Search

Finding correspondences between 3D shapes is an important and long-standing problem in computer vision, graphics and beyond. While approaches based on machine learning dominate modern 3D shape matching, almost all existing (learning-based) methods require that at least one of the involved shapes is complete. In contrast, the most challenging and arguably most practically relevant setting of matching partially observed shapes, is currently underexplored. One important factor is that existing datasets contain only a small number of shapes (typically below 100), which are unable to serve data-hungry machine learning approaches, particularly in the unsupervised regime. In addition, the type of partiality present in existing datasets is often artificial and far from realistic. To address these limitations and to encourage research on these relevant settings, we provide a generic and flexible framework for the procedural generation of challenging partial shape matching scenarios. Our framework allows for a virtually infinite generation of partial shape matching instances from a finite set of shapes with complete geometry. Further, we manually create cross-dataset correspondences between seven existing (complete geometry) shape matching datasets, leading to a total of 2543 shapes. Based on this, we propose several challenging partial benchmark settings, for which we evaluate respective state-of-the-art methods as baselines.

We derive formulae for Gram matrices arising in the Nyman--Beurling reformulation of the Riemann hypothesis. The development naturally leads upon series of the form $S(x) = \sum_{n\ge 1} R(nx)$ and their reciprocity relations. We give integral representations of these series; and we present decompositions of the quadratic forms associated with the Gram matrices along with a discussion of the components' properties., Comment: two minor amendments

Finding correspondences between 3D shapes is an important and long-standing problem in computer vision, graphics and beyond. A prominent challenge are partial-to-partial shape matching settings, which occur when the shapes to match are only observed incompletely (e.g. from 3D scanning). Although partial-to-partial matching is a highly relevant setting in practice, it is rarely explored. Our work bridges the gap between existing (rather artificial) 3D full shape matching and partial-to-partial real-world settings by exploiting geometric consistency as a strong constraint. We demonstrate that it is indeed possible to solve this challenging problem in a variety of settings. For the first time, we achieve geometric consistency for partial-to-partial matching, which is realized by a novel integer non-linear program formalism building on triangle product spaces, along with a new pruning algorithm based on linear integer programming. Further, we generate a new inter-class dataset for partial-to-partial shape-matching. We show that our method outperforms current SOTA methods on both an established intra-class dataset and our novel inter-class dataset.

Routing problems are a common optimization problem in industrial applications, which occur on a large scale in supply chain planning. Due to classical limitations for solving NP-hard problems, quantum computing hopes to improve upon speed or solution quality. Several suggestions have been made for encodings of routing problems to solve them with variational quantum algorithms. However, for an end user it is hard to decide a priori which encoding will give the best solutions according to their needs. In this work, we investigate different encodings for the Travelling Salesperson Problem. We compare their scaling and performance when using the Quantum Approximate Optimization Algorithm and the Variational Quantum Eigensolver and provide a clear guide for users when to choose which encoding. For small instances, we find evidence that the permutation encoding can yield good results since it does not suffer from feasibility issues., Comment: Accepted at IEEE QSW 2024

The challenge to achieve practical quantum computing considering current hardware size and gate fidelity is the sensitivity to errors and noise. Recent work has shown that by learning the underlying noise model capturing qubit cross-talk, error mitigation can push the boundary of practical quantum computing. This has been accomplished using Sparse Pauli-Lindblad models only on devices with a linear topology connectivity (i.e. superconducting qubit devices). In this work we extend the theoretical requirement for learning such noise models on hardware with full connectivity (i.e. ion trap devices)., Comment: 6 pages, 3 figures

Finding correspondences between 3D shapes is a crucial problem in computer vision and graphics, which is for example relevant for tasks like shape interpolation, pose transfer, or texture transfer. An often neglected but essential property of matchings is geometric consistency, which means that neighboring triangles in one shape are consistently matched to neighboring triangles in the other shape. Moreover, while in practice one often has only access to partial observations of a 3D shape (e.g. due to occlusion, or scanning artifacts), there do not exist any methods that directly address geometrically consistent partial shape matching. In this work we fill this gap by proposing to integrate state-of-the-art deep shape features into a novel integer linear programming partial shape matching formulation. Our optimization yields a globally optimal solution on low resolution shapes, which we then refine using a coarse-to-fine scheme. We show that our method can find more reliable results on partial shapes in comparison to existing geometrically consistent algorithms (for which one first has to fill missing parts with a dummy geometry). Moreover, our matchings are substantially smoother than learning-based state-of-the-art shape matching methods.

As 3rd-person pronoun usage shifts to include novel forms, e.g., neopronouns, we need more research on identity-inclusive NLP. Exclusion is particularly harmful in one of the most popular NLP applications, machine translation (MT). Wrong pronoun translations can discriminate against marginalized groups, e.g., non-binary individuals (Dev et al., 2021). In this ``reality check'', we study how three commercial MT systems translate 3rd-person pronouns. Concretely, we compare the translations of gendered vs. gender-neutral pronouns from English to five other languages (Danish, Farsi, French, German, Italian), and vice versa, from Danish to English. Our error analysis shows that the presence of a gender-neutral pronoun often leads to grammatical and semantic translation errors. Similarly, gender neutrality is often not preserved. By surveying the opinions of affected native speakers from diverse languages, we provide recommendations to address the issue in future MT research., Comment: Accepted to ACL

Flows in networks (or graphs) play a significant role in numerous computer vision tasks. The scalar-valued edges in these graphs often lead to a loss of information and thereby to limitations in terms of expressiveness. For example, oftentimes high-dimensional data (e.g. feature descriptors) are mapped to a single scalar value (e.g. the similarity between two feature descriptors). To overcome this limitation, we propose a novel formalism for non-separable multi-dimensional network flows. By doing so, we enable an automatic and adaptive feature selection strategy - since the flow is defined on a per-dimension basis, the maximizing flow automatically chooses the best matching feature dimensions. As a proof of concept, we apply our formalism to the multi-object tracking problem and demonstrate that our approach outperforms scalar formulations on the MOT16 benchmark in terms of robustness to noise.

Many relevant problems in industrial settings result in NP-hard optimization problems, such as the Capacitated Vehicle Routing Problem (CVRP) or its reduced variant, the Travelling Salesperson Problem (TSP). Even with today's most powerful classical algorithms, the CVRP is challenging to solve classically. Quantum computing may offer a way to improve the time to solution, although the question remains open as to whether Noisy Intermediate-Scale Quantum (NISQ) devices can achieve a practical advantage compared to classical heuristics. The most prominent algorithms proposed to solve combinatorial optimization problems in the NISQ era are the Quantum Approximate Optimization Algorithm (QAOA) and the more general Variational Quantum Eigensolver (VQE). However, implementing them in a way that reliably provides high-quality solutions is challenging, even for toy examples. In this work, we discuss decomposition and formulation aspects of the CVRP and propose an application-driven way to measure solution quality. Considering current hardware constraints, we reduce the CVRP to a clustering phase and a set of TSPs. For the TSP, we extensively test both QAOA and VQE and investigate the influence of various hyperparameters, such as the classical optimizer choice and strength of constraint penalization. Results of QAOA are generally of limited quality because the algorithm does not reach the energy threshold for feasible TSP solutions, even when considering various extensions such as recursive, warm-start and constraint-preserving mixer QAOA. On the other hand, the VQE reaches the energy threshold and shows a better performance. Our work outlines the obstacles to quantum-assisted solutions for real-world optimization problems and proposes perspectives on how to overcome them., Comment: Submitted to the IEEE for possible publication

Optimization problems are ubiquitous in various industrial settings, and multi-knapsack optimization is one recurrent task faced daily by several industries. The advent of quantum computing has opened a new paradigm for computationally intensive tasks, with promises of delivering better and faster solutions for specific classes of problems. This work presents a comprehensive study of quantum computing approaches for multi-knapsack problems, by investigating some of the most prominent and state-of-the-art quantum algorithms using different quantum software and hardware tools. The performance of the quantum approaches is compared for varying hyperparameters. We consider several gate-based quantum algorithms, such as QAOA and VQE, as well as quantum annealing, and present an exhaustive study of the solutions and the estimation of runtimes. Additionally, we analyze the impact of warm-starting QAOA to understand the reasons for the better performance of this approach. We discuss the implications of our results in view of utilizing quantum optimization for industrial applications in the future. In addition to the high demand for better quantum hardware, our results also emphasize the necessity of more and better quantum optimization algorithms, especially for multi-knapsack problems., Comment: 20 pages

Short-range interactions and long-range contacts drive the 3D folding of structured proteins. The proteins' structure has a direct impact on their biological function. However, nearly 40% of the eukaryotes proteome is composed of intrinsically disordered proteins (IDPs) and protein regions that fluctuate between ensembles of numerous conformations. Therefore, to understand their biological function, it is critical to depict how the structural ensemble statistics correlate to the IDPs' amino acid sequence. Here, using small-angle x-ray scattering (SAXS) and time-resolved F\"orster resonance energy transfer (trFRET), we study the intra-molecular structural heterogeneity of the neurofilament low intrinsically disordered tail domain (NFLt). Using theoretical results of polymer physics, we find that the Flory scaling exponent of NFLt sub-segments correlates linearly with their net charge, ranging from statistics of ideal to self-avoiding chains. Surprisingly, measuring the same segments in the context of the whole NFLt protein, we find that regardless of the peptide sequence, the segments' structural statistics are more expanded than when measured independently. Our findings show that while polymer physics can, to some level, relate the IDP's sequence to its ensemble conformations, long-range contacts between distant amino acids play a crucial role in determining intra-molecular structures. This emphasizes the necessity of advanced polymer theories to fully describe IDPs ensembles with the hope it will allow us to model their biological function.

Understanding the interplay between various design strategies (for instance, bonding heterogeneity and lone pair induced anharmonicity) to achieve ultralow lattice thermal conductivity ($\kappa_l$) is indispensable for discovering novel functional materials for thermal energy applications. In the present study, we investigate layered PbXF (X = Cl, Br, I), which offers bonding heterogeneity through the layered crystal structure, anharmonicity through the Pb$^{2+}$ $6s^2$ lone pair, and phonon softening through the mass difference between F and Pb/X. The weak inter-layer van der Waals bonding and the strong intra-layer ionic bonding with partial covalent bonding result in a significant bonding heterogeneity and a poor phonon transport in the out-of-plane direction. Large average Gr\"uneisen parameters ($\geq$ 2.5) demonstrate strong anharmonicity. The computed phonon dispersions show flat bands, which suggest short phonon lifetimes, especially for PbIF. Enhanced Born effective charges are due to cross-band-gap hybridization. PbIF shows lattice instability at a small volume expansion of 0.1$\%$. The $\kappa_l$ values obtained by the two channel transport model are 20-50$\%$ higher than those obtained by solving the Boltzmann transport equation. Overall, ultralow $\kappa_l$ values are found at 300 K, especially for PbIF. We propose that the interplay of bonding heterogeneity, lone pair induced anharmonicity, and constituent elements with high mass difference aids the design of low $\kappa_l$ materials for thermal energy applications.

Semiconductor supply chains are described by significant demand fluctuation that increases as one moves up the supply chain, the so-called bullwhip effect. To counteract, semiconductor manufacturers aim to optimize capacity utilization, to deliver with shorter lead times and exploit this to generate revenue. Additionally, in a competitive market, firms seek to maintain customer relationships while applying revenue management strategies such as dynamic pricing. Price change potentially generates conflicts with customers. In this paper, we present KnowGraph-PM, a knowledge graph-based dynamic pricing model. The semantic model uses the potential of faster delivery and shorter lead times to define premium prices, thus entail increased profits based on the customer profile. The knowledge graph enables the integration of customer-related information, e.g., customer class and location to customer order data. The pricing algorithm is realized as a SPARQL query that relies on customer profile and order behavior to determine the corresponding price premium. We evaluate the approach by calculating the revenue generated after applying the pricing algorithm. Based on competency questions that translate to SPARQL queries, we validate the created knowledge graph. We demonstrate that semantic data integration enables customer-tailored revenue management.

Supply Chains (SCs) are subject to disruptive events that potentially hinder the operational performance. Disruption Management Process (DMP) relies on the analysis of integrated heterogeneous data sources such as production scheduling, order management and logistics to evaluate the impact of disruptions on the SC. Existing approaches are limited as they address DMP process steps and corresponding data sources in a rather isolated manner which hurdles the systematic handling of a disruption originating anywhere in the SC. Thus, we propose MARE a semantic disruption management and resilience evaluation framework for integration of data sources included in all DMP steps, i.e. Monitor/Model, Assess, Recover and Evaluate. MARE, leverages semantic technologies i.e. ontologies, knowledge graphs and SPARQL queries to model and reproduce SC behavior under disruptive scenarios. Also, MARE includes an evaluation framework to examine the restoration performance of a SC applying various recovery strategies. Semantic SC DMP, put forward by MARE, allows stakeholders to potentially identify the measures to enhance SC integration, increase the resilience of supply networks and ultimately facilitate digitalization.

Supply Chain (SC) modeling is essential to understand and influence SC behavior, especially for increasingly globalized and complex SCs. Existing models address various SC notions, e.g., processes, tiers and production, in an isolated manner limiting enriched analysis granted by integrated information systems. Moreover, the scarcity of real-world data prevents the benchmarking of the overall SC performance in different circumstances, especially wrt. resilience during disruption. We present SENS, an ontology-based Knowlegde-Graph (KG) equipped with SPARQL implementations of KPIs to incorporate an end-to-end perspective of the SC including standardized SCOR processes and metrics. Further, we propose SENS-GEN, a highly configurable data generator that leverages SENS to create synthetic semantic SC data under multiple scenario configurations for comprehensive analysis and benchmarking applications. The evaluation shows that the significantly improved simulation and analysis capabilities, enabled by SENS, facilitate grasping, controlling and ultimately enhancing SC behavior and increasing resilience in disruptive scenarios.

Alkaline-earth metal oxides, in particular MgO and CaO dominate Earths lower mantle, therefore, exploring high pressure behavior of this class of compounds is of significant geophysical research interest. Among all these compounds, BaO exhibits rich polymorphism in the pressure range of 0-1.5 Mbar. Static enthalpy calculations revealed that BaO undergoes a pressure induced structural phase transition from NaCl-type (B1) $\rightarrow$ NiAs-type (B8) $\rightarrow$ distorted CsCl-type (d-B2) $\rightarrow$ CsCl-type (B2) at 5.1, 19.5, 120 GPa respectively. B1 $\rightarrow$ B8 $\&$ B8 $\rightarrow$ d-B2 transitions are found to be first order in nature whereas d-B2$\rightarrow$ B2 is a second order or weak first order phase transition. Interestingly, d-B2 phase shows stability over a wide pressure range, $\sim$19.5-113 GPa. Mechanical and dynamical stabilities of ambient and high pressure phases are demonstrated through computed elastic constants and phonon dispersion curves, respectively. Under high pressure, significant phonon softening and soft phonon mode along M-direction are observed for B8, d-B2 and B2 phases, respectively. Pressure dependent Raman spectra suggest a phase transition from d-B2 to Raman inactive phase under pressure. Overall, the present study provides a comprehensive understanding of underlying mechanisms behind pressure-induced structural phase transitions in BaO.

Finding shortest paths in a graph is relevant for numerous problems in computer vision and graphics, including image segmentation, shape matching, or the computation of geodesic distances on discrete surfaces. Traditionally, the concept of a shortest path is considered for graphs with scalar edge weights, which makes it possible to compute the length of a path by adding up the individual edge weights. Yet, graphs with scalar edge weights are severely limited in their expressivity, since oftentimes edges are used to encode significantly more complex interrelations. In this work we compensate for this modelling limitation and introduce the novel graph-theoretic concept of a shortest path in a graph with matrix-valued edges. To this end, we define a meaningful way for quantifying the path length for matrix-valued edges, and we propose a simple yet effective algorithm to compute the respective shortest path. While our formalism is universal and thus applicable to a wide range of settings in vision, graphics and beyond, we focus on demonstrating its merits in the context of 3D multi-shape analysis., Comment: published at 3DV

Materials with an intrinsic (ultra)low lattice thermal conductivity (k$_L$) are critically important for the development of efficient energy conversion devices. In the present work, we have investigated microscopic origins of low k$_L$ behavior in BaO, BaS and MgTe by exploring lattice dynamics and phonon transport of 16 iso-structural MX (Mg, Ca, Sr, Ba and X = O, S, Se and Te) compounds in the rocksalt (NaCl)-type structure by comparing their lattice transport properties with the champion thermoeletric iso-structural material, PbTe. Anomalous trends are observed for k$_L$ in MX compounds except the MgX series in contrast to the expected trend from their atomic mass. The underlying mechanisms for such low k$_L$ behavior in relatively low atomic mass systems namely BaO, BaS and MgTe compounds are thoroughly analyzed. We propose the following dominant factors that might be responsible for low k$_L$ behavior in these materials: 1) softening of transverse acoustic (TA) phonon modes despite low atomic mass, 2) low lying optic (LLO) phonon modes fall deep into acoustic mode region which enhances overlap between longitudinal acoustic (LA) and LLO phonon modes which increases scattering phase space, 3) short phonon lifetimes and high scattering rates, 4) relatively high density (\r{ho}) and large Gr\"uneisen parameter. Moreover, tensile strain also causes a further reduction in k$_L$ for BaO, BaS and MgTe through phonon softening and near ferroelectric instability. Our comprehensive study on 16 binary MX compounds might provide a pathway for designing (ultra)low k$_L$ materials even with simple crystal systems through phonon engineering., Comment: 40 pages, 10 figures

Amphiphilic molecules and their self-assembled structures have long been the target of extensive research due to their potential applications in fields ranging from materials design to biomedical and cosmetic applications. Increasing demands for functional complexity have been met with challenges in biochemical engineering, driving researchers to innovate in the design of new amphiphiles. An emerging class of molecules, namely, peptide amphiphiles, combines key advantages and circumvents some of the disadvantages of conventional phospholipids and block-copolymers. Herein, we present new peptide amphiphiles comprised of an intrinsically disordered peptide conjugated to two variants of hydrophobic dendritic domains. These molecules termed intrinsically disordered peptide amphiphiles (IDPA), exhibit a sharp pH-induced micellar phase-transition from low-dispersity spheres to extremely elongated worm-like micelles. We present an experimental characterization of the transition and propose a theoretical model to describe the pH-response. We also present the potential of the shape transition to serve as a mechanism for the design of a cargo hold-and-release application. Such amphiphilic systems demonstrate the power of tailoring the interactions between disordered peptides for various stimuli-responsive biomedical applications., Comment: 27 pages, 4 figures, supplementary information file

The human proteome is enriched in proteins that do not fold into a stable 3D structure. These intrinsically disordered proteins (IDPs) spontaneously fluctuate between a large number of configurations in their native form. Remarkably, the disorder does not lead to dysfunction as with denatured folded proteins. In fact, unlike denatured proteins, recent evidences strongly suggest that multiple biological functions stem from such structural plasticity. Here, focusing on the nanoscopic length-scale, we review the latest advances in IDP research and discuss some of the future directions in this highly promising field., Comment: 15 pages, 5 figures