Manuel Schrauth
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Manuel Schrauth has a total of 20 co-authors. Here are the top co-authors by number of shared publications:
- Jefferson S. E. Portela
University of Würzburg
5 shared publications - Florian Goth
Würzburg Institute for Theoretical Physics and Astrophysics, University of Würzburg
3 shared publications Dietmar Herdt
University of Würzburg
2 shared publicationsFelix Dusel
University of British Columbia
2 shared publications- Felix Kreyß
Fraunhofer Institute for Integrated Circuits (IIS), 91058 Erlangen, Germany
2 shared publications - Moritz Schreiber
Fraunhofer Institute for Integrated Circuits (IIS), 91058 Erlangen, Germany
2 shared publications
Latest Publications
- Open Access
- 28 November 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15 (23) , 12615
Self-Organized Neural Network Inference in Dynamic Edge Networks
Inference of large machine learning models can quickly exceed the capabilities of edge devices in terms of performance, memory or energy consumption. When offloading computations to a cloud server is not possible or feasible, for instance, due to data sovereignty concerns or latency constraints, a solution can be to distribute the inference load across multiple devices in a local edge network. We propose an approach which is capable of orchestrating multi-stage inference tasks in a mobile ad-hoc network consisting of heterogeneous devices in a self-organized and fully distributed manner. As individual edge devices may be battery-powered and volatile, the framework ensures a high degree of reliability even in dynamic environments. In particular, new nodes are automatically and seamlessly integrated into the ensemble, rendering the approach highly scalable. Moreover, resilience against spontaneous node dropouts or connection failures is implemented through adaptive task rerouting. Finally, by enabling complex inference tasks to be processed in small segments on the most suitable hardware available in the network, the ensemble is able to attain considerable pipelining performance and energy efficiency.
- 28 November 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15 (23) , 12615
Open AccessSelf-Organized Neural Network Inference in Dynamic Edge Networks
Inference of large machine learning models can quickly exceed the capabilities of edge devices in terms of performance, memory or energy consumption. When offloading computations to a cloud server is not possible or feasible, for instance, due to data sovereignty concerns or latency constraints, a solution can be to distribute the inference load across multiple devices in a local edge network. We propose an approach which is capable of orchestrating multi-stage inference tasks in a mobile ad-hoc network consisting of heterogeneous devices in a self-organized and fully distributed manner. As individual edge devices may be battery-powered and volatile, the framework ensures a high degree of reliability even in dynamic environments. In particular, new nodes are automatically and seamlessly integrated into the ensemble, rendering the approach highly scalable. Moreover, resilience against spontaneous node dropouts or connection failures is implemented through adaptive task rerouting. Finally, by enabling complex inference tasks to be processed in small segments on the most suitable hardware available in the network, the ensemble is able to attain considerable pipelining performance and energy efficiency.
- Open Access
Hyperbolic tiling neighborhoods in O(1) time
Tilings of the hyperbolic plane are of significant interest among many branches of mathematics, physics and computer science. Yet, their construction remains a non-trivial task. Current approaches primarily use tree-based recursive algorithms, which are fundamentally limited: they do not readily yield the neighborhood graph representing cell adjacencies, which is however required for many applications. We introduce a novel approach that allows to build hyperbolic tilings and their associated graph structure simultaneously, using only combinatoric rules without requiring an explicit coordinate representation. This allows to generate arbitrarily large, exact hyperbolic graphs, with an algorithmic complexity that does not depend on the lattice size. We provide an easy-to-use implementation which substantially outperforms existing methods, hence rendering ultra large-scale numerical simulations on these geometric structures accessible for the scientific community.
Open AccessHyperbolic tiling neighborhoods in O(1) time
Tilings of the hyperbolic plane are of significant interest among many branches of mathematics, physics and computer science. Yet, their construction remains a non-trivial task. Current approaches primarily use tree-based recursive algorithms, which are fundamentally limited: they do not readily yield the neighborhood graph representing cell adjacencies, which is however required for many applications. We introduce a novel approach that allows to build hyperbolic tilings and their associated graph structure simultaneously, using only combinatoric rules without requiring an explicit coordinate representation. This allows to generate arbitrarily large, exact hyperbolic graphs, with an algorithmic complexity that does not depend on the lattice size. We provide an easy-to-use implementation which substantially outperforms existing methods, hence rendering ultra large-scale numerical simulations on these geometric structures accessible for the scientific community.
- Open Access
- 28 August 2024
- Published by Stichting SciPost in SciPost Physics Codebases
- p. 034
HYPERTILING — a high performance Python library for the generation and visualization of hyperbolic lattices
HYPERTILING is a high-performance Python library for the generation and visualization of regular hyperbolic lattices embedded in the Poincaré disk model. Using highly optimized, efficient algorithms, hyperbolic tilings with millions of vertices can be created in a matter of minutes on a single workstation computer. Facilities including computation of adjacent vertices, dynamic lattice manipulation, refinements, as well as powerful plotting and animation capabilities are provided to support advanced uses of hyperbolic graphs. In this manuscript, we present a comprehensive exploration of the package, encompassing its mathematical foundations, usage examples, applications, and a detailed description of its implementation.
- 28 August 2024
- Published by Stichting SciPost in SciPost Physics Codebases
- p. 034
Open AccessHYPERTILING — a high performance Python library for the generation and visualization of hyperbolic lattices
HYPERTILING is a high-performance Python library for the generation and visualization of regular hyperbolic lattices embedded in the Poincaré disk model. Using highly optimized, efficient algorithms, hyperbolic tilings with millions of vertices can be created in a matter of minutes on a single workstation computer. Facilities including computation of adjacent vertices, dynamic lattice manipulation, refinements, as well as powerful plotting and animation capabilities are provided to support advanced uses of hyperbolic graphs. In this manuscript, we present a comprehensive exploration of the package, encompassing its mathematical foundations, usage examples, applications, and a detailed description of its implementation.
- Open Access
- 28 August 2024
- Published by Stichting SciPost in SciPost Physics Codebases
- p. 034
Codebase release 1.3 for HYPERTILING
HYPERTILING is a high-performance Python library for the generation and visualization of regular hyperbolic lattices embedded in the Poincaré disk model. Using highly optimized, efficient algorithms, hyperbolic tilings with millions of vertices can be created in a matter of minutes on a single workstation computer. Facilities including computation of adjacent vertices, dynamic lattice manipulation, refinements, as well as powerful plotting and animation capabilities are provided to support advanced uses of hyperbolic graphs. In this manuscript, we present a comprehensive exploration of the package, encompassing its mathematical foundations, usage examples, applications, and a detailed description of its implementation.
- 28 August 2024
- Published by Stichting SciPost in SciPost Physics Codebases
- p. 034
Open AccessCodebase release 1.3 for HYPERTILING
HYPERTILING is a high-performance Python library for the generation and visualization of regular hyperbolic lattices embedded in the Poincaré disk model. Using highly optimized, efficient algorithms, hyperbolic tilings with millions of vertices can be created in a matter of minutes on a single workstation computer. Facilities including computation of adjacent vertices, dynamic lattice manipulation, refinements, as well as powerful plotting and animation capabilities are provided to support advanced uses of hyperbolic graphs. In this manuscript, we present a comprehensive exploration of the package, encompassing its mathematical foundations, usage examples, applications, and a detailed description of its implementation.
- 21 July 2023
- Published by Springer Nature
- p. 145-154
Enabling Distributed Inference of Large Neural Networks on Resource Constrained Edge Devices using Ad Hoc Networks
- 21 July 2023
- Published by Springer Nature
- p. 145-154
Enabling Distributed Inference of Large Neural Networks on Resource Constrained Edge Devices using Ad Hoc Networks
- Open Access
- 2 March 2023
- Published by American Physical Society (APS) in Physical Review Letters
- Vol. 130 (9) , 091604
Breitenlohner-Freedman Bound on Hyperbolic Tilings
We establish how the Breitenlohner-Freedman (BF) bound is realized on tilings of two-dimensional Euclidean Anti–de Sitter space. For the continuum, the BF bound states that on Anti–de Sitter spaces, fluctuation modes remain stable for small negative mass squared . This follows from a real and positive total energy of the gravitational system. For finite cutoff , we solve the Klein-Gordon equation numerically on regular hyperbolic tilings. When , we find that the continuum BF bound is approached in a manner independent of the tiling. We confirm these results via simulations of a hyperbolic electric circuit. Moreover, we propose a novel circuit including active elements that allows us to further scan values of above the BF bound.- 2 March 2023
- Published by American Physical Society (APS) in Physical Review Letters
- Vol. 130 (9) , 091604
Open AccessBreitenlohner-Freedman Bound on Hyperbolic Tilings
We establish how the Breitenlohner-Freedman (BF) bound is realized on tilings of two-dimensional Euclidean Anti–de Sitter space. For the continuum, the BF bound states that on Anti–de Sitter spaces, fluctuation modes remain stable for small negative mass squared . This follows from a real and positive total energy of the gravitational system. For finite cutoff , we solve the Klein-Gordon equation numerically on regular hyperbolic tilings. When , we find that the continuum BF bound is approached in a manner independent of the tiling. We confirm these results via simulations of a hyperbolic electric circuit. Moreover, we propose a novel circuit including active elements that allows us to further scan values of above the BF bound. - 16 December 2019
- Published by American Physical Society (APS) in Physical Review E
- Vol. 100 (6) , 062118
Universality of continuous phase transitions on random Voronoi graphs
The Voronoi construction is ubiquitous across the natural sciences and engineering. In statistical mechanics, however, only its dual, the Delaunay triangulation, has been considered in the investigation of critical phenomena. In this paper we set to fill this gap by studying three prominent systems of classical statistical mechanics, the equilibrium spin-1/2 Ising model, the nonequilibrium contact process, and the conserved stochastic sandpile model on two-dimensional random Voronoi graphs. Particular motivation comes from the fact that these graphs have vertices of constant coordination number, making it possible to isolate topological effects of quenched disorder from node-intrinsic coordination number disorder. Using large-scale numerical simulations and finite-size scaling techniques, we are able to demonstrate that all three systems belong to their respective clean universality classes. Therefore, quenched disorder introduced by the randomness of the lattice is irrelevant and does not influence the character of the phase transitions. We report the critical points to considerable precision and, for the Ising model, also the first correction-to-scaling exponent.
- 16 December 2019
- Published by American Physical Society (APS) in Physical Review E
- Vol. 100 (6) , 062118
Universality of continuous phase transitions on random Voronoi graphs
The Voronoi construction is ubiquitous across the natural sciences and engineering. In statistical mechanics, however, only its dual, the Delaunay triangulation, has been considered in the investigation of critical phenomena. In this paper we set to fill this gap by studying three prominent systems of classical statistical mechanics, the equilibrium spin-1/2 Ising model, the nonequilibrium contact process, and the conserved stochastic sandpile model on two-dimensional random Voronoi graphs. Particular motivation comes from the fact that these graphs have vertices of constant coordination number, making it possible to isolate topological effects of quenched disorder from node-intrinsic coordination number disorder. Using large-scale numerical simulations and finite-size scaling techniques, we are able to demonstrate that all three systems belong to their respective clean universality classes. Therefore, quenched disorder introduced by the randomness of the lattice is irrelevant and does not influence the character of the phase transitions. We report the critical points to considerable precision and, for the Ising model, also the first correction-to-scaling exponent.
- Open Access
Relevance of topological disorder on the directed percolation phase transition
Despite decades of research, the precise role of topological disorder in critical phenomena has yet to be fully understood. A major contribution has been the work by Barghathi and Vojta, which uses spatial correlations to explain puzzling earlier results. However, due to its reliance on coordination number fluctuations, their criterion cannot be applied to constant-coordination lattices, raising the question, for which classes of transitions this type of disorder can be a relevant perturbation. In order to cast light on this question, we investigate the non-equilibrium phase transition of the two-dimensional contact process on different types of spatial random graphs with varying or constant coordination number. Using large-scale numerical simulations, we find the disorder to be relevant, as the dynamical scaling behaviour turns out to be non-conventional, thus ruling out the directed percolation universality class. We conjecture that the relevance of topological disorder is linked to how strongly connected the lattice is. Based on this assumption, we design two analysis tools that succeed in qualitatively distinguishing relevant from non-relevant topological disorders, supporting our conjecture and possibly pointing the way to a more complete relevance criterion.
Open AccessRelevance of topological disorder on the directed percolation phase transition
Despite decades of research, the precise role of topological disorder in critical phenomena has yet to be fully understood. A major contribution has been the work by Barghathi and Vojta, which uses spatial correlations to explain puzzling earlier results. However, due to its reliance on coordination number fluctuations, their criterion cannot be applied to constant-coordination lattices, raising the question, for which classes of transitions this type of disorder can be a relevant perturbation. In order to cast light on this question, we investigate the non-equilibrium phase transition of the two-dimensional contact process on different types of spatial random graphs with varying or constant coordination number. Using large-scale numerical simulations, we find the disorder to be relevant, as the dynamical scaling behaviour turns out to be non-conventional, thus ruling out the directed percolation universality class. We conjecture that the relevance of topological disorder is linked to how strongly connected the lattice is. Based on this assumption, we design two analysis tools that succeed in qualitatively distinguishing relevant from non-relevant topological disorders, supporting our conjecture and possibly pointing the way to a more complete relevance criterion.
- 4 September 2018
- Published by American Physical Society (APS) in Physical Review Letters
- Vol. 121 (10) , 100601
Violation of the Harris-Barghathi-Vojta Criterion
In 1974, Harris proposed his celebrated criterion: Continuous phase transitions in -dimensional systems are stable against quenched spatial randomness whenever , where is the clean critical exponent of the correlation length. Forty years later, motivated by violations of the Harris criterion for certain lattices such as Voronoi-Delaunay triangulations of random point clouds, Barghathi and Vojta put forth a modified criterion for topologically disordered systems: , where is the disorder decay exponent, which measures how fast coordination number fluctuations decay with increasing length scale. Here we present a topologically disordered lattice with coordination number fluctuations that decay as slowly as those of conventional uncorrelated randomness, but for which the clean universal behavior is preserved, hence violating even the modified criterion.- 4 September 2018
- Published by American Physical Society (APS) in Physical Review Letters
- Vol. 121 (10) , 100601
Violation of the Harris-Barghathi-Vojta Criterion
In 1974, Harris proposed his celebrated criterion: Continuous phase transitions in -dimensional systems are stable against quenched spatial randomness whenever , where is the clean critical exponent of the correlation length. Forty years later, motivated by violations of the Harris criterion for certain lattices such as Voronoi-Delaunay triangulations of random point clouds, Barghathi and Vojta put forth a modified criterion for topologically disordered systems: , where is the disorder decay exponent, which measures how fast coordination number fluctuations decay with increasing length scale. Here we present a topologically disordered lattice with coordination number fluctuations that decay as slowly as those of conventional uncorrelated randomness, but for which the clean universal behavior is preserved, hence violating even the modified criterion.- Open Access
Diffusion and first-passage characteristics on a dynamically evolving support
We propose a generalized diffusion equation for a flat Euclidean space subjected to a continuous infinitesimal scale transform. For the special cases of an algebraic or exponential expansion/contraction, governed by time-dependent scale factors $a(t)\sim t^\lambda$ and $a(t)\sim \exp(\mu t)$, the partial differential equation is solved analytically and the asymptotic scaling behavior, as well as the dynamical exponents, are derived. Whereas in the algebraic case the two processes (diffusion and expansion) compete and a crossover is observed, we find that for exponential dynamics the expansion dominates on all time scales. For the case of contracting spaces, an algebraic evolution slows down the overall dynamics, reflected in terms of a new effective diffusion constant, whereas an exponential contraction neutralizes the diffusive behavior entirely and leads to a stationary state. Furthermore, we derive various first-passage properties and describe four qualitatively different regimes of (strong) recurrent/transient behavior depending on the scale factor exponent.Open AccessDiffusion and first-passage characteristics on a dynamically evolving support
We propose a generalized diffusion equation for a flat Euclidean space subjected to a continuous infinitesimal scale transform. For the special cases of an algebraic or exponential expansion/contraction, governed by time-dependent scale factors $a(t)\sim t^\lambda$ and $a(t)\sim \exp(\mu t)$, the partial differential equation is solved analytically and the asymptotic scaling behavior, as well as the dynamical exponents, are derived. Whereas in the algebraic case the two processes (diffusion and expansion) compete and a crossover is observed, we find that for exponential dynamics the expansion dominates on all time scales. For the case of contracting spaces, an algebraic evolution slows down the overall dynamics, reflected in terms of a new effective diffusion constant, whereas an exponential contraction neutralizes the diffusive behavior entirely and leads to a stationary state. Furthermore, we derive various first-passage properties and describe four qualitatively different regimes of (strong) recurrent/transient behavior depending on the scale factor exponent.