic sign recognition that combines the advantages of data-driven and analytical modeling: GAN-based texture generation enables data-driven dirt and wear artifacts, rendering unique and realistic traffic sign surfaces, while the analytical scene modulation achieves physically correct lighting and allows detailed parameterization. In particular, the latter opens up applications in the context of explainable AI (XAI) and robustness tests due to the possibility of evaluating the sensitivity to parameter changes, which we demonstrate with experiments. Our resulting synthetic traffic sign recognition dataset Synset Signset Germany contains a total of 105500 images of 211 different German traffic sign classes, including newly published (2020) and thus comparatively rare traffic signs. In addition to a mask and a segmentation image, we also provide extensive metadata including the stochastically selected environment and imaging effect parameters for each image. We evaluate the degree of realism of Synset Signset Germany on the real-world German Traffic Sign Recognition Benchmark (GTSRB) and in comparison to CATERED, a state-of-the-art synthetic traffic sign recognition dataset.

cs.CV↗

arXiv2025

Graphical Tests of Causality

Bell inequalities limit the possible observations of non-communicating parties. Here, we present analogous inequalities for any number of communicating parties under the causal constraints of static causal order, definite causal order, and bi-causal order. All derived inequalities are remarkably simple. They correspond to upper bounds on the winning chance in graphical games: Given a specific directed graph over the parties, the parties are challenged to communicate along a randomly chosen arc. In the case of definite causal order, every game that we find is specified by a kefalopoda digraph. Based on this we define weakly causal correlations as those that satisfy all kefalopoda inequalities. We show that the problem of deciding whether some correlations are weakly causal is solvable in polynomial time in the number of parties.

quant-ph↗

arXiv2025

Flow of dynamical causal structures with an application to correlations

Causal models capture cause-effect relations both qualitatively - via the graphical causal structure - and quantitatively - via the model parameters. They offer a powerful framework for analyzing and constructing processes. Here, we introduce a tool - the flow of causal structures - to visualize and explore the dynamical aspect of classical-deterministic processes, arguably like those present in general relativity. The flow describes all possible ways in which the causal structure of a process can evolve. We also present an algorithm to construct its supergraph - the superflow - from the causal structure only, without invoking the model parameters. As an application, we show that if all leaves of a flow are trivial, then the corresponding process produces causal correlations only, i.e., correlations where future data cannot influence past events. This strengthens the result that processes, where every directed cycle in their causal structure is chordless, establish causal correlations only. We also discuss the main difficulties for the quantum generalization of the present algorithms.

quant-ph↗

arXiv2025

Quantum-foundational implications of information erasure upon measurement

A projective measurement cannot decrease the von Neumann entropy if the outcome is ignored. However, under certain sound assumptions and using the quantum violation of Leggett-Garg inequalities, we have previously demonstrated that this property is not inherited by a classical simulation of such a measurement process. In the simulation, a measurement erases prior information by partially resetting the system, suggesting that the quantum-state update following a measurement cannot be entirely epistemic. The erasure of information has been proved by assuming that the maximally mixed quantum state corresponds to maximal ignorance of the classical state. A more intricate proof employed the weaker hypothesis that the entropy is finite at some stage of the simulation. In this paper, we focus on the quantum-foundational implications of this theorem. We first provide a simple proof by directly using the second hypothesis. Second, we identify information erasure as the mechanism breaking the time symmetry in ontological theories. This symmetry break has been previously proved by Pusey and Leifer. Third, we show that information erasure and, thus, symmetry break can be avoided by employing a branching a la many-worlds theory. The information flow and the time asymmetry are transferred to the measurement devices and the subsequent comparison of results, which inherently involve time-asymmetric processes. Thus, causality and the absence of information erasure suggest that measurements have multiple actual outcomes. Similarly, Deutsch and Hayden argued that Bell's theorem leads to the same conclusion if locality is given for granted. We conclude by showing that the problem of the clumsiness loophole in an experimental Leggett-Garg test of macrorealism is mitigated by the information-erasure theorem.

quant-ph↗

arXiv2025

Local theories with parallel realities and the epistemic view of the quantum state

Hidden-variable theories effectively solve the measurement problem. However, a serious issue of this route towards a realistic completion of quantum theory is raised by Bell's proof that the resulting theories are nonlocal. A possible resolution is to reject the assumption that measurements have single actual outcomes. Indeed, relaxing this premise, Deutsch and Hayden showed that Bell's theorem can be evaded by delaying the buildup of the correlations until the parties compare their outcomes at a meeting point. However, the Deutsch-Hayden theory, which is deterministic and psi-ontic, leads to an infinite information flow towards the meeting point. Furthermore, alternative branches are weighted by amplitudes, leading to interpretative issues. In this paper, we introduce a general framework that combines the randomness of single-world theories with the coexistence of diverse instances, as found in many-worlds theory. This framework incorporates the existing theories as limiting cases. We explore how this hybrid approach addresses key challenges of single-world and Deutsch-Hayden theories. On one hand, the multiplicity of coexisting instances allows us to circumvent nonlocality and, possibly, contextuality. On the other hand, randomness makes it possible to derive quantum probabilities from unweighted counts of instances and ensemble averages. Furthermore, it can lead to a reduction of the information flow. We illustrate this framework with a local model for two spatially separate maximally entangled qubits. The model requires two unweighted instances and a finite information flow -- just one bit per measurement is communicated to the meeting point. Setting aside its foundational motivations, this framework has also relevance in quantum communication complexity and leads to novel technical questions, potentially providing new insights into some peculiarities of entanglement.

quant-ph↗

arXiv2025

Temperature as Joules per Bit

In statistical mechanics, entropy is defined as a fundamental quantity. However, its unit, J/K, involves that of temperature, which is only subsequently defined - and defined in terms of entropy. This circularity arises with the introduction of Boltzmann's constant into the very expression of entropy. The J/K carried by the constant prevents entropy from finding a unit of its own while simultaneously obfuscating its informational nature. Following the precepts of information theory, we argue that entropy is well measured in bits and coincides with information capacity at thermodynamic equilibrium. Consequently, not only is the temperature of a system in equilibrium expressed in J/bit, but it acquires a clear meaning: It is the cost in energy to increase its information capacity by 1 bit. Viewing temperature as joules per bit uncovers the strong duality exhibited by Gibbs long ago between available capacity and free energy. It also simplifies Landauer's cost and clarifies that it is a cost of displacement, not of erasure. Replacing the kelvin with the bit as an SI unit would remove Boltzmann's constant from the seven defining constants.

quant-ph↗

arXiv2025

On multivariate polynomials achievable with quantum signal processing

Quantum signal processing (QSP) is a framework which was proven to unify and simplify a large number of known quantum algorithms, as well as discovering new ones. QSP allows one to transform a signal embedded in a given unitary using polynomials. Characterizing which polynomials can be achieved with QSP protocols is an important part of the power of this technique, and while such a characterization is well-understood in the case of univariate signals, it is unclear which multivariate polynomials can be constructed when the signal is a vector, rather than a scalar. This work uses a slightly different formalism than what is found in the literature, and uses it to find simpler necessary conditions for decomposability, as well as a sufficient condition -- the first, to the best of our knowledge, proven for a (generally inhomogeneous) multivariate polynomial in the context of quantum signal processing.

quant-ph↗

arXiv2024

Solving an Industrially Relevant Quantum Chemistry Problem on Quantum Hardware

Quantum chemical calculations are among the most promising applications for quantum computing. Implementations of dedicated quantum algorithms on available quantum hardware were so far, however, mostly limited to comparatively simple systems without strong correlations. As such, they can also be addressed by classically efficient single-reference methods. In this work, we calculate the lowest energy eigenvalue of active space Hamiltonians of industrially relevant and strongly correlated metal chelates on trapped ion quantum hardware, and integrate the results into a typical industrial quantum chemical workflow to arrive at chemically meaningful properties. We are able to achieve chemical accuracy by training a variational quantum algorithm on quantum hardware, followed by a classical diagonalization in the subspace of states measured as outputs of the quantum circuit. This approach is particularly measurement-efficient, requiring 600 single-shot measurements per cost function evaluation on a ten qubit system, and allows for efficient post-processing to handle erroneous runs.

quant-ph↗

arXiv2022

Unconditional Proofs-of-Work and Other Possibilities of Thermodynamic Cryptography

In line with advances in recent years about realizing cryptographic functionalities in an information-theoretically secure way from physical phenomena and laws, we propose here to obtain useful tasks from the sole assumption of limited free energy. Specifically, based on that assumption -- resulting in a setting loosely related to Maurer's bounded-storage model -- we derive protocols for unconditional proofs-of-thermodynamical-work, secret sharing of free energy, unforgeable money, and proofs-of-position. While our schemes can be considered classical and not quantum per se, they are resistant against both classes of adversaries.

quant-ph↗

arXiv2022

An Algebraic-Geometry Approach to Prime Factorization

New algorithms for prime factorization that outperform the existing ones or take advantage of particular properties of the prime factors can have a practical impact on present implementations of cryptographic algorithms that rely on the complexity of factorization. Currently used keys are chosen on the basis of the present algorithmic knowledge and, thus, can potentially be subject to future breaches. For this reason, it is worth to investigate new approaches which have the potentiality of giving a computational advantage. The problem has also relevance in quantum computation, as an efficient quantum algorithm for prime factorization already exists. Thus, better classical asymptotic complexity can provide a better understanding of the advantages offered by quantum computers. In this paper, we reduce the factorization problem to the search of points of parametrizable varieties, in particular curves, over finite fields. The varieties are required to have an arbitrarily large number of intersection points with some hypersurface over the base field. For a subexponential or poly- nomial factoring complexity, the number of parameters have to scale sublinearly in the space dimension n and the complexity of computing a point given the parameters has to be subexponential or polynomial, respectively. We outline a procedure for building these varieties, which is illustrated with two constructions. In one case, we show that there are varieties whose points can be evaluated efficiently given a number of parameters not greater than n/2. In the other case, the bound is dropped to n/3. Incidentally, the first construction resembles a kind of retro-causal model. Retro-causality is considered one possible explanation of quantum weirdness.

cs.CR↗

arXiv2022

Key Agreement and Oblivious Transfer from Free-Energy Limitations

We propose one of the very few constructive consequences of the second law of thermodynamics. More specifically, we present protocols for secret-key establishment and multiparty computation the security of which is based fundamentally on Landauer's principle. The latter states that the erasure cost of each bit of information is at least kTln2 (where k is Boltzmann's constant and T is the absolute temperature of the environment). Albeit impractical, our protocols explore the limits of reversible computation, and the only assumption about the adversary is her inability to access a quantity of free energy that is exponential in the one of the honest participants. Our results generalize to the quantum realm.

quant-ph↗

arXiv2022

Thermodynamics as Combinatorics: A Toy Theory

We discuss a simple toy model which allows, in a natural way, for deriving central facts from thermodynamics such as its fundamental laws, including Carnot's version of the second principle. Our viewpoint represents thermodynamic systems as binary strings, and it links their temperature to their Hamming weight. From this, we can reproduce the possibility of negative temperatures, the notion of equilibrium as the coïncidence of two notions of temperature - statistical versus structural -, as well as the zeroth law of thermodynamics (transitivity of the thermal-equilibrium relation), which we find to be redundant, as other authors, yet at the same time not to be universally valid.

quant-ph↗

arXiv2020

On the Advantage of Irreversible Processes in Single-System Games

The CHSH no-signalling game studies Bell nonlocality by showcasing a gap between the win rates of classical strategies, quantum-entangled strategies, and no-signalling strategies. Similarly, the CHSH* single-system game explores the advantage of irreversible processes by showcasing a gap between the win rates of classical reversible strategies, quantum reversible strategies, and irreversible strategies. The irreversible process of erasure rules supreme for the CHSH* single-system game, but this ``erasure advantage'' does not necessarily extend to every single-system game: We introduce the 32-Game, in which reversibility is irrelevant and only the distinction between classical and quantum operations matters. We showcase our new insight by modifying the CHSH* game to make it erasure-immune, while conserving its quantum advantage. We conclude by the reverse procedure: We tune the 32-Game to make it erasure-vulnerable, and erase its quantum advantage in the process. The take-home message is that, when the size of the single-system is too small for Alice to encode her whole input, quantum advantage and erasure advantage can happen independently.

quant-ph↗

arXiv2020

Measuring Measuring

Measurements play a crucial role in doing physics: Their results provide the basis on which we adopt or reject physical theories. In this note, we examine the effect of subjecting measurements themselves to our experience. We require that our contact with the world is empirically warranted. Therefore, we study theories that satisfy the following assumption: Interactions are accounted for so that they are empirically traceable, and observations necessarily go with such an interaction with the observed system. Examining, with regard to these assumptions, an abstract representation of measurements with tools from quantum logic leads us to contextual theories. Contextuality becomes a means to render interactions, thus also measurements, empirically tangible. The measurement becomes problematic---also beyond quantum mechanics---if one tries to commensurate the assumption of tangible interactions with the notion of a spectator theory, i.e., with the idea that measurement results are read off without effect. The problem, thus, presents itself as the collision of different epistemological stances with repercussions beyond quantum mechanics.

quant-ph↗