Search

Recently, AMD platforms have not supported offloading C++17 PSTL (StdPar) programs to the GPU. Our previous work highlights how StdPar is able to achieve good performance across NVIDIA and Intel GPU platforms. In that work, we acknowledged AMD's past effort such as HCC, which unfortunately is deprecated and does not support newer hardware platforms. Recent developments by AMD, Codeplay, and AdaptiveCpp (previously known as hipSYCL or OpenSYCL) have enabled multiple paths for StdPar programs to run on AMD GPUs. This informal report discusses our experiences and evaluation of currently available StdPar implementations for AMD GPUs. We conduct benchmarks using our suite of HPC mini-apps with ports in many heterogeneous programming models, including StdPar. We then compare the performance of StdPar, using all available StdPar compilers, to contemporary heterogeneous programming models supported on AMD GPUs: HIP, OpenCL, Thrust, Kokkos, OpenMP, SYCL. Where appropriate, we discuss issues encountered and workarounds applied during our evaluation. Finally, the StdPar model discussed in this report largely depends on Unified Shared Memory (USM) performance and very few AMD GPUs have proper support for this feature. As such, this report demonstrates a proof-of-concept host-side userspace pagefault solution for models that use the HIP API. We discuss performance improvements achieved with our solution using the same set of benchmarks., Comment: This is a preliminary report intended as early feedback for interested vendors. Report contains 11 pages and 16 figures

We present the discovery of Ursa Major III/UNIONS 1, the least luminous known satellite of the Milky Way, which is estimated to have an absolute V-band magnitude of $+2.2^{+0.4}_{-0.3}$ mag, equivalent to a total stellar mass of 16$^{+6}_{-5}$ M$_{\odot}$. Ursa Major III/UNIONS 1 was uncovered in the deep, wide-field Ultraviolet Near Infrared Optical Northern Survey (UNIONS) and is consistent with an old ($\tau > 11$ Gyr), metal-poor ([Fe/H] $\sim -2.2$) stellar population at a heliocentric distance of $\sim$ 10 kpc. Despite being compact ($r_{\text{h}} = 3\pm1$ pc) and composed of so few stars, we confirm the reality of Ursa Major III/UNIONS 1 with Keck II/DEIMOS follow-up spectroscopy and identify 11 radial velocity members, 8 of which have full astrometric data from $Gaia$ and are co-moving based on their proper motions. Based on these 11 radial velocity members, we derive an intrinsic velocity dispersion of $3.7^{+1.4}_{-1.0}$ km s$^{-1}$ but some caveats preclude this value from being interpreted as a direct indicator of the underlying gravitational potential at this time. Primarily, the exclusion of the largest velocity outlier from the member list drops the velocity dispersion to $1.9^{+1.4}_{-1.1}$ km s$^{-1}$, and the subsequent removal of an additional outlier star produces an unresolved velocity dispersion. While the presence of binary stars may be inflating the measurement, the possibility of a significant velocity dispersion makes Ursa Major III/UNIONS 1 a high priority candidate for multi-epoch spectroscopic follow-ups to deduce to true nature of this incredibly faint satellite., Comment: 21 pages, 9 figures, 3 tables; Accepted for publication in ApJ

The recently discovered stellar system Ursa Major III/UNIONS 1 (UMa3/U1) is the faintest known Milky Way satellite to date. With a stellar mass of $16^{+6}_{-5}\,\rm M_\odot$ and a half-light radius of $3\pm1$pc, it is either the darkest galaxy ever discovered or the faintest self-gravitating star cluster known to orbit the Galaxy. Its line-of-sight velocity dispersion suggests the presence of dark matter, although current measurements are inconclusive because of the unknown contribution to the dispersion of potential binary stars. We use $N$-body simulations to show that, if self-gravitating, the system could not survive in the Milky Way tidal field for much longer than a single orbit (roughly 0.4Gyr), which strongly suggests that the system is stabilized by the presence of large amounts of dark matter. If UMa3/U1 formed at the center of a ~$10^9\rm M_\odot$ cuspy LCDM halo, its velocity dispersion would be predicted to be of order ~1km/s. This is roughly consistent with the current estimate, which, neglecting binaries, places $\sigma_{\rm los}$ in the range 1 to 4km/s. Because of its dense cusp, such a halo should be able to survive the Milky Way tidal field, keeping UMa3/U1 relatively unscathed until the present time. This implies that UMa3/U1 is plausibly the faintest and densest dwarf galaxy satellite of the Milky Way, with important implications for alternative dark matter models and for the minimum halo mass threshold for luminous galaxy formation in the LCDM cosmology. Our results call for multi-epoch high-resolution spectroscopic follow-up to confirm the dark matter content of this extraordinary system., Comment: updated to match accepted version

We report on the discovery of a significant and compact over-density of old and metal-poor stars in the KiDS survey (data release 4). The discovery is confirmed by deeper HSC-SSC data revealing the old Main Sequence Turn-Off of a stellar system located at a distance from the sun of $D_{\sun}=145^{+14}_{-13}$~kpc in the direction of the Sextans constellation. The system has absolute integrated magnitude ($M_V=-3.9^{+0.4}_{-0.3}$), half-light radius ($r_h=193^{+61}_{-46}$~pc), and ellipticity ($e=0.46^{+0.11}_{-0.15}$) typical of Ultra Faint Dwarf galaxies (UFDs). The central surface brightness is near the lower limits of known local dwarf galaxies of similar integrated luminosity, as expected for stellar systems that escaped detection until now. The distance of the newly found system suggests that it is likely a satellite of our own Milky Way, consequently, we tentatively baptise it Sextans~II (KiDS-UFD-1)., Comment: 7 pages, 5 figures. Accepted for publication on A&A Letters

The primary tool for analysing groups acting on trees is Bass--Serre Theory. It is comprised of two parts: a decomposition result, in which an action is decomposed via a graph of groups, and a construction result, in which graphs of groups are used to build examples of groups acting on trees. The usefulness of the latter for constructing new examples of `large' (e.g. nondiscrete) groups acting on trees is severely limited. There is a pressing need for new examples of such groups as they play an important role in the theory of locally compact groups. An alternative `local-to-global' approach to the study of groups acting on trees has recently emerged, inspired by a paper of Marc Burger and Shahar Mozes, based on groups that are `universal' with respect to some specified `local' action. In recent work, the authors of this survey article have developed a general theory of universal groups of local actions, that behaves, in many respects, like Bass--Serre Theory. We call this the theory of local action diagrams. The theory is powerful enough to completely describe all closed groups of automorphisms of trees that enjoy Tits' Independence Property (P). This article is an introductory survey of the local-to-global behaviour of groups acting on trees and the theory of local action diagrams. The article contains many ideas for future research projects., Comment: Survey article based on Simon M Smith's lecture at Groups St Andrews 2022

Dwarf galaxies are valuable laboratories for dynamical studies related to dark matter and galaxy evolution, yet it is currently unknown just how physically extended their stellar components are. Satellites orbiting the Galaxy's potential may undergo tidal stripping by the host, or alternatively, may themselves have accreted smaller systems whose debris populates the dwarf's own stellar halo. Evidence of these past interactions, if present, is best searched for in the outskirts of the satellite. However, foreground contamination dominates the signal at these large radial distances, making observation of stars in these regions difficult. In this work, we introduce an updated algorithm for application to Gaia data that identifies candidate member stars of dwarf galaxies, based on spatial, color-magnitude and proper motion information, and which allows for an outer component to the stellar distribution. Our method shows excellent consistency with spectroscopically confirmed members from the literature despite having no requirement for radial velocity information. We apply the algorithm to all $\sim$60 Milky Way dwarf galaxy satellites, and we find 9 dwarfs (Bo\"otes 1, Bo\"otes 3, Draco 2, Grus 2, Segue 1, Sculptor, Tucana 2, Tucana 3, and Ursa Minor) that exhibit evidence for a secondary, low-density outer profile. We identify many member stars which are located beyond 5 half-light radii (and in some cases, beyond 10). We argue these distant stars are likely tracers of dwarf stellar haloes or tidal streams, though ongoing spectroscopic follow-up will be required to determine the origin of these extended stellar populations., Comment: 25 pages, 13 figures, 4 tables, published in MNRAS

The formation of "stellar halos" in dwarf galaxies have been discussed in terms of early mergers or Galactic tides, although fluctuations in the gravitational potential due to stellar feedback is also a possible candidate mechanism. A Bayesian algorithm is used to find new candidate members in the extreme outskirts of the Sculptor dwarf galaxy. Precise metallicities and radial velocities for two distant stars are measured from their spectra taken with the Gemini South GMOS spectrograph. The radial velocity, proper motion and metallicity of these targets are consistent with Sculptor membership. As a result, the known boundary of the Sculptor dwarf extends now out to an elliptical distance of $\sim10$ half-light radii, which corresponds to a projected physical distance of $\sim3$ kpc. As reported in earlier work, the overall distribution of radial velocities and metallicities indicate the presence of a more spatially and kinematically dispersed metal-poor population that surrounds the more concentrated and colder metal-rich stars. Sculptor's density profile shows a "kink" in its logarithmic slope at a projected distance of $\sim25$ arcmin (620 pc), which we interpret as evidence that Galactic tides have helped to populate the distant outskirts of the dwarf. We discuss further ways to test and validate this tidal interpretation for the origin of these distant stars., Comment: Accepted for publication on MNRAS. Figure 1 now displays all the candidate members. arXiv admin note: text overlap with arXiv:2301.13214

Learning algorithms, like Quality-Diversity (QD), can be used to acquire repertoires of diverse robotics skills. This learning is commonly done via computer simulation due to the large number of evaluations required. However, training in a virtual environment generates a gap between simulation and reality. Here, we build upon the Reset-Free QD (RF-QD) algorithm to learn controllers directly on a physical robot. This method uses a dynamics model, learned from interactions between the robot and the environment, to predict the robot's behaviour and improve sample efficiency. A behaviour selection policy filters out uninteresting or unsafe policies predicted by the model. RF-QD also includes a recovery policy that returns the robot to a safe zone when it has walked outside of it, allowing continuous learning. We demonstrate that our method enables a physical quadruped robot to learn a repertoire of behaviours in two hours without human supervision. We successfully test the solution repertoire using a maze navigation task. Finally, we compare our approach to the MAP-Elites algorithm. We show that dynamics awareness and a recovery policy are required for training on a physical robot for optimal archive generation. Video available at https://youtu.be/BgGNvIsRh7Q, Comment: 5 pages, 2 figures, 1 linked video, to be presented as a poster at the Genetic and Evolutionary Computation Conference Companion (GECCO 2023 Companion), July, 2023, Lisbon, Portugal

Filamentary structures have been found nearly ubiquitously in molecular clouds and yet their formation and evolution is still poorly understood. We examine a segment of Taurus Molecular Cloud 1 (TMC-1) that appears as a single, narrow filament in continuum emission from dust. We use the Regularized Optimization for Hyper-Spectral Analysis (ROHSA), a Gaussian decomposition algorithm which enforces spatial coherence when fitting multiple velocity components simultaneously over a data cube. We analyze HC$_5$N (9-8) line emission as part of the Green Bank Ammonia Survey (GAS) and identify three velocity-coherent components with ROHSA. The two brightest components extend the length of the filament, while the third component is fainter and clumpier. The brightest component has a prominent transverse velocity gradient of $2.7 \pm 0.1$ km s$^{-1}$ pc$^{-1}$ that we show to be indicative of gravitationally induced inflow. In the second component, we identify regularly spaced emission peaks along its length. We show that the local minima between pairs of adjacent HC$_5$N peaks line up closely with submillimetre continuum emission peaks, which we argue is evidence for fragmentation along the spine of TMC-1. While coherent velocity components have been described as separate physical structures in other star-forming filaments, we argue that the two bright components identified in HC$_5$N emission in TMC-1 are tracing two layers in one filament: a lower density outer layer whose material is flowing under gravity towards the higher density inner layer of the filament., Comment: 17 pages, 8 figures; Accepted for publication to MNRAS

We present a Quality-Diversity benchmark suite for Deep Neuroevolution in Reinforcement Learning domains for robot control. The suite includes the definition of tasks, environments, behavioral descriptors, and fitness. We specify different benchmarks based on the complexity of both the task and the agent controlled by a deep neural network. The benchmark uses standard Quality-Diversity metrics, including coverage, QD-score, maximum fitness, and an archive profile metric to quantify the relation between coverage and fitness. We also present how to quantify the robustness of the solutions with respect to environmental stochasticity by introducing corrected versions of the same metrics. We believe that our benchmark is a valuable tool for the community to compare and improve their findings. The source code is available online: https://github.com/adaptive-intelligent-robotics/QDax, Comment: Accepted at GECCO Workshop on Quality Diversity Algorithm Benchmarks

In real-world environments, robots need to be resilient to damages and robust to unforeseen scenarios. Quality-Diversity (QD) algorithms have been successfully used to make robots adapt to damages in seconds by leveraging a diverse set of learned skills. A high diversity of skills increases the chances of a robot to succeed at overcoming new situations since there are more potential alternatives to solve a new task.However, finding and storing a large behavioural diversity of multiple skills often leads to an increase in computational complexity. Furthermore, robot planning in a large skill space is an additional challenge that arises with an increased number of skills. Hierarchical structures can help reducing this search and storage complexity by breaking down skills into primitive skills. In this paper, we introduce the Hierarchical Trial and Error algorithm, which uses a hierarchical behavioural repertoire to learn diverse skills and leverages them to make the robot adapt quickly in the physical world. We show that the hierarchical decomposition of skills enables the robot to learn more complex behaviours while keeping the learning of the repertoire tractable. Experiments with a hexapod robot show that our method solves a maze navigation tasks with 20% less actions in simulation, and 43% less actions in the physical world, for the most challenging scenarios than the best baselines while having 78% less complete failures., Comment: arXiv admin note: substantial text overlap with arXiv:2204.05726

We present the discovery of Bo\"otes V, a new ultra-faint dwarf galaxy candidate. This satellite is detected as a resolved overdensity of stars during an ongoing search for new Local Group dwarf galaxy candidates in the UNIONS photometric dataset. It has a physical half-light radius of 26.9$^{+7.5}_{-5.4}$ pc, a $V$-band magnitude of $-$4.5 $\pm$ 0.4 mag, and resides at a heliocentric distance of approximately 100 kpc. We use Gaia DR3 astrometry to identify member stars, characterize the systemic proper motion, and confirm the reality of this faint stellar system. The brightest star in this system was followed up using Gemini GMOS-N long-slit spectroscopy and is measured to have a metallicity of [Fe/H] $=$ $-$2.85 $\pm$ 0.10 dex and a heliocentric radial velocity of $v_r$ = 5.1 $\pm$ 13.4 km s$^{-1}$. Bo\"otes V is larger (in terms of scale radius), more distant, and more metal-poor than the vast majority of globular clusters. It is likely that Bo\"otes V is an ultra-faint dwarf galaxy, though future spectroscopic studies will be necessary to definitively classify this object., Comment: 13 pages, 7 figures, 3 tables. Accepted for publication in the AAS Journals. Please note that this paper was submitted in coordination with the work of William Cerny et al. 2022. These authors independently discovered this same satellite so our two research groups have coordinated the submission of these discovery papers

Adaptation capabilities, like damage recovery, are crucial for the deployment of robots in complex environments. Several works have demonstrated that using repertoires of pre-trained skills can enable robots to adapt to unforeseen mechanical damages in a few minutes. These adaptation capabilities are directly linked to the behavioural diversity in the repertoire. The more alternatives the robot has to execute a skill, the better are the chances that it can adapt to a new situation. However, solving complex tasks, like maze navigation, usually requires multiple different skills. Finding a large behavioural diversity for these multiple skills often leads to an intractable exponential growth of the number of required solutions. In this paper, we introduce the Hierarchical Trial and Error algorithm, which uses a hierarchical behavioural repertoire to learn diverse skills and leverages them to make the robot more adaptive to different situations. We show that the hierarchical decomposition of skills enables the robot to learn more complex behaviours while keeping the learning of the repertoire tractable. The experiments with a hexapod robot show that our method solves maze navigation tasks with 20% less actions in the most challenging scenarios than the best baseline while having 57% less complete failures., Comment: Accepted at GECCO 2022

Learning functional neuroanatomy requires forming mental representations of 3D structure, but forming such representations from 2D textbook diagrams can be challenging. We address this challenge in the UW Virtual Brain Project by developing 3D narrated diagrams, which are interactive, guided tours through 3D models of perceptual systems. Lessons can be experienced in virtual realty (VR) or on a personal computer monitor (PC). We predicted participants would learn from lessons presented on both VR and PC devices (comparing pre-test/post-test scores), but that VR would be more effective for achieving both content-based learning outcomes (i.e test performance) and experience-based learning outcomes (i.e., reported enjoyment and ease of use). All participants received lessons about the visual system and auditory system, one in VR and one on a PC(order counterbalanced). We assessed content learning using a drawing/labeling task on paper (2D drawing) in Experiment 1 and a Looking Glass autostereoscopic display (3D drawing) in Experiment 2. In both experiments, we found that the UW Virtual Brain Project lessons were effective for teaching functional neuroanatomy, with no difference between devices. However, participants reported VR was more enjoyable and easier to use. We also evaluated the VR lessons in our Classroom Implementation during an undergraduate course on perception. Students reported that the VR lessons helped them make progress on course learning outcomes, especially for learning system pathways. They suggested lessons could be improved byadding more examples and providing more time to explore in VR., Comment: To appear in Translational Issues in Psychological Science

Understanding a controller's performance in different scenarios is crucial for robots that are going to be deployed in safety-critical tasks. If we do not have a model of the dynamics of the world, which is often the case in complex domains, we may need to approximate a performance function of the robot based on its interaction with the environment. Such a performance function gives us insights into the behaviour of the robot, allowing us to fine-tune the controller with manual interventions. In high-dimensionality systems, where the actionstate space is large, fine-tuning a controller is non-trivial. To overcome this problem, we propose a performance function whose domain is defined by external features and parameters of the controller. Attainment regions are defined over such a domain defined by feature-parameter pairs, and serve the purpose of enabling prediction of successful execution of the task. The use of the feature-parameter space -in contrast to the action-state space- allows us to adapt, explain and finetune the controller over a simpler (i.e., lower dimensional space). When the robot successfully executes the task, we use the attainment regions to gain insights into the limits of the controller, and its robustness. When the robot fails to execute the task, we use the regions to debug the controller and find adaptive and counterfactual changes to the solutions. Another advantage of this approach is that we can generalise through the use of Gaussian processes regression of the performance function in the high-dimensional space. To test our approach, we demonstrate learning an approximation to the performance function in simulation, with a mobile robot traversing different terrain conditions. Then, with a sample-efficient method, we propagate the attainment regions to a physical robot in a similar environment., Comment: 6 pages, 3 figures. To be published in the International Conference on Intelligent Robots and Systems, IROS, 2021

We propose a portable framework to infer shareholders' preferences and influences on firms' prosocial decisions and the costs these decisions impose on firms and shareholders. Using quasi-experimental variations from the media coverage of firms' annual general meetings, we find that shareholders support costly prosocial decisions, such as covid-related donations and private sanctions on Russia, if they can earn image gains from them. In contrast, shareholders that the public cannot readily associate with specific firms, like financial corporations with large portfolios, oppose them. These prosocial expenditures crowd out investments at exposed firms, reducing productivity and earnings by between 1 and 3\%: pursuing the values of some shareholders comes at the cost of others, which the shareholders' monitoring motivated by heterogeneous preferences could prevent.

We propose an architecture for training generative models of counterfactual conditionals of the form, 'can we modify event A to cause B instead of C?', motivated by applications in robot control. Using an 'adversarial training' paradigm, an image-based deep neural network model is trained to produce small and realistic modifications to an original image in order to cause user-defined effects. These modifications can be used in the design process of image-based robust control - to determine the ability of the controller to return to a working regime by modifications in the input space, rather than by adaptation. In contrast to conventional control design approaches, where robustness is quantified in terms of the ability to reject noise, we explore the space of counterfactuals that might cause a certain requirement to be violated, thus proposing an alternative model that might be more expressive in certain robotics applications. So, we propose the generation of counterfactuals as an approach to explanation of black-box models and the envisioning of potential movement paths in autonomous robotic control. Firstly, we demonstrate this approach in a set of classification tasks, using the well known MNIST and CelebFaces Attributes datasets. Then, addressing multi-dimensional regression, we demonstrate our approach in a reaching task with a physical robot, and in a navigation task with a robot in a digital twin simulation., Comment: 8 pages, 11 figures. To be published in the 10th IEEE International Conference on Development and Learning (ICDL), Valparaiso, Chile

Semi-supervised learning improves the performance of supervised machine learning by leveraging methods from unsupervised learning to extract information not explicitly available in the labels. Through the design of a system that enables a robot to learn inspection strategies from a human operator, we present a hybrid semi-supervised system capable of learning interpretable and verifiable models from demonstrations. The system induces a controller program by learning from immersive demonstrations using sequential importance sampling. These visual servo controllers are parametrised by proportional gains and are visually verifiable through observation of the position of the robot in the environment. Clustering and effective particle size filtering allows the system to discover goals in the state space. These goals are used to label the original demonstration for end-to-end learning of behavioural models. The behavioural models are used for autonomous model predictive control and scrutinised for explanations. We implement causal sensitivity analysis to identify salient objects and generate counterfactual conditional explanations. These features enable decision making interpretation and post hoc discovery of the causes of a failure. The proposed system expands on previous approaches to program synthesis by incorporating repellers in the attribution prior of the sampling process. We successfully learn the hybrid system from an inspection scenario where an unmanned ground vehicle has to inspect, in a specific order, different areas of the environment. The system induces an interpretable computer program of the demonstration that can be synthesised to produce novel inspection behaviours. Importantly, the robot successfully runs the synthesised program on an unseen configuration of the environment while presenting explanations of its autonomous behaviour., Comment: In Proceedings AREA 2020, arXiv:2007.11260

A 1970 article of J. Tits concerning groups acting on trees introduced an independence property $(\mathrm{P})$ as a condition to produce the first examples of nonlinear nondiscrete locally compact simple groups, answering a question of J. P. Serre. This property has become very important in the recent development of the theory of totally disconnected, locally compact (t.d.l.c.) groups, with the majority of new constructions of compactly generated simple t.d.l.c. groups using $(\mathrm{P})$ or related ideas. In this paper we aim to advance the local-to-global theory of groups acting on trees by developing a `local action' complement to classical Bass--Serre theory. We describe, for a closed group $G$ of automorphisms of a (not necessarily locally finite) tree $T$ something called a local action diagram: a graph decorated with the local actions of $G$. A local action diagram plays a role in our theory that is analogous to a graph of groups in Bass--Serre theory. In place of the universal cover of a graph of groups, we define the universal group of a local action diagram. In this context, the groups $\mathbf{U}(F)$ and $\mathbf{U}(F_1, F_2)$ play analogous roles to the HNN extension and amalgamated free product respectively in Bass--Serre theory. We then show how to determine whether the universal group has certain properties, such as geometric density, compact generation and simplicity, directly from the local action diagram. Our theory allows us to completely describe all closed groups of automorphisms of trees with Tits' independence property $(\mathrm{P})$: they are precisely the universal groups of local action diagrams., Comment: 68 pages, with an appendix by Stephan Tornier

Contraction Clustering (RASTER) is a single-pass algorithm for density-based clustering of 2D data. It can process arbitrary amounts of data in linear time and in constant memory, quickly identifying approximate clusters. It also exhibits good scalability in the presence of multiple CPU cores. RASTER exhibits very competitive performance compared to standard clustering algorithms, but at the cost of decreased precision. Yet, RASTER is limited to batch processing and unable to identify clusters that only exist temporarily. In contrast, S-RASTER is an adaptation of RASTER to the stream processing paradigm that is able to identify clusters in evolving data streams. This algorithm retains the main benefits of its parent algorithm, i.e. single-pass linear time cost and constant memory requirements for each discrete time step within a sliding window. The sliding window is efficiently pruned, and clustering is still performed in linear time. Like RASTER, S-RASTER trades off an often negligible amount of precision for speed. Our evaluation shows that competing algorithms are at least 50% slower. Furthermore, S-RASTER shows good qualitative results, based on standard metrics. It is very well suited to real-world scenarios where clustering does not happen continually but only periodically., Comment: 24 pages, 5 figures, 2 tables

This paper is about the structure of infinite primitive permutation groups and totally disconnected locally compact groups ("tdlc groups'"). The permutation groups we investigate are subdegree-finite (i.e. all orbits of point stabilisers are finite). Automorphism groups of connected, locally finite graphs are examples of subdegree-finite permutation groups. The tdlc groups we investigate all have a maximal subgroup that is compact and open. Tdlc groups with few open subgroups (recently studied by Pierre-Emmanuel Caprace and Timoth\'ee Marquis) are examples of such groups. We prove a classification result, and use it to show that every closed, subdegree-finite primitive group is a primitive subgroup of a product H Wr F_1 [X] F_2 Wr ... [X] F_{m-1} Wr F_m, where H is a closed, subdegree-finite and primitive group that is either finite or one-ended and almost topologically simple. The groups F_i are transitive and finite and m is finite. The product [X] here denotes the recently discovered box product, and all wreath products here act via their product action. We apply this permutational result to tdlc groups. If G is a tdlc group then it contains a compact open subgroup V. The permutation group induced by the action of G on the coset space G/V is called the Schlichting completion of the pair (G,V), and is denoted G//V. Knowledge of this action of G on G/V underpins many recent influential results in tdlc theory. We show that if V is a maximal subgroup of G, and G is non-compact, then G//V is subject to a topological interpretation of our result for primitive groups. We use this to show that in this case, if G//V is nondiscrete, preserves no nontrivial homogenous cartesian decomposition on G/V, and does not split nontrivially over a compact open subgroup, then the monolith of G//V is a nondiscrete, one-ended, topologically simple, compactly generated tdlc group., Comment: Typo in abstract corrected. Draft, comments welcome

Mandarin Chinese is an increasingly popular world language and object of study, and while there are numerous online character learning apps and flashcard systems, very little research has been done on inductive or autonomous learning in the realm of collocation acquisition. I propose a new Chinese implementation of a trusted corpus-based platform, currently available for English and several other languages, accompanied and enhanced by an adaptive approach to Chinese segmentation approach, whereby different ways of carving up a given sentence are selectively displayed to the learner. [For the complete volume, "Flipping the Blend through MOOCs, MALL and OIL--New Directions in CALL," see ED585277.]

Clustering is an essential data mining tool for analyzing and grouping similar objects. In big data applications, however, many clustering algorithms are infeasible due to their high memory requirements and/or unfavorable runtime complexity. In contrast, Contraction Clustering (RASTER) is a single-pass algorithm for identifying density-based clusters with linear time complexity. Due to its favorable runtime and the fact that its memory requirements are constant, this algorithm is highly suitable for big data applications where the amount of data to be processed is huge. It consists of two steps: (1) a contraction step which projects objects onto tiles and (2) an agglomeration step which groups tiles into clusters. This algorithm is extremely fast in both sequential and parallel execution. Our quantitative evaluation shows that a sequential implementation of RASTER performs significantly better than various standard clustering algorithms. Furthermore, the parallel speedup is significant: on a contemporary workstation, an implementation in Rust processes a batch of 500 million points with 1 million clusters in less than 50 seconds on one core. With 8 cores, the algorithm is about four times faster., Comment: 19 pages; journal paper extending a previous conference publication (cf. https://doi.org/10.1007/978-3-319-72926-8_6)

OODIDA (On-board/Off-board Distributed Data Analytics) is a platform for distributed real-time analytics, targeting fleets of reference vehicles in the automotive industry. Its users are data analysts. The bulk of the data analytics tasks are performed by clients (on-board), while a central cloud server performs supplementary tasks (off-board). OODIDA can be automatically packaged and deployed, which necessitates restarting parts of the system, or all of it. As this is potentially disruptive, we added the ability to execute user-defined Python modules on clients as well as the server. These modules can be replaced without restarting any part of the system; they can even be replaced between iterations of an ongoing assignment. This feature is referred to as active-code replacement. It facilitates use cases such as iterative A/B testing of machine learning algorithms or modifying experimental algorithms on-the-fly. Consistency of results is achieved by majority vote, which prevents tainted state. Active-code replacement can be done in less than a second in an idealized setting whereas a standard deployment takes many orders of magnitude more time. The main contribution of this paper is the description of a relatively straightforward approach to active-code replacement that is very user-friendly. It enables a data analyst to quickly execute custom code on the cloud server as well as on client devices. Sensible safeguards and design decisions ensure that this feature can be used by non-specialists who are not familiar with the implementation of OODIDA in general or this feature in particular. As a consequence of adding the active-code replacement feature, OODIDA is now very well-suited for rapid prototyping., Comment: 24 pages, 4 figures, 3 code listings, 1 table

A fleet of connected vehicles easily produces many gigabytes of data per hour, making centralized (off-board) data processing impractical. In addition, there is the issue of distributing tasks to on-board units in vehicles and processing them efficiently. Our solution to this problem is OODIDA (On-board/Off-board Distributed Data Analytics), which is a platform that tackles both task distribution to connected vehicles as well as concurrent execution of tasks on arbitrary subsets of edge clients. Its message-passing infrastructure has been implemented in Erlang/OTP, while the end points use a language-independent JSON interface. Computations can be carried out in arbitrary programming languages. The message-passing infrastructure of OODIDA is highly scalable, facilitating the execution of large numbers of concurrent tasks., Comment: 28 pages, 9 figures, 2 algorithms, 2 code listings, 1 table