Showing 1–50 of 207 results for author: Becker, P

Spectral evolution of RX J0440.9+4431 during the 2022-2023 giant outburst observed with Insight-HXMT

Authors: P. P. Li, Peter A. Becker, L. Tao

Abstract: In 2022-2023, the X-ray pulsar RX J0440.9+4431 underwent a Type II giant outburst, reaching a peak luminosity L_x ~ 4*10^{37} erg/s. In this work, we utilize Insight-HXMT data to analyze the spectral evolution of RX J0440.9+4431 during the giant outburst. By analysing the variation of the X-ray spectrum during the outburst using standard phenomenological models, we find that as the luminosity appr… ▽ More In 2022-2023, the X-ray pulsar RX J0440.9+4431 underwent a Type II giant outburst, reaching a peak luminosity L_x ~ 4*10^{37} erg/s. In this work, we utilize Insight-HXMT data to analyze the spectral evolution of RX J0440.9+4431 during the giant outburst. By analysing the variation of the X-ray spectrum during the outburst using standard phenomenological models, we find that as the luminosity approaches the critical luminosity, the spectrum became flatter, with the photon enhancement predominantly concentrated around ~ 2 keV and 20-40 keV. The same behavior has also been noted in Type II outbursts from other sources. While the phenomenological models provide good fits to the spectrum, this approach is sometimes difficult to translate into direct insight into the details of the fundamental accretion physics. Hence we have also analyzed spectra obtained during high and low phases of the outburst using a new, recently-developed physics-based theoretical model, which allows us to study the variations of physical parameters such as temperature, density, and magnetic field strength during the outburst. Application of the theoretical model reveals that the observed spectrum is dominated by Comptonized bremsstrahlung emission emitted from the column walls in both the high and low states. We show that the spectral flattening observed at high luminosities results from a decrease in the electron temperature, combined with a compactification of the emission zone, which reduces the efficiency of bulk Comptonization. We also demonstrate that when the source is at maximum luminosity, the spectrum tends to harden around the peak of the pulse profile, and we discuss possible theoretical explanations for this behavior. We argue that the totality of the behavior in this source can be explained if the accretion column is in a quasi-critical state when at the maximum luminosity observed during the outburst. △ Less

Submitted 21 June, 2024; originally announced July 2024.

Comments: 11 pages, 4 figures, accepted for publication in A&A

KalMamba: Towards Efficient Probabilistic State Space Models for RL under Uncertainty

Authors: Philipp Becker, Niklas Freymuth, Gerhard Neumann

Abstract: Probabilistic State Space Models (SSMs) are essential for Reinforcement Learning (RL) from high-dimensional, partial information as they provide concise representations for control. Yet, they lack the computational efficiency of their recent deterministic counterparts such as S4 or Mamba. We propose KalMamba, an efficient architecture to learn representations for RL that combines the strengths of… ▽ More Probabilistic State Space Models (SSMs) are essential for Reinforcement Learning (RL) from high-dimensional, partial information as they provide concise representations for control. Yet, they lack the computational efficiency of their recent deterministic counterparts such as S4 or Mamba. We propose KalMamba, an efficient architecture to learn representations for RL that combines the strengths of probabilistic SSMs with the scalability of deterministic SSMs. KalMamba leverages Mamba to learn the dynamics parameters of a linear Gaussian SSM in a latent space. Inference in this latent space amounts to standard Kalman filtering and smoothing. We realize these operations using parallel associative scanning, similar to Mamba, to obtain a principled, highly efficient, and scalable probabilistic SSM. Our experiments show that KalMamba competes with state-of-the-art SSM approaches in RL while significantly improving computational efficiency, especially on longer interaction sequences. △ Less

Submitted 21 June, 2024; originally announced June 2024.

Iterative Sizing Field Prediction for Adaptive Mesh Generation From Expert Demonstrations

Authors: Niklas Freymuth, Philipp Dahlinger, Tobias Würth, Philipp Becker, Aleksandar Taranovic, Onno Grönheim, Luise Kärger, Gerhard Neumann

Abstract: Many engineering systems require accurate simulations of complex physical systems. Yet, analytical solutions are only available for simple problems, necessitating numerical approximations such as the Finite Element Method (FEM). The cost and accuracy of the FEM scale with the resolution of the underlying computational mesh. To balance computational speed and accuracy meshes with adaptive resolutio… ▽ More Many engineering systems require accurate simulations of complex physical systems. Yet, analytical solutions are only available for simple problems, necessitating numerical approximations such as the Finite Element Method (FEM). The cost and accuracy of the FEM scale with the resolution of the underlying computational mesh. To balance computational speed and accuracy meshes with adaptive resolution are used, allocating more resources to critical parts of the geometry. Currently, practitioners often resort to hand-crafted meshes, which require extensive expert knowledge and are thus costly to obtain. Our approach, Adaptive Meshing By Expert Reconstruction (AMBER), views mesh generation as an imitation learning problem. AMBER combines a graph neural network with an online data acquisition scheme to predict the projected sizing field of an expert mesh on a given intermediate mesh, creating a more accurate subsequent mesh. This iterative process ensures efficient and accurate imitation of expert mesh resolutions on arbitrary new geometries during inference. We experimentally validate AMBER on heuristic 2D meshes and 3D meshes provided by a human expert, closely matching the provided demonstrations and outperforming a single-step CNN baseline. △ Less

Submitted 20 June, 2024; originally announced June 2024.

Comments: Accepted as a workshop paper in AI4Science@ICML 2024

The giant outburst of EXO 2030+375 II: Broadband spectroscopy and evolution

Authors: R. Ballhausen, P. Thalhammer, P. Pradhan, E. Sokolova-Lapa, J. Stierhof, K. Pottschmidt, J. Wilms, J. B. Coley, P. Kretschmar, F. Fuerst, P. Becker, B. West, C. Malacaria, M. T. Wolff, R. Rothschild, R. Staubert

Abstract: In 2021, the high-mass X-ray binary EXO 2030+375 underwent a giant X-ray outburst, the first since 2006, that reached a peak flux of ${\sim}600\,\mathrm{mCrab}$ (3-50\,keV). The goal of this work is to study the spectral evolution over the course of the outburst, search for possible cyclotron resonance scattering features (CRSFs), and to associate spectral components with the emission pattern of t… ▽ More In 2021, the high-mass X-ray binary EXO 2030+375 underwent a giant X-ray outburst, the first since 2006, that reached a peak flux of ${\sim}600\,\mathrm{mCrab}$ (3-50\,keV). The goal of this work is to study the spectral evolution over the course of the outburst, search for possible cyclotron resonance scattering features (CRSFs), and to associate spectral components with the emission pattern of the accretion column. We used broadband spectra taken with the Nuclear Spectroscopic Telescope Array (NuSTAR), the Neutron Star Interior Composition Explorer (NICER), and Chandra near the peak and during the decline phase of the outburst. We describe the data with established empirical continuum models and perform pulse-phase-resolved spectroscopy. We compare the spectral evolution with pulse phase using a proposed geometrical emission model. We find a significant spectral hardening toward lower luminosity, a behavior that is expected for super-critical sources. The continuum shape and evolution cannot be described by a simple power-law model with exponential cutoff; it requires additional absorption or emission components. We can confirm the presence of a narrow absorption feature at ${\sim}10\,\mathrm{keV}$ in both NuSTAR observations. The absence of harmonics puts into question the interpretation of this feature as a CRSF. The empirical spectral components cannot be directly associated with identified emission components from the accretion column. △ Less

Submitted 18 June, 2024; originally announced June 2024.

Comments: 12 pages, 8 figures, accepted in A&A

The giant outburst of EXO 2030+375 I: Spectral and pulse profile evolution

Authors: P. Thalhammer, R. Ballhausen, E. Sokolova-Lapa, J. Stierhof, A. Zainab, R. Staubert, K. Pottschmidt, J. B. Coley, R. E. Rothschild, G. K. Jaisawal, B. West, P. A. Becker, P. Pradhan, P. Kretschmar, J. Wilms

Abstract: The Be X-ray binary EXO 2030+375 went through its third recorded giant outburst from June 2021 to early 2022. We present the results of both spectral and timing analysis based on NICER monitoring, covering the 2-10 keV flux range from 20 to 310 mCrab. Dense monitoring with observations carried out about every second day and a total exposure time of 160 ks allowed us to closely track the source evo… ▽ More The Be X-ray binary EXO 2030+375 went through its third recorded giant outburst from June 2021 to early 2022. We present the results of both spectral and timing analysis based on NICER monitoring, covering the 2-10 keV flux range from 20 to 310 mCrab. Dense monitoring with observations carried out about every second day and a total exposure time of 160 ks allowed us to closely track the source evolution over the outburst. Changes in spectral shape and pulse profiles showed a stable luminosity dependence during the rise and decline. The same type of dependence has been seen in past outbursts. The pulse profile is characterized by several distinct peaks and dips. The profiles show a clear dependence on luminosity with a stark transition at a luminosity of 2x10^36 erg/s, indicating a change in the emission pattern. Using relativistic ray-tracing, we demonstrate how anisotropic beaming of emission from an accretion channel with constant geometrical configuration can give rise to the observed pulse profiles over a range of luminosities. △ Less

Submitted 31 May, 2024; originally announced May 2024.

Comments: 15 pages, 16 figures, accepted by A&A

Theoretical Analysis of the RX J0209.6-7427 X-ray Spectrum During Giant Outburst

Authors: Brent F. West, Peter A. Becker, Georgios Vasilopoulos

Abstract: We model the spectral formation occurring in the binary X-ray pulsar RX~J0209.6-7427 during the 2019 super-Eddington outburst. Using a theoretical model previously developed by the authors, we are able to produce spectra that closely resemble the phase-averaged X-ray spectra observed using NuSTAR and Insight-HXMT during low and high luminosity states of the outburst, respectively. The theoretical… ▽ More We model the spectral formation occurring in the binary X-ray pulsar RX~J0209.6-7427 during the 2019 super-Eddington outburst. Using a theoretical model previously developed by the authors, we are able to produce spectra that closely resemble the phase-averaged X-ray spectra observed using NuSTAR and Insight-HXMT during low and high luminosity states of the outburst, respectively. The theoretical model simulates the accretion of fully ionized gas in a dipole magnetic field, and includes a complete description of the radiation hydrodynamics, matter distribution, and spectral formation. Type II X-ray outbursts provide an opportunity to study accretion over a large range of luminosities for the same neutron star. The analysis performed here represents the first time both the outburst low and high states of an accretion-powered X-ray pulsar are modeled using a physics-based model rather than standard phenomenological fitting with arbitrary mathematical functions. We find the outer polar cap radius remains constant and the column is more fully-filled with increasing luminosity, Comptonized bremsstrahlung dominates the formation of the phase-averaged X-ray spectrum, and a negative correlation exists between cyclotron centroid energy and luminosity, as expected. The super-Eddington nature of the outburst is rendered possible due to the low scattering cross section for photons propagating parallel to the magnetic field. We also find emission through the column top dominates in both the low and high states, implying the pulse profiles should have a roughly sinusoidal shape, which agrees with observed properties of ultra-luminous X-ray pulsars. △ Less

Submitted 24 April, 2024; originally announced April 2024.

Comments: 11 pages, 5 figures

On the number of minimal forts of a graph

Authors: Paul Becker, Thomas R. Cameron, Derek Hanely, Boon Ong, Joseph P. Previte

Abstract: In 2018, a fort of a graph was introduced as a non-empty subset of vertices in which no vertex outside of the set has exactly one neighbor in the set. Since then, forts have been used to characterize zero forcing sets, model the zero forcing number as an integer program, and generate lower bounds on the zero forcing number of a Cartesian product. In this article, we investigate the number of minim… ▽ More In 2018, a fort of a graph was introduced as a non-empty subset of vertices in which no vertex outside of the set has exactly one neighbor in the set. Since then, forts have been used to characterize zero forcing sets, model the zero forcing number as an integer program, and generate lower bounds on the zero forcing number of a Cartesian product. In this article, we investigate the number of minimal forts of a graph, where a fort is minimal if every proper subset is not a fort. In particular, we show that the number of minimal forts of a graph of order at least six is strictly less than Sperner's bound, a famous bound due to Emanuel Sperner (1928) on the size of a collection of subsets where no subset contains another. Then, we derive an explicit formula for the number of minimal forts for several families of graphs, including the path, cycle, and spider graphs. Moreover, we show that the asymptotic growth rate of the number of minimal forts of the spider graph is bounded above by that of the path graph. We conjecture that the asymptotic growth rate of the path graph is extremal over all trees. Finally, we develop methods for constructing minimal forts of graph products using the minimal forts of the original graphs. In the process, we derive explicit formulas and lower bounds on the number of minimal forts for additional families of graphs, such as the wheel, sunlet, and windmill graphs. Most notably, we show that the family of windmill graphs has an exponential number of minimal forts with a maximum asymptotic growth rate of cube root of three, which is the largest asymptotic growth rate we have observed. We conjecture that there exist families of graphs with a larger asymptotic growth rate. △ Less

Submitted 8 April, 2024; originally announced April 2024.

Comments: 21 pages, 6 figures, 1 table

MSC Class: 05C15 (Primary) 05C30; 05C35; 05C50; 05C57; 05C76; 90C35 (Secondary)

Witnessing Quantum Entanglement Using Resonant Inelastic X-ray Scattering

Authors: Tianhao Ren, Yao Shen, Sophia F. R. TenHuisen, Jennifer Sears, Wei He, Mary H. Upton, Diego Casa, Petra Becker, Matteo Mitrano, Mark P. M. Dean, Robert M. Konik

Abstract: Although entanglement is both a central ingredient in our understanding of quantum many-body systems and an essential resource for quantum technologies, we only have a limited ability to quantify entanglement in real quantum materials. Thus far, entanglement metrology in quantum materials has been limited to measurements involving Hermitian operators, such as the detection of spin entanglement usi… ▽ More Although entanglement is both a central ingredient in our understanding of quantum many-body systems and an essential resource for quantum technologies, we only have a limited ability to quantify entanglement in real quantum materials. Thus far, entanglement metrology in quantum materials has been limited to measurements involving Hermitian operators, such as the detection of spin entanglement using inelastic neutron scattering. Here, we devise a method to extract the quantum Fisher information (QFI) from non-Hermitian operators and formulate an entanglement witness for resonant inelastic x-ray scattering (RIXS). Our approach is then applied to the model iridate dimer system Ba$_3$CeIr$_2$O$_9$ and used to directly test for entanglement of the electronic orbitals between neighboring Ir sites. We find that entanglement is challenging to detect under standard conditions, but that it could be achieved by analyzing the outgoing x-ray polarization or via specific choices of momentum and energy. Our protocol provides a new handle for entanglement detection, which offers routes to related types of entanglement witness (such as orbitally-resolved measurements) and to the generalization to out-of-equilibrium settings accessed in ultrafast settings. △ Less

Submitted 8 April, 2024; originally announced April 2024.

Comments: 10 pages, 8 figures

Vlearn: Off-Policy Learning with Efficient State-Value Function Estimation

Authors: Fabian Otto, Philipp Becker, Ngo Anh Vien, Gerhard Neumann

Abstract: Existing off-policy reinforcement learning algorithms often rely on an explicit state-action-value function representation, which can be problematic in high-dimensional action spaces due to the curse of dimensionality. This reliance results in data inefficiency as maintaining a state-action-value function in such spaces is challenging. We present an efficient approach that utilizes only a state-va… ▽ More Existing off-policy reinforcement learning algorithms often rely on an explicit state-action-value function representation, which can be problematic in high-dimensional action spaces due to the curse of dimensionality. This reliance results in data inefficiency as maintaining a state-action-value function in such spaces is challenging. We present an efficient approach that utilizes only a state-value function as the critic for off-policy deep reinforcement learning. This approach, which we refer to as Vlearn, effectively circumvents the limitations of existing methods by eliminating the necessity for an explicit state-action-value function. To this end, we introduce a novel importance sampling loss for learning deep value functions from off-policy data. While this is common for linear methods, it has not been combined with deep value function networks. This transfer to deep methods is not straightforward and requires novel design choices such as robust policy updates, twin value function networks to avoid an optimization bias, and importance weight clip**. We also present a novel analysis of the variance of our estimate compared to commonly used importance sampling estimators such as V-trace. Our approach improves sample complexity as well as final performance and ensures consistent and robust performance across various benchmark tasks. Eliminating the state-action-value function in Vlearn facilitates a streamlined learning process, enabling more effective exploration and exploitation in complex environments. △ Less

Submitted 20 June, 2024; v1 submitted 7 March, 2024; originally announced March 2024.

Rotational phase transitions in antifluorite-type osmate and iridate compounds

Authors: A. Bertin, L. Kiefer, P. Becker, L. Bohatý, M. Braden

Abstract: We present temperature-dependent single-crystal diffraction results on seven antifluorite-type $A_2MeX_6$ compounds with $Me$=Os or Ir: K$_2$OsCl$_6$, $A_2$OsBr$_6$ with $A$=K, Rb, Cs and NH$_4$, and K$_2$Ir$X_6$ with $X$=Cl and Br. The structural transitions in this family arise from $MeX_6$ octahedron rotations that generate a rich variety of symmetries depending on the rotation axis and stackin… ▽ More We present temperature-dependent single-crystal diffraction results on seven antifluorite-type $A_2MeX_6$ compounds with $Me$=Os or Ir: K$_2$OsCl$_6$, $A_2$OsBr$_6$ with $A$=K, Rb, Cs and NH$_4$, and K$_2$Ir$X_6$ with $X$=Cl and Br. The structural transitions in this family arise from $MeX_6$ octahedron rotations that generate a rich variety of symmetries depending on the rotation axis and stacking schemes. In order to search for local distortions in the high-symmetry phase we perform refinements of anharmonic atomic displacement parameters with comprehensive data sets. Even at temperatures close to the onset of structural distortions, these refinements only yield a small improvement indicating only small anharmonic effects. The phase transitions in these antifluorites are essentially of displacive character. However, some harmonic displacement parameters are very large reflecting soft phonon modes with the softening covering large parts of the Brillouin zone. The occurrence of the rotational transitions in the antifluorite-type family can be remarkably well analyzed in terms of a tolerance factor of ionic radii. △ Less

Submitted 23 February, 2024; originally announced February 2024.

Comments: 18 pages, 8 figures

Journal ref: A Bertin et al 2024 J. Phys.: Condens. Matter 36 245402

Spin-orbit coupling in a half-filled $t_{2g}$ shell: the case of $5d^3$ K$_2$ReCl$_6$

Authors: P. Warzanowski, M. Magnaterra, G. Schlicht, Q. Faure, Ch. J. Sahle, P. Becker, L. Bohatý, M. Moretti Sala, G. Monaco, M. Hermanns, P. H. M. van Loosdrecht, M. Grüninger

Abstract: The half-filled $t_{2g}$ shell of the $t_{2g}^3$ configuration usually, in LS coupling, hosts a S = 3/2 ground state with quenched orbital moment. This state is not Jahn-Teller active. Sufficiently large spin-orbit coupling $ζ$ has been predicted to change this picture by mixing in orbital moment, giving rise to a sizable Jahn-Teller distortion. In $5d^3$ K$_2$ReCl$_6$ we study the electronic exci… ▽ More The half-filled $t_{2g}$ shell of the $t_{2g}^3$ configuration usually, in LS coupling, hosts a S = 3/2 ground state with quenched orbital moment. This state is not Jahn-Teller active. Sufficiently large spin-orbit coupling $ζ$ has been predicted to change this picture by mixing in orbital moment, giving rise to a sizable Jahn-Teller distortion. In $5d^3$ K$_2$ReCl$_6$ we study the electronic excitations using resonant inelastic x-ray scattering (RIXS) and optical spectroscopy. We observe on-site intra-$t_{2g}$ excitations below 2 eV and corresponding overtones with two intra-$t_{2g}$ excitations on adjacent sites, the Mott gap at 2.7 eV, $t_{2g}$-to-$e_g$ excitations above 3 eV, and charge-transfer excitations at still higher energy. The intra-$t_{2g}$ excitation energies are a sensitive measure of $ζ$ and Hund's coupling $J_H$. The sizable value of $ζ\approx$ 0.29 eV places K$_2$ReCl$_6$ into the intermediate coupling regime, but $ζ/J_H \approx 0.6$ is not sufficiently large to drive a pronounced Jahn-Teller effect. We discuss the ground state wavefunction in a Kanamori picture and find that the S = 3/2 multiplet still carries about 97 % of the weight. However, the finite admixture of orbital moment allows for subtle effects. We discuss small temperature-induced changes of the optical data and find evidence for a lowering of the ground state by about 3 meV below the structural phase transitions. △ Less

Submitted 19 November, 2023; originally announced November 2023.

Comments: 16 pages, 14 figures

Information-Theoretic Trust Regions for Stochastic Gradient-Based Optimization

Authors: Philipp Dahlinger, Philipp Becker, Maximilian Hüttenrauch, Gerhard Neumann

Abstract: Stochastic gradient-based optimization is crucial to optimize neural networks. While popular approaches heuristically adapt the step size and direction by rescaling gradients, a more principled approach to improve optimizers requires second-order information. Such methods precondition the gradient using the objective's Hessian. Yet, computing the Hessian is usually expensive and effectively using… ▽ More Stochastic gradient-based optimization is crucial to optimize neural networks. While popular approaches heuristically adapt the step size and direction by rescaling gradients, a more principled approach to improve optimizers requires second-order information. Such methods precondition the gradient using the objective's Hessian. Yet, computing the Hessian is usually expensive and effectively using second-order information in the stochastic gradient setting is non-trivial. We propose using Information-Theoretic Trust Region Optimization (arTuRO) for improved updates with uncertain second-order information. By modeling the network parameters as a Gaussian distribution and using a Kullback-Leibler divergence-based trust region, our approach takes bounded steps accounting for the objective's curvature and uncertainty in the parameters. Before each update, it solves the trust region problem for an optimal step size, resulting in a more stable and faster optimization process. We approximate the diagonal elements of the Hessian from stochastic gradients using a simple recursive least squares approach, constructing a model of the expected Hessian over time using only first-order information. We show that arTuRO combines the fast convergence of adaptive moment-based optimization with the generalization capabilities of SGD. △ Less

Submitted 31 October, 2023; originally announced October 2023.

Multiferroic domain relaxation in (NH4)2[FeCl5(H2O)]

Authors: Sebastian Biesenkamp, Karin Schmalzl, Petra Becker, Ladislav Bohaty, Markus Braden

Abstract: The molecular compound (NH4)2[FeCl5(H2O)] is a type-II multiferroic material, in which incommensurate cycloidal order directly induces ferroelectric polarization. The multiferroic domain kinetics in (NH4)2[FeCl5(H2O)] were studied by time-resolved neutron diffraction experiments utilizing neutron polarization analysis. The temperature and electric-field dependent multiferroic relaxation obeys the… ▽ More The molecular compound (NH4)2[FeCl5(H2O)] is a type-II multiferroic material, in which incommensurate cycloidal order directly induces ferroelectric polarization. The multiferroic domain kinetics in (NH4)2[FeCl5(H2O)] were studied by time-resolved neutron diffraction experiments utilizing neutron polarization analysis. The temperature and electric-field dependent multiferroic relaxation obeys the simple combined Arrhenius-Merz law, which was reported to describe domain kinetics in the prototype multiferroics TbMnO3 and NaFeGe2O6. However, the characteristic time scale of the multiferroic relaxation is considerably larger than those in TbMnO3 or NaFeGe2O6. Temperature-dependent diffraction on (NH4)2[FeCl5(H2O)] reveals the emergence of higher-order and commensurate magnetic contributions upon cooling in the multiferroic phase in zero field. The good agreement with studies of higher-harmonic contributions in the deuterated material indicates that the isotopes only posses a minor impact on the magnetic ordering. But in contrast to similar observations in multiferroic MnWO4, this anharmonic modification of magnetic ordering does not depin multiferroic domain walls or alter the temperature dependence of the multiferroic relaxation. △ Less

Submitted 29 August, 2023; originally announced August 2023.

Comments: 10 pages, 9 Figures

Journal ref: Phys. Rev. B 108, 094417 (2023)

High mobility transport in isotopically-enriched $^{12}$C and $^{13}$C exfoliated graphene

Authors: Shuichi Iwakiri, Jakob Miller, Florian Lang, Jakob Prettenthaler, Takashi Taniguchi, Kenji Watanabe, Sung Sik Lee, Pascal Becker, Detlef Günther, Thomas Ihn, Klaus Ensslin

Abstract: Graphene quantum dots are promising candidates for qubits due to weak spin-orbit and hyperfine interactions. The hyperfine interaction, controllable via isotopic purification, could be the key to further improving the coherence. Here, we use isotopically enriched graphite crystals of both $^{12}$C and $^{13}$C grown by high-pressure-high-temperature method to exfoliate graphene layers. We fabricat… ▽ More Graphene quantum dots are promising candidates for qubits due to weak spin-orbit and hyperfine interactions. The hyperfine interaction, controllable via isotopic purification, could be the key to further improving the coherence. Here, we use isotopically enriched graphite crystals of both $^{12}$C and $^{13}$C grown by high-pressure-high-temperature method to exfoliate graphene layers. We fabricated Hall bar devices and performed quantum transport measurements, revealing mobilities exceeding $10^{5}$$\textrm{cm}^{2}/Vs$ and a long mean free path of microns, which are as high as natural graphene. Shubnikov-de Haas oscillations, quantum Hall effect up to the filling factor of one, and Brown-Zak oscillations due to the alignment of hBN and graphene are observed thanks to the high mobility. These results constitute a material platform for physics and engineering of isotopically-enriched graphene qubits. △ Less

Submitted 16 August, 2023; originally announced August 2023.

Comments: 6 pages, 2 figures

Journal ref: Phys. Rev. Research 5, 043212 (2023)

Electronic excitations in $5d^4$ J=0 Os$^{4+}$ halides studied by RIXS and optical spectroscopy

Authors: P. Warzanowski, M. Magnaterra, P. Stein, G. Schlicht, Q. Faure, Ch. J. Sahle, T. Lorenz, P. Becker, L. Bohaty, M. Moretti Sala, G. Monaco, P. H. M. van Loosdrecht, M. Grüninger

Abstract: We demonstrate that the cubic antifluorite-type halides K$_2$OsCl$_6$, K$_2$OsBr$_6$, and Rb$_2$OsBr$_6$ are excellent realizations of non-magnetic J=0 compounds. The magnetic susceptibility shows the corresponding Van-Vleck type behavior and no sign of defects. We investigate the electronic excitations with two complementary techniques, resonant inelastic x-ray scattering (RIXS) and optical spect… ▽ More We demonstrate that the cubic antifluorite-type halides K$_2$OsCl$_6$, K$_2$OsBr$_6$, and Rb$_2$OsBr$_6$ are excellent realizations of non-magnetic J=0 compounds. The magnetic susceptibility shows the corresponding Van-Vleck type behavior and no sign of defects. We investigate the electronic excitations with two complementary techniques, resonant inelastic x-ray scattering (RIXS) and optical spectroscopy. This powerful combination allows us to thoroughly study, e.g., on-site intra-$t_{2g}$ excitations and $t_{2g}$-to-$e_g$ excitations as well as inter-site excitations across the Mott gap and an exciton below the gap. In this way, we determine the electronic parameters with high accuracy, altogether yielding a comprehensive picture. In K$_2$OsCl$_6$, we find the spin-orbit coupling constant $ζ$=0.34 eV, Hund's coupling $J_H$=0.43 eV, the onset of excitations across the Mott gap at $Δ$=2.2 eV, the cubic crystal-field splitting 10Dq=3.3 eV, and the charge-transfer energy $Δ_{CT}$=4.6 eV. With $J_H/ζ$=1.3, K$_2$OsCl$_6$ is in the intermediate-coupling regime. In a $t_{2g}$-only Kanamori picture, the above values correspond to $ζ^{eff}$=0.41 eV and $J_H^{eff}$=0.28 eV, which is very close to results reported for related $5d^4$ iridates. In the tetragonal phase at 5 K, the non-cubic crystal field causes a peak splitting of the J=1 state as small as 4 meV. Compared to K$_2$OsCl$_6$, the bromides K$_2$OsBr$_6$ and Rb$_2$OsBr$_6$ show about 12-14 % smaller values of 10Dq and $Δ_{CT}$, while the spin-orbit-entangled intra-$t_{2g}$ excitations below 2 eV and hence $ζ$ and $J_H$ are reduced by less than 4 %. Furthermore, the Mott gap in K$_2$OsBr$_6$ is reduced to about 1.8 eV. △ Less

Submitted 21 June, 2023; originally announced June 2023.

Comments: 14 pages, 14 figures

Journal ref: Phys. Rev. B 108, 125120 (2023)

Beyond Deep Ensembles: A Large-Scale Evaluation of Bayesian Deep Learning under Distribution Shift

Authors: Florian Seligmann, Philipp Becker, Michael Volpp, Gerhard Neumann

Abstract: Bayesian deep learning (BDL) is a promising approach to achieve well-calibrated predictions on distribution-shifted data. Nevertheless, there exists no large-scale survey that evaluates recent SOTA methods on diverse, realistic, and challenging benchmark tasks in a systematic manner. To provide a clear picture of the current state of BDL research, we evaluate modern BDL algorithms on real-world da… ▽ More Bayesian deep learning (BDL) is a promising approach to achieve well-calibrated predictions on distribution-shifted data. Nevertheless, there exists no large-scale survey that evaluates recent SOTA methods on diverse, realistic, and challenging benchmark tasks in a systematic manner. To provide a clear picture of the current state of BDL research, we evaluate modern BDL algorithms on real-world datasets from the WILDS collection containing challenging classification and regression tasks, with a focus on generalization capability and calibration under distribution shift. We compare the algorithms on a wide range of large, convolutional and transformer-based neural network architectures. In particular, we investigate a signed version of the expected calibration error that reveals whether the methods are over- or under-confident, providing further insight into the behavior of the methods. Further, we provide the first systematic evaluation of BDL for fine-tuning large pre-trained models, where training from scratch is prohibitively expensive. Finally, given the recent success of Deep Ensembles, we extend popular single-mode posterior approximations to multiple modes by the use of ensembles. While we find that ensembling single-mode approximations generally improves the generalization capability and calibration of the models by a significant margin, we also identify a failure mode of ensembles when finetuning large transformer-based language models. In this setting, variational inference based approaches such as last-layer Bayes By Backprop outperform other methods in terms of accuracy by a large margin, while modern approximate inference algorithms such as SWAG achieve the best calibration. △ Less

Submitted 24 October, 2023; v1 submitted 21 June, 2023; originally announced June 2023.

Comments: Code at https://github.com/Feuermagier/Beyond_Deep_Ensembles

A Mean-Field Method for Generic Conductance-Based Integrate-and-Fire Neurons with Finite Timescales

Authors: Marcelo P. Becker, Marco A. P. Idiart

Abstract: The construction of transfer functions in theoretical neuroscience plays an important role in determining the spiking rate behavior of neurons in networks. These functions can be obtained through various fitting methods, but the biological relevance of the parameters is not always clear. However, for stationary inputs, such functions can be obtained without the adjustment of free parameters by usi… ▽ More The construction of transfer functions in theoretical neuroscience plays an important role in determining the spiking rate behavior of neurons in networks. These functions can be obtained through various fitting methods, but the biological relevance of the parameters is not always clear. However, for stationary inputs, such functions can be obtained without the adjustment of free parameters by using mean-field methods. In this work, we expand current Fokker-Planck approaches to account for the concurrent influence of colored and multiplicative noise terms on generic conductance-based integrate-and-fire neurons. We reduce the resulting stochastic system from the application of the diffusion approximation to a one-dimensional Langevin equation. An effective Fokker-Planck is then constructed using Fox Theory, which is solved numerically to obtain the transfer function. The solution is capable of reproducing the transfer function behavior of simulated neurons across a wide range of parameters. The method can also be easily extended to account for different sources of noise with various multiplicative terms, and it can be used in other types of problems in principle. △ Less

Submitted 24 May, 2023; originally announced May 2023.

Comments: 11 pages, 6 figures, research article

Curriculum-Based Imitation of Versatile Skills

Authors: Maximilian Xiling Li, Onur Celik, Philipp Becker, Denis Blessing, Rudolf Lioutikov, Gerhard Neumann

Abstract: Learning skills by imitation is a promising concept for the intuitive teaching of robots. A common way to learn such skills is to learn a parametric model by maximizing the likelihood given the demonstrations. Yet, human demonstrations are often multi-modal, i.e., the same task is solved in multiple ways which is a major challenge for most imitation learning methods that are based on such a maximu… ▽ More Learning skills by imitation is a promising concept for the intuitive teaching of robots. A common way to learn such skills is to learn a parametric model by maximizing the likelihood given the demonstrations. Yet, human demonstrations are often multi-modal, i.e., the same task is solved in multiple ways which is a major challenge for most imitation learning methods that are based on such a maximum likelihood (ML) objective. The ML objective forces the model to cover all data, it prevents specialization in the context space and can cause mode-averaging in the behavior space, leading to suboptimal or potentially catastrophic behavior. Here, we alleviate those issues by introducing a curriculum using a weight for each data point, allowing the model to specialize on data it can represent while incentivizing it to cover as much data as possible by an entropy bonus. We extend our algorithm to a Mixture of (linear) Experts (MoE) such that the single components can specialize on local context regions, while the MoE covers all data points. We evaluate our approach in complex simulated and real robot control tasks and show it learns from versatile human demonstrations and significantly outperforms current SOTA methods. A reference implementation can be found at https://github.com/intuitive-robots/ml-cur △ Less

Submitted 11 April, 2023; originally announced April 2023.

Journal ref: 2023 IEEE International Conference on Robotics and Automation (ICRA)

Room temperature quantum bit storage exceeding 39 minutes using ionized donors in 28-silicon

Authors: Kamyar Saeedi, Stephanie Simmons, Jeff Z. Salvail, Phillip Dluhy, Helge Riemann, Nikolai V. Abrosimov, Peter Becker, Hans-Joachim Pohl, John J. L. Morton, Mike L. W. Thewalt

Abstract: Quantum memories capable of storing and retrieving coherent information for extended times at room temperature would enable a host of new technologies. Electron and nuclear spin qubits using shallow neutral donors in semiconductors have been studied extensively but are limited to low temperatures ($\le$10 K); however, the nuclear spins of ionized donors have potential for high temperature operatio… ▽ More Quantum memories capable of storing and retrieving coherent information for extended times at room temperature would enable a host of new technologies. Electron and nuclear spin qubits using shallow neutral donors in semiconductors have been studied extensively but are limited to low temperatures ($\le$10 K); however, the nuclear spins of ionized donors have potential for high temperature operation. We use optical methods and dynamical decoupling to realize this potential for an ensemble of 31P donors in isotopically purified 28Si and observe a room temperature coherence time of over 39 minutes. We further show that a coherent spin superposition can be cycled from 4.2 K to room temperature and back, and report a cryogenic coherence time of 3 hours in the same system. △ Less

Submitted 30 March, 2023; originally announced March 2023.

Comments: 5 pages, 4 figures

Journal ref: Science (2013) Vol 342, Issue 6160 pp. 830-833

Combining Reconstruction and Contrastive Methods for Multimodal Representations in RL

Authors: Philipp Becker, Sebastian Mossburger, Fabian Otto, Gerhard Neumann

Abstract: Learning self-supervised representations using reconstruction or contrastive losses improves performance and sample complexity of image-based and multimodal reinforcement learning (RL). Here, different self-supervised loss functions have distinct advantages and limitations depending on the information density of the underlying sensor modality. Reconstruction provides strong learning signals but is… ▽ More Learning self-supervised representations using reconstruction or contrastive losses improves performance and sample complexity of image-based and multimodal reinforcement learning (RL). Here, different self-supervised loss functions have distinct advantages and limitations depending on the information density of the underlying sensor modality. Reconstruction provides strong learning signals but is susceptible to distractions and spurious information. While contrastive approaches can ignore those, they may fail to capture all relevant details and can lead to representation collapse. For multimodal RL, this suggests that different modalities should be treated differently based on the amount of distractions in the signal. We propose Contrastive Reconstructive Aggregated representation Learning (CoRAL), a unified framework enabling us to choose the most appropriate self-supervised loss for each sensor modality and allowing the representation to better focus on relevant aspects. We evaluate CoRAL's benefits on a wide range of tasks with images containing distractions or occlusions, a new locomotion suite, and a challenging manipulation suite with visually realistic distractions. Our results show that learning a multimodal representation by combining contrastive and reconstruction-based losses can significantly improve performance and solve tasks that are out of reach for more naive representation learning approaches and other recent baselines. △ Less

Submitted 26 June, 2024; v1 submitted 10 February, 2023; originally announced February 2023.

Comments: Published in "Reinforcement Learning Conference (RLC)", August 2024

NFRsTDO v1.2's Terms, Properties, and Relationships -- A Top-Domain Non-Functional Requirements Ontology

Authors: Luis Olsina, María Fernanda Papa, Pablo Becker

Abstract: This preprint specifies and defines all the Terms, Properties, and Relationships of NFRsTDO (Non-Functional Requirements Top-Domain Ontology). NFRsTDO v1.2, whose UML conceptualization is shown in Figure 1 is a slightly updated version of its predecessor, namely NFRsTDO v1.1. NFRsTDO is an ontology mainly devoted to quality (non-functional) requirements and quality/cost views, which is placed at t… ▽ More This preprint specifies and defines all the Terms, Properties, and Relationships of NFRsTDO (Non-Functional Requirements Top-Domain Ontology). NFRsTDO v1.2, whose UML conceptualization is shown in Figure 1 is a slightly updated version of its predecessor, namely NFRsTDO v1.1. NFRsTDO is an ontology mainly devoted to quality (non-functional) requirements and quality/cost views, which is placed at the top-domain level in the context of a multilayer ontological architecture called FCD-OntoArch (Foundational, Core, Domain, and instance Ontological Architecture for sciences). Figure 2 depicts its five tiers, which entail Foundational, Core, Top-Domain, Low-Domain, and Instance. Each level is populated with ontological components or, in other words, ontologies. Ontologies at the same level can be related to each other, except at the foundational level, where only ThingFO (Thing Foundational Ontology) is found. In addition, ontologies' terms and relationships at lower levels can be semantically enriched by ontologies' terms and relationships from the higher levels. NFRsTDO's terms and relationships are mainly extended/reused from ThingFO, SituationCO (Situation Core Ontology), ProcessCO (Process Core Ontology), and FRsTDO (Functional Requirements Top-Domain Ontology). Stereotypes are the used mechanism for enriching NFRsTDO terms. Note that annotations of updates from the previous version (NFRsTDO v1.1) to the current one (v1.2) can be found in Appendix A. △ Less

Submitted 2 February, 2023; originally announced February 2023.

Comments: 9 pages and 2 figures

K-D Bonsai: ISA-Extensions to Compress K-D Trees for Autonomous Driving Tasks

Authors: Pedro H. E. Becker, José María Arnau, Antonio González

Abstract: Autonomous Driving (AD) systems extensively manipulate 3D point clouds for object detection and vehicle localization. Thereby, efficient processing of 3D point clouds is crucial in these systems. In this work we propose K-D Bonsai, a technique to cut down memory usage during radius search, a critical building block of point cloud processing. K-D Bonsai exploits value similarity in the data structu… ▽ More Autonomous Driving (AD) systems extensively manipulate 3D point clouds for object detection and vehicle localization. Thereby, efficient processing of 3D point clouds is crucial in these systems. In this work we propose K-D Bonsai, a technique to cut down memory usage during radius search, a critical building block of point cloud processing. K-D Bonsai exploits value similarity in the data structure that holds the point cloud (a k-d tree) to compress the data in memory. K-D Bonsai further compresses the data using a reduced floating-point representation, exploiting the physically limited range of point cloud values. For easy integration into nowadays systems, we implement K-D Bonsai through Bonsai-extensions, a small set of new CPU instructions to compress, decompress, and operate on points. To maintain baseline safety levels, we carefully craft the Bonsai-extensions to detect precision loss due to compression, allowing re-computation in full precision to take place if necessary. Therefore, K-D Bonsai reduces data movement, improving performance and energy efficiency, while guaranteeing baseline accuracy and programmability. We evaluate K-D Bonsai over the euclidean cluster task of Autoware.ai, a state-of-the-art software stack for AD. We achieve an average of 9.26% improvement in end-to-end latency, 12.19% in tail latency, and a reduction of 10.84% in energy consumption. Differently from expensive accelerators proposed in related work, K-D Bonsai improves radius search with minimal area increase (0.36%). △ Less

Submitted 30 August, 2023; v1 submitted 1 February, 2023; originally announced February 2023.

MSC Class: Article No. 18; 2018 Related DOI: https://doi.org/10.1145/3243176.3243184 Focus to learn more

Journal ref: ISCA'23 Proceedings of the 50th Annual International Symposium on Computer Architecture, Article No. 20, 2023

A new three-ranged Gogny interaction in touch with pion exchange

Authors: L. Batail, D. Davesne, S. Péru, P. Becker, A. Pastore, J. Navarro

Abstract: We present a new Gogny-type finite-range effective interaction including a third Gaussian in the central term. Based on simple arguments valid for an arbitrary radial form factor, the three ranges are fixed from physical grounds, relating them to one-boson exchange interactions. Moreover, some parameters of the longest range are fixed through the G-matrix elements of the One Pion Exchange Potentia… ▽ More We present a new Gogny-type finite-range effective interaction including a third Gaussian in the central term. Based on simple arguments valid for an arbitrary radial form factor, the three ranges are fixed from physical grounds, relating them to one-boson exchange interactions. Moreover, some parameters of the longest range are fixed through the G-matrix elements of the One Pion Exchange Potential. On top of giving a fairly good description of atomic nuclei properties, the resulting interaction leads to a remarkable improvement of some infinite matter properties that are relevant for astrophysical calculations. △ Less

Submitted 1 December, 2022; originally announced December 2022.

An Analytical Fourier-Transformation Model for the Production of Hard and Soft X-Ray Time Lags in AGNs: Application to 1H 0707-495

Authors: David C. Baughman, Peter A. Becker

Abstract: The variability of the X-ray emission from active galactic nuclei is often characterized using time lags observed between soft and hard energy bands in the detector. The time lags are usually computed using the complex cross spectrum, which is based on the Fourier transforms of the hard and soft time series data. It has been noted that some active galactic nuclei display soft X-ray time lags, in a… ▽ More The variability of the X-ray emission from active galactic nuclei is often characterized using time lags observed between soft and hard energy bands in the detector. The time lags are usually computed using the complex cross spectrum, which is based on the Fourier transforms of the hard and soft time series data. It has been noted that some active galactic nuclei display soft X-ray time lags, in addition to the more ubiquitous hard lags. Hard time lags are thought to be produced via propagating fluctuations, spatial reverberation, or via the thermal Comptonization of soft seed photons injected into a hot electron cloud. The physical origin of the soft lags has been a subject of debate over the last decade. Currently, the reverberation interpretation is recognized as a leading theory. In this paper, we explore the alternative possibility that the soft X-ray time lags result partially from the thermal and bulk Comptonization of monochromatic seed photons, which in the case of the narrow-line Seyfert 1 galaxy 1H 0707-495, may correlate with fluorescence of iron L-line emission. In our model, the seed photons are injected into a hot, quasi-spherical corona in the inner region of the accretion flow. We develop an exact, time-dependent analytical model for the thermal and bulk Comptonization of the seed photons based on a Fourier-transformed radiation transport equation, and we demonstrate that the model successfully reproduces both the hard and soft time lags observed from 1H 0707-495. △ Less

Submitted 26 November, 2022; originally announced November 2022.

Journal ref: Astrophysical Journal, 2022, volume 932, page 113

A Generalized Analytical Model For Thermal And Bulk Comptonization In Accretion-Powered X-Ray Pulsars

Authors: Peter A. Becker, Michael T. Wolff

Abstract: We develop a new theoretical model describing the formation of the radiation spectrum in accretion-powered X-ray pulsars as a result of bulk and thermal Comptonization of photons in the accretion column. The new model extends the previous model developed by the authors in four ways: (1) we utilize a conical rather than cylindrical geometry; (2) the radiation components emitted from the column wall… ▽ More We develop a new theoretical model describing the formation of the radiation spectrum in accretion-powered X-ray pulsars as a result of bulk and thermal Comptonization of photons in the accretion column. The new model extends the previous model developed by the authors in four ways: (1) we utilize a conical rather than cylindrical geometry; (2) the radiation components emitted from the column wall and the column top are computed separately; (3) the model allows for a non-zero impact velocity at the stellar surface; and (4) the velocity profile of the gas merges with Newtonian free-fall far from the star. We show that these extensions allow the new model to simulate sources over a wide range of accretion rates. The model is based on a rigorous mathematical approach in which we obtain an exact series solution for the Green's function describing the reprocessing of monochromatic seed photons. Emergent spectra are then computed by convolving the Green's function with bremsstrahlung, cyclotron, and blackbody photon sources. The range of the new model is demonstrated via applications to the high-luminosity source Her X-1, and the low-luminosity source X Per. The new model suggests that the observed increase in spectral hardness associated with increasing luminosity in Her X-1 may be due to a decrease in the surface impact velocity, which increases the $P$d$V$ work done on the radiation field by the gas. △ Less

Submitted 25 November, 2022; originally announced November 2022.

Journal ref: Astrophysical Journal, 2022, volume 939, page 67

RIXS interferometry and the role of disorder in the quantum magnet Ba$_3$Ti$_{3-x}$Ir$_{x}$O$_9$

Authors: M. Magnaterra, M. Moretti Sala, G. Monaco, P. Becker, M. Hermanns, P. Warzanowski, T. Lorenz, D. I. Khomskii, P. H. M. van Loosdrecht, J. van den Brink, M. Grüninger

Abstract: Motivated by several claims of spin-orbit driven spin-liquid physics in hexagonal Ba$_3$Ti$_{3-x}$Ir$_x$O$_9$ hosting Ir2O9 dimers, we report on resonant inelastic x-ray scattering (RIXS) at the Ir L3 edge for different x. We demonstrate that magnetism in Ba$_3$Ti$_{3-x}$Ir$_x$O$_9$ is governed by an unconventional realization of strong disorder, where cation disorder affects the character of the… ▽ More Motivated by several claims of spin-orbit driven spin-liquid physics in hexagonal Ba$_3$Ti$_{3-x}$Ir$_x$O$_9$ hosting Ir2O9 dimers, we report on resonant inelastic x-ray scattering (RIXS) at the Ir L3 edge for different x. We demonstrate that magnetism in Ba$_3$Ti$_{3-x}$Ir$_x$O$_9$ is governed by an unconventional realization of strong disorder, where cation disorder affects the character of the local moments. RIXS interferometry, studying the RIXS intensity over a broad range of transferred momentum q, is ideally suited to assign different excitations to different Ir sites. We find pronounced Ir-Ti site mixing. Both ions are distributed over two crystallographically inequivalent sites, giving rise to a coexistence of quasimolecular singlet states on Ir2O9 dimers and spin-orbit entangled j=1/2 moments of 5d$^5$ Ir$^{4+}$ ions. RIXS reveals different kinds of strong magnetic couplings for different bonding geometries, highlighting the role of cation disorder for the suppression of long-range magnetic order in this family of compounds. △ Less

Submitted 24 November, 2022; originally announced November 2022.

Comments: 12 pages, 9 figures

Journal ref: Phys. Rev. Research 5, 013167 (2023)

On Uncertainty in Deep State Space Models for Model-Based Reinforcement Learning

Authors: Philipp Becker, Gerhard Neumann

Abstract: Improved state space models, such as Recurrent State Space Models (RSSMs), are a key factor behind recent advances in model-based reinforcement learning (RL). Yet, despite their empirical success, many of the underlying design choices are not well understood. We show that RSSMs use a suboptimal inference scheme and that models trained using this inference overestimate the aleatoric uncertainty of… ▽ More Improved state space models, such as Recurrent State Space Models (RSSMs), are a key factor behind recent advances in model-based reinforcement learning (RL). Yet, despite their empirical success, many of the underlying design choices are not well understood. We show that RSSMs use a suboptimal inference scheme and that models trained using this inference overestimate the aleatoric uncertainty of the ground truth system. We find this overestimation implicitly regularizes RSSMs and allows them to succeed in model-based RL. We postulate that this implicit regularization fulfills the same functionality as explicitly modeling epistemic uncertainty, which is crucial for many other model-based RL approaches. Yet, overestimating aleatoric uncertainty can also impair performance in cases where accurately estimating it matters, e.g., when we have to deal with occlusions, missing observations, or fusing sensor modalities at different frequencies. Moreover, the implicit regularization is a side-effect of the inference scheme and not the result of a rigorous, principled formulation, which renders analyzing or improving RSSMs difficult. Thus, we propose an alternative approach building on well-understood components for modeling aleatoric and epistemic uncertainty, dubbed Variational Recurrent Kalman Network (VRKN). This approach uses Kalman updates for exact smoothing inference in a latent space and Monte Carlo Dropout to model epistemic uncertainty. Due to the Kalman updates, the VRKN can naturally handle missing observations or sensor fusion problems with varying numbers of observations per time step. Our experiments show that using the VRKN instead of the RSSM improves performance in tasks where appropriately capturing aleatoric uncertainty is crucial while matching it in the deterministic standard benchmarks. △ Less

Submitted 17 October, 2022; originally announced October 2022.

Comments: Published in TMLR, October 2022

Inferring Versatile Behavior from Demonstrations by Matching Geometric Descriptors

Authors: Niklas Freymuth, Nicolas Schreiber, Philipp Becker, Aleksandar Taranovic, Gerhard Neumann

Abstract: Humans intuitively solve tasks in versatile ways, varying their behavior in terms of trajectory-based planning and for individual steps. Thus, they can easily generalize and adapt to new and changing environments. Current Imitation Learning algorithms often only consider unimodal expert demonstrations and act in a state-action-based setting, making it difficult for them to imitate human behavior i… ▽ More Humans intuitively solve tasks in versatile ways, varying their behavior in terms of trajectory-based planning and for individual steps. Thus, they can easily generalize and adapt to new and changing environments. Current Imitation Learning algorithms often only consider unimodal expert demonstrations and act in a state-action-based setting, making it difficult for them to imitate human behavior in case of versatile demonstrations. Instead, we combine a mixture of movement primitives with a distribution matching objective to learn versatile behaviors that match the expert's behavior and versatility. To facilitate generalization to novel task configurations, we do not directly match the agent's and expert's trajectory distributions but rather work with concise geometric descriptors which generalize well to unseen task configurations. We empirically validate our method on various robot tasks using versatile human demonstrations and compare to imitation learning algorithms in a state-action setting as well as a trajectory-based setting. We find that the geometric descriptors greatly help in generalizing to new task configurations and that combining them with our distribution-matching objective is crucial for representing and reproducing versatile behavior. △ Less

Submitted 9 November, 2022; v1 submitted 17 October, 2022; originally announced October 2022.

Comments: Accepted as a poster at the 6th Conference on Robot Learning (CoRL), 2022

Single-crystal study of the honeycomb XXZ magnet BaCo$_2$(PO$_4$)$_2$ in magnetic fields

Authors: Xiao Wang, Rohit Sharma, Petra Becker, Ladislav Bohatý, Thomas Lorenz

Abstract: We present a study of high-quality BaCo$_2$(PO$_4$)$_2$ single crystals via magnetization, heat-capacity, thermal-expansion and magnetostriction measurements. Sharp anomalies in the thermodynamic properties at $T_N=3.4\,$K reveal a long-range antiferromagnetic order in these single-crystalline samples, which is absent in polycrystalline BaCo$_2$(PO$_4$)$_2$. The temperature dependent magnetic susc… ▽ More We present a study of high-quality BaCo$_2$(PO$_4$)$_2$ single crystals via magnetization, heat-capacity, thermal-expansion and magnetostriction measurements. Sharp anomalies in the thermodynamic properties at $T_N=3.4\,$K reveal a long-range antiferromagnetic order in these single-crystalline samples, which is absent in polycrystalline BaCo$_2$(PO$_4$)$_2$. The temperature dependent magnetic susceptibilities for in-plane and out-of-plane magnetic fields are strongly anisotropic and reveal a pronounced easy-plane anisotropy. A Curie-Weiss analysis implies strong orbital magnetism, as it is known from the sister compound BaCo$_2$(AsO$_4$)$_2$ that is discussed as a potential Kitaev spin-liquid material. When applying in-plane magnetic fields at low temperature, BaCo$_2$(PO$_4$)$_2$ is driven to another ordered phase at a critical field $H_{C1}\approx 0.11\,$T and then undergoes a further field-induced transition to a highly polarized paramagnetic phase at $H_{C2}\approx 0.3\,$T, which is again similar to the case of BaCo$_2$(AsO$_4$)$_2$. In addition, our lowest-temperature data reveal that the field-induced transitions in BaCo$_2$(PO$_4$)$_2$ become dominated by thermally assisted domain-wall motion. △ Less

Submitted 7 February, 2023; v1 submitted 30 September, 2022; originally announced September 2022.

Comments: 6 pages, 4 figures

Journal ref: Physical Review Materials 7, 024402 (2023)

Local symmetry breaking and low-energy continuum in K$_2$ReCl$_6 $

Authors: P. Stein, T. C. Koethe, L. Bohaty, P. Becker, M. Grüninger, P. H. M. van Loosdrecht

Abstract: Using polarization selective spontaneous Raman scattering, we have investigated the 5$d$ transition metal compound K$_2$ReCl$_6$ which displays a series of structural phase transitions. We observe a violation of the Raman selection rules in the cubic high temperature phase as well as a low-energy scattering continuum persistent throughout the investigated temperature range from 300 down to 5 K. Th… ▽ More Using polarization selective spontaneous Raman scattering, we have investigated the 5$d$ transition metal compound K$_2$ReCl$_6$ which displays a series of structural phase transitions. We observe a violation of the Raman selection rules in the cubic high temperature phase as well as a low-energy scattering continuum persistent throughout the investigated temperature range from 300 down to 5 K. The continuum couples to one of the phonon modes at temperatures above the lowest structural phase transition at 76 K. We propose a common origin of these observations caused by local fluctuations in the orientation of the ReCl$_6$ octahedra which locally break the long-range cubic symmetry. Consistent results from the related non-magnetic compound K$_2$SnCl$_6$ support this interpretation. △ Less

Submitted 7 June, 2023; v1 submitted 13 September, 2022; originally announced September 2022.

Comments: 6 pages, 7 figures

Journal ref: Physical Review B 107, 214301 (2023)

Interplay of magnetic order and ferroelasticity in the spin-orbit coupled antiferromagnet K$_2$ReCl$_6$

Authors: Alexandre Bertin, Tusharkanti Dey, Daniel Brüning, Dmitry Gorkov, Kevin Jenni, Astin Krause, Petra Becker, Ladislav Bohaty, Daniel Khomskii, Vladimir Pomjakushin, Lukas Keller, Markus Braden, Thomas Lorenz

Abstract: The magnetic and structural phase transitions occurring in K$_2$ReCl$_6$ were studied by macroscopic and microscopic techniques. Structural phase transitions associated with rotations of the ReCl$_6$ octahedra lower the symmetry from cubic to monoclinic, form ferroelastic domains, and are visible in susceptibility, specific heat and thermal expansion measurements. In the antiferromagnetically orde… ▽ More The magnetic and structural phase transitions occurring in K$_2$ReCl$_6$ were studied by macroscopic and microscopic techniques. Structural phase transitions associated with rotations of the ReCl$_6$ octahedra lower the symmetry from cubic to monoclinic, form ferroelastic domains, and are visible in susceptibility, specific heat and thermal expansion measurements. In the antiferromagnetically ordered state slightly below $T_{\rm N}$=12\,K these domains can be rearranged by a magnetic field inducing a relative elongation of the polydomain crystal parallel to the field of 0.6\%. At zero field the magnetic structure in K$_2$ReCl$_6$ does not exhibit a weak ferromagnetic component, but at large magnetic field a distinct magnetic structure with a finite weak ferromagnetic component is stabilized. High magnetic fields rearrange the domains in the crystal to align the weak ferromagnetic moment parallel to the field. The altered domain structure with the crystal elongation is abruptly suppressed at lower temperature but persists upon heating to well above $T_{\rm N}$. However, heating above the lowest structural phase transition and successive cooling restore the initial shape. △ Less

Submitted 8 March, 2024; v1 submitted 22 July, 2022; originally announced July 2022.

Comments: 14 pages, 10 figures; In this revised version of our manuscript, neutron powder diffraction data have been included and the discussion of the multidomain structure has been extended

Journal ref: Phys. Rev. B 109, 094409 (2024)

Quasimolecular electronic structure of the spin-liquid candidate Ba$_3$InIr$_2$O$_9$

Authors: A. Revelli, M. Moretti Sala, G. Monaco, M. Magnaterra, J. Attig, L. Peterlini, T. Dey, A. A. Tsirlin, P. Gegenwart, T. Fröhlich, M. Braden, C. Grams, J. Hemberger, P. Becker, P. H. M. van Loosdrecht, D. I. Khomskii, J. van den Brink, M. Hermanns, M. Grüninger

Abstract: The mixed-valent iridate Ba3InIr2O9 has been discussed as a promising candidate for quantum spin-liquid behavior. The compound exhibits Ir$^{4.5+}$ ions in face-sharing IrO6 octahedra forming Ir2O9 dimers with three t2g holes per dimer. Our results establish Ba3InIr2O9 as a cluster Mott insulator. Strong intra-dimer hop** delocalizes the three t2g holes in quasi-molecular dimer states while inte… ▽ More The mixed-valent iridate Ba3InIr2O9 has been discussed as a promising candidate for quantum spin-liquid behavior. The compound exhibits Ir$^{4.5+}$ ions in face-sharing IrO6 octahedra forming Ir2O9 dimers with three t2g holes per dimer. Our results establish Ba3InIr2O9 as a cluster Mott insulator. Strong intra-dimer hop** delocalizes the three t2g holes in quasi-molecular dimer states while inter-dimer charge fluctuations are suppressed by Coulomb repulsion. The magnetism of Ba3InIr2O9 emerges from spin-orbit entangled quasi-molecular moments with yet unexplored interactions, opening up a new route to unconventional magnetic properties of 5d compounds. Using single-crystal x-ray diffraction we find the monoclinic space group C2/c already at room temperature. Dielectric spectroscopy shows insulating behavior. Resonant inelastic x-ray scattering (RIXS) reveals a rich excitation spectrum below 1.5 eV with a sinusoidal dynamical structure factor that unambiguously demonstrates the quasi-molecular character of the electronic states. Below 0.3 eV, we observe a series of excitations. According to exact diagonalization calculations, such low-energy excitations reflect the proximity of Ba3InIr2O9 to a hop**-induced phase transition based on the condensation of a quasi-molecular spin-orbit exciton. The dimer ground state roughly hosts two holes in a bonding j=1/2 orbital and the third hole in a bonding j=3/2 orbital. △ Less

Submitted 28 November, 2022; v1 submitted 11 July, 2022; originally announced July 2022.

Comments: 14 pages, 10 figures, V2: published version, minor changes

Journal ref: Phys. Rev. B 106, 155107 (2022)

An additive framework for kirigami design

Authors: Levi H. Dudte, Gary P. T. Choi, Kaitlyn P. Becker, L. Mahadevan

Abstract: We present an additive approach for the inverse design of kirigami-based mechanical metamaterials by focusing on the empty (negative) spaces instead of the solid tiles. By considering each negative space as a four-bar linkage, we identify a simple recursive relationship between adjacent linkages, yielding an efficient method for creating kirigami patterns. This allows us to solve the kirigami desi… ▽ More We present an additive approach for the inverse design of kirigami-based mechanical metamaterials by focusing on the empty (negative) spaces instead of the solid tiles. By considering each negative space as a four-bar linkage, we identify a simple recursive relationship between adjacent linkages, yielding an efficient method for creating kirigami patterns. This allows us to solve the kirigami design problem using elementary linear algebra, with compatibility, reconfigurability and rigid-deployability encoded into an iterative procedure involving simple matrix multiplications. The resulting linear design strategy circumvents the solution of a non-convex global optimization problem and allows us to control the degrees of freedom in the deployment angle field, linkage offsets and boundary conditions. We demonstrate this by creating a large variety of rigid-deployable, compact, reconfigurable kirigami patterns. We then realize our kirigami designs physically using two simple but effective fabrication strategies with very different materials. Altogether, our additive approaches present routes for efficient mechanical metamaterial design and fabrication based on ori/kirigami art forms. △ Less

Submitted 25 May, 2023; v1 submitted 5 July, 2022; originally announced July 2022.

Journal ref: Nature Computational Science, 3, 443-454 (2023)

Hidden Parameter Recurrent State Space Models For Changing Dynamics Scenarios

Authors: Vaisakh Shaj, Dieter Buchler, Rohit Sonker, Philipp Becker, Gerhard Neumann

Abstract: Recurrent State-space models (RSSMs) are highly expressive models for learning patterns in time series data and system identification. However, these models assume that the dynamics are fixed and unchanging, which is rarely the case in real-world scenarios. Many control applications often exhibit tasks with similar but not identical dynamics which can be modeled as a latent variable. We introduce… ▽ More Recurrent State-space models (RSSMs) are highly expressive models for learning patterns in time series data and system identification. However, these models assume that the dynamics are fixed and unchanging, which is rarely the case in real-world scenarios. Many control applications often exhibit tasks with similar but not identical dynamics which can be modeled as a latent variable. We introduce the Hidden Parameter Recurrent State Space Models (HiP-RSSMs), a framework that parametrizes a family of related dynamical systems with a low-dimensional set of latent factors. We present a simple and effective way of learning and performing inference over this Gaussian graphical model that avoids approximations like variational inference. We show that HiP-RSSMs outperforms RSSMs and competing multi-task models on several challenging robotic benchmarks both on real-world systems and simulations. △ Less

Submitted 12 October, 2023; v1 submitted 29 June, 2022; originally announced June 2022.

Comments: Published at the International Conference on Learning Representations, ICLR 2022

End-to-End Learning of Hybrid Inverse Dynamics Models for Precise and Compliant Impedance Control

Authors: Moritz Reuss, Niels van Duijkeren, Robert Krug, Philipp Becker, Vaisakh Shaj, Gerhard Neumann

Abstract: It is well-known that inverse dynamics models can improve tracking performance in robot control. These models need to precisely capture the robot dynamics, which consist of well-understood components, e.g., rigid body dynamics, and effects that remain challenging to capture, e.g., stick-slip friction and mechanical flexibilities. Such effects exhibit hysteresis and partial observability, rendering… ▽ More It is well-known that inverse dynamics models can improve tracking performance in robot control. These models need to precisely capture the robot dynamics, which consist of well-understood components, e.g., rigid body dynamics, and effects that remain challenging to capture, e.g., stick-slip friction and mechanical flexibilities. Such effects exhibit hysteresis and partial observability, rendering them, particularly challenging to model. Hence, hybrid models, which combine a physical prior with data-driven approaches are especially well-suited in this setting. We present a novel hybrid model formulation that enables us to identify fully physically consistent inertial parameters of a rigid body dynamics model which is paired with a recurrent neural network architecture, allowing us to capture unmodeled partially observable effects using the network memory. We compare our approach against state-of-the-art inverse dynamics models on a 7 degree of freedom manipulator. Using data sets obtained through an optimal experiment design approach, we study the accuracy of offline torque prediction and generalization capabilities of joint learning methods. In control experiments on the real system, we evaluate the model as a feed-forward term for impedance control and show the feedback gains can be drastically reduced to achieve a given tracking accuracy. △ Less

Submitted 27 May, 2022; originally announced May 2022.

Comments: Accepted for publication at Robotics: Science and System XVIII (RSS), year 2022. Paper length is 13 pages (i.e. 9 pages of technical content, 1 page of the Bibliography/References and 3 pages of Appendix)

Multi-Stage Phase-Segregation of Mixed Halide Perovskites under Illumination: A Quantitative Comparison of Experimental Observations and Thermodynamic Models

Authors: Klara Suchan, Justus Just, Pascal Becker, Carolin Rehermann, Aboma Merdasa, Roland Mainz, Ivan G. Scheblykin, Eva Unger

Abstract: Photo- and charge-carrier induced ion migration is a major challenge when utilizing metal halide perovskite semiconductors for optoelectronic applications. For mixed iodide/bromide perovskites, the compositional instability due to light- or electrical bias induced phase- segregation restricts the exploitation of the entire bandgap range. Previous experimental and theoretical work suggests that exc… ▽ More Photo- and charge-carrier induced ion migration is a major challenge when utilizing metal halide perovskite semiconductors for optoelectronic applications. For mixed iodide/bromide perovskites, the compositional instability due to light- or electrical bias induced phase- segregation restricts the exploitation of the entire bandgap range. Previous experimental and theoretical work suggests that excited states or charge-carriers trigger the process but the exact mechanism is still under debate. To identify the mechanism and cause of light-induced phase-segregation phenomena we investigate the full compositional range of methylammonium lead bromide/iodide samples, MAPb(Br$_x$I$_{1-x}$)$_3$ with $x = 0\ldots 1$, by simultaneous in-situ X-ray diffraction and photoluminescence spectroscopy during illumination. The quantitative comparison of composition-dependent in-situ XRD and PL shows that at excitation densities of 1 sun, only the initial stage of photo-segregation can be rationalized with the previously established thermodynamic models. However, we observe a progression of the phase-segregation that can only be rationalized by considering long-lived accumulative photo-induced material alterations. We suggest that (additional) photo-induced defects, possibly halide vacancies and interstitials, need to be considered to fully rationalize light-induced phase-segregation and anticipate our findings to provide crucial insight for the development of more sophisticated models. △ Less

Submitted 22 May, 2022; originally announced May 2022.

Understanding the Domain Gap in LiDAR Object Detection Networks

Authors: Jasmine Richter, Florian Faion, Di Feng, Paul Benedikt Becker, Piotr Sielecki, Claudius Glaeser

Abstract: In order to make autonomous driving a reality, artificial neural networks have to work reliably in the open-world. However, the open-world is vast and continuously changing, so it is not technically feasible to collect and annotate training datasets which accurately represent this domain. Therefore, there are always domain gaps between training datasets and the open-world which must be understood.… ▽ More In order to make autonomous driving a reality, artificial neural networks have to work reliably in the open-world. However, the open-world is vast and continuously changing, so it is not technically feasible to collect and annotate training datasets which accurately represent this domain. Therefore, there are always domain gaps between training datasets and the open-world which must be understood. In this work, we investigate the domain gaps between high-resolution and low-resolution LiDAR sensors in object detection networks. Using a unique dataset, which enables us to study sensor resolution domain gaps independent of other effects, we show two distinct domain gaps - an inference domain gap and a training domain gap. The inference domain gap is characterised by a strong dependence on the number of LiDAR points per object, while the training gap shows no such dependence. These fndings show that different approaches are required to close these inference and training domain gaps. △ Less

Submitted 21 April, 2022; originally announced April 2022.

Comments: 14. Uni-DAS e.V. Workshop Fahrerassistenz und automatisiertes Fahren

A Walking Space Robot for On-Orbit Satellite Servicing: The ReCoBot

Authors: Stefan Scherzinger, Jakob Weinland, Robert Wilbrandt, Pascal Becker, Arne Roennau, Rüdiger Dillmann

Abstract: A key factor in the economic efficiency of satellites is their availability in orbit. Replacing standardized building blocks, such as empty fuel tanks or outdated electronic modules, could greatly extend the satellites' lifetime. This, however, requires flexible robots that can locomote on the surface of these satellites for optimal accessibility and manipulation. This paper introduces ReCoBot, a… ▽ More A key factor in the economic efficiency of satellites is their availability in orbit. Replacing standardized building blocks, such as empty fuel tanks or outdated electronic modules, could greatly extend the satellites' lifetime. This, however, requires flexible robots that can locomote on the surface of these satellites for optimal accessibility and manipulation. This paper introduces ReCoBot, a 7-axis walking space manipulator for locomotion and manipulation. The robot can connect to compatible structures with its symmetric ends and provides interfaces for manual teleoperation and motion planning with a constantly changing base and tip. We build on open-source robotics software and easily available components to evaluate the overall concept with an early stage demonstrator. The proposed manipulator has a length of 1.20 m and a weight of 10.4 kg and successfully locomotes over a satellite mockup in our lab environment. △ Less

Submitted 18 March, 2022; originally announced March 2022.

Comments: 7 pages, 9 figures, submitted to the 18th IEEE International Conference on Automation Science and Engineering (CASE)

Motion Macro Programming on Assistive Robotic Manipulators: Three Skill Types for Everyday Tasks

Authors: Stefan Scherzinger, Pascal Becker, Arne Roennau, Rüdiger Dillmann

Abstract: Assistive robotic manipulators are becoming increasingly important for people with disabilities. Teleoperating the manipulator in mundane tasks is part of their daily lives. Instead of steering the robot through all actions, applying self-recorded motion macros could greatly facilitate repetitive tasks. Dynamic Movement Primitives (DMP) are a powerful method for skill learning via teleoperation. F… ▽ More Assistive robotic manipulators are becoming increasingly important for people with disabilities. Teleoperating the manipulator in mundane tasks is part of their daily lives. Instead of steering the robot through all actions, applying self-recorded motion macros could greatly facilitate repetitive tasks. Dynamic Movement Primitives (DMP) are a powerful method for skill learning via teleoperation. For this use case, however, they need simple heuristics to specify where to start, stop, and parameterize a skill without a background in computer science and academic sensor setups for autonomous perception. To achieve this goal, this paper provides the concept of local, global, and hybrid skills that form a modular basis for composing single-handed tasks of daily living. These skills are specified implicitly and can easily be programmed by users themselves, requiring only their basic robotic manipulator. The paper contributes all details for robot-agnostic implementations. Experiments validate the developed methods for exemplary tasks, such as scratching an itchy spot, sorting objects on a desk, and feeding a piggy bank with coins. The paper is accompanied by an open-source implementation at https://github.com/fzi-forschungszentrum-informatik/ArNe △ Less

Submitted 12 May, 2023; v1 submitted 18 February, 2022; originally announced February 2022.

Comments: 8 pages, 10 figures, accepted to the IEEE 20th International Conference on Ubiquitous Robots (UR 2023), Honolulu, USA

Specializing Versatile Skill Libraries using Local Mixture of Experts

Authors: Onur Celik, Dongzhuoran Zhou, Ge Li, Philipp Becker, Gerhard Neumann

Abstract: A long-cherished vision in robotics is to equip robots with skills that match the versatility and precision of humans. For example, when playing table tennis, a robot should be capable of returning the ball in various ways while precisely placing it at the desired location. A common approach to model such versatile behavior is to use a Mixture of Experts (MoE) model, where each expert is a context… ▽ More A long-cherished vision in robotics is to equip robots with skills that match the versatility and precision of humans. For example, when playing table tennis, a robot should be capable of returning the ball in various ways while precisely placing it at the desired location. A common approach to model such versatile behavior is to use a Mixture of Experts (MoE) model, where each expert is a contextual motion primitive. However, learning such MoEs is challenging as most objectives force the model to cover the entire context space, which prevents specialization of the primitives resulting in rather low-quality components. Starting from maximum entropy reinforcement learning (RL), we decompose the objective into optimizing an individual lower bound per mixture component. Further, we introduce a curriculum by allowing the components to focus on a local context region, enabling the model to learn highly accurate skill representations. To this end, we use local context distributions that are adapted jointly with the expert primitives. Our lower bound advocates an iterative addition of new components, where new components will concentrate on local context regions not covered by the current MoE. This local and incremental learning results in a modular MoE model of high accuracy and versatility, where both properties can be scaled by adding more components on the fly. We demonstrate this by an extensive ablation and on two challenging simulated robot skill learning tasks. We compare our achieved performance to LaDiPS and HiREPS, a known hierarchical policy search method for learning diverse skills. △ Less

Submitted 10 January, 2022; v1 submitted 8 December, 2021; originally announced December 2021.

Comments: published at CoRL 2021 London

Switching Recurrent Kalman Networks

Authors: Giao Nguyen-Quynh, Philipp Becker, Chen Qiu, Maja Rudolph, Gerhard Neumann

Abstract: Forecasting driving behavior or other sensor measurements is an essential component of autonomous driving systems. Often real-world multivariate time series data is hard to model because the underlying dynamics are nonlinear and the observations are noisy. In addition, driving data can often be multimodal in distribution, meaning that there are distinct predictions that are likely, but averaging c… ▽ More Forecasting driving behavior or other sensor measurements is an essential component of autonomous driving systems. Often real-world multivariate time series data is hard to model because the underlying dynamics are nonlinear and the observations are noisy. In addition, driving data can often be multimodal in distribution, meaning that there are distinct predictions that are likely, but averaging can hurt model performance. To address this, we propose the Switching Recurrent Kalman Network (SRKN) for efficient inference and prediction on nonlinear and multi-modal time-series data. The model switches among several Kalman filters that model different aspects of the dynamics in a factorized latent state. We empirically test the resulting scalable and interpretable deep state-space model on toy data sets and real driving data from taxis in Porto. In all cases, the model can capture the multimodal nature of the dynamics in the data. △ Less

Submitted 16 November, 2021; originally announced November 2021.

Comments: Machine Learning for Autonomous Driving Workshop at the 35th Conference on Neural Information Processing Systems (NeurIPS 2021), Sydney, Australia

Versatile Inverse Reinforcement Learning via Cumulative Rewards

Authors: Niklas Freymuth, Philipp Becker, Gerhard Neumann

Abstract: Inverse Reinforcement Learning infers a reward function from expert demonstrations, aiming to encode the behavior and intentions of the expert. Current approaches usually do this with generative and uni-modal models, meaning that they encode a single behavior. In the common setting, where there are various solutions to a problem and the experts show versatile behavior this severely limits the gene… ▽ More Inverse Reinforcement Learning infers a reward function from expert demonstrations, aiming to encode the behavior and intentions of the expert. Current approaches usually do this with generative and uni-modal models, meaning that they encode a single behavior. In the common setting, where there are various solutions to a problem and the experts show versatile behavior this severely limits the generalization capabilities of these methods. We propose a novel method for Inverse Reinforcement Learning that overcomes these problems by formulating the recovered reward as a sum of iteratively trained discriminators. We show on simulated tasks that our approach is able to recover general, high-quality reward functions and produces policies of the same quality as behavioral cloning approaches designed for versatile behavior. △ Less

Submitted 15 November, 2021; originally announced November 2021.

Comments: Accepted as a workshop paper in 4th Robot Learning Workshop: Self-Supervised and Lifelong Learning @NeurIPS 2021

ACM Class: I.2.6

Designing Quality Requirements, Metrics and Indicators for Core Ontologies: Results of a Comparative Study for Process Core Ontologies

Authors: Luis Olsina, Maria Fernanda Papa, Guido Tebes, Pablo Becker

Abstract: This preprint specifies quality requirements for a core ontology whose ontological elements such as terms, non-taxonomic relationships, among others, are based on a foundational ontology. The quality requirements are represented in a quality model that is structured in the form of a requirements tree composed of characteristics and attributes to be measured and evaluated. An attribute represents a… ▽ More This preprint specifies quality requirements for a core ontology whose ontological elements such as terms, non-taxonomic relationships, among others, are based on a foundational ontology. The quality requirements are represented in a quality model that is structured in the form of a requirements tree composed of characteristics and attributes to be measured and evaluated. An attribute represents an atomic aspect of an entity, that is, an elementary non-functional requirement that can be measured by a direct or indirect metric and evaluated by an elementary indicator. In contrast, characteristics that model less atomic aspects of an entity cannot be measured by metrics, but rather are evaluated by derived indicators generally modeled by an aggregation function. Therefore, this preprint shows the design of direct and indirect metrics in addition to the design of elementary indicators, which are used to implement measurement and evaluation activities to obtain the results of a quality requirements tree. In particular, this document shows the applicability of the designed metrics and indicators that are used by a evaluation and comparison strategy. Two process core ontologies were preselected, evaluated and compared in order to adopt strengths in the target entity named ProcessCO. The data and information resulting from this study are also recorded, as well as the outcomes of the revaluation after improvement of the target entity. △ Less

Submitted 1 October, 2021; originally announced October 2021.

Comments: 48 pgs

Fitting strategies of accretion column models and application to the broadband spectrum of Cen X-3

Authors: Philipp Thalhammer, Matthias Bissinger, Ralf Ballhausen, Katja Pottschmidt, Michael T. Wolff, Jakob Stierhof, Ekaterina Sokolova-Lapa, Felix Fürst, Christian Malacaria, Amy Gottlieb, Diana M. Marcu-Cheatham, Peter A. Becker, Jörn Wilms

Abstract: Due to the complexity of modeling the radiative transfer inside the accretion columns of neutron star binaries, their X-ray spectra are still commonly described with phenomenological models, for example, a cutoff power law. While the behavior of these models is well understood and they allow for a comparison of different sources and studying source behavior, the extent to which the underlying phys… ▽ More Due to the complexity of modeling the radiative transfer inside the accretion columns of neutron star binaries, their X-ray spectra are still commonly described with phenomenological models, for example, a cutoff power law. While the behavior of these models is well understood and they allow for a comparison of different sources and studying source behavior, the extent to which the underlying physics can be derived from the model parameters is very limited. During recent years, several physically motivated spectral models have been developed to overcome these limitations. Their application, however, is generally computationally much more expensive and they require a high number of parameters which are difficult to constrain. Previous works have presented an analytical solution to the radiative transfer equation inside the accretion column assuming a velocity profile that is linear in the optical depth. An implementation of this solution that is both fast and accurate enough to be fitted to observed spectra is available as a model in XSPEC. The main difficulty of this implementation is that some solutions violate energy conservation and therefore have to be rejected by the user. We propose a novel fitting strategy that ensures energy conservation during the $χ^2$-minimization which simplifies the application of the model considerably. We demonstrate this approach as well a study of possible parameter degeneracies with a comprehensive Markov-chain Monte Carlo analysis of the complete parameter space for a combined NuSTAR and Swift/XRT dataset of Cen X-3. The derived accretion-flow structure features a small column radius of $\sim$63 m and a spectrum dominated by bulk-Comptonization of bremsstrahlung seed photons, in agreement with previous studies. △ Less

Submitted 29 September, 2021; originally announced September 2021.

Comments: 14 pages, 7 figures, A&A accepted

Journal ref: A&A 656, A105 (2021)

ProcessCO v1.3's Terms, Properties, Relationships and Axioms - A Core Ontology for Processes

Authors: Pablo Becker, Luis Olsina

Abstract: The present preprint specifies and defines all Terms, Properties, Relationships and Axioms of ProcessCO (Process Core Ontology). ProcessCO is an ontology devoted mainly for Work Entities and related terms, which is placed at the core level in the context of a multilayer ontological architecture called FCD-OntoArch (Foundational, Core, and Domain Ontological Architecture for Sciences). This is a fi… ▽ More The present preprint specifies and defines all Terms, Properties, Relationships and Axioms of ProcessCO (Process Core Ontology). ProcessCO is an ontology devoted mainly for Work Entities and related terms, which is placed at the core level in the context of a multilayer ontological architecture called FCD-OntoArch (Foundational, Core, and Domain Ontological Architecture for Sciences). This is a five-layered ontological architecture, which considers Foundational, Core, Domain and Instance levels, where the domain level is split down in two sub-levels, namely: Top-domain and Low-domain. Ontologies at the same level can be related to each other, except for the foundational level where only ThingFO (Thing Foundational Ontology) is found. In addition, ontologies' terms and relationships at lower levels can be semantically enriched by ontologies' terms and relationships from the higher levels. Note that both ThingFO and ontologies at the core level such as ProcessCO, SituationCO, among others, are domain independent with respect to their terms. Stereotypes are the mechanism used for enriching ProcessCO terms mainly from the ThingFO ontology. Note that in the end of this document, we address the ProcessCO vs. ThingFO non-taxonomic relationship verification matrix. Additionally, note that annotations of updates from the previous version (ProcessCO v1.2) to the current one (v1.3) can be found in Appendix A. For instance, 6 axioms were added. △ Less

Submitted 5 August, 2021; originally announced August 2021.

Comments: 12 pages

SituationCO v1.2's Terms, Properties, Relationships and Axioms -- A Core Ontology for Particular and Generic Situations

Authors: Luis Olsina, Guido Tebes, Pablo Becker

Abstract: The current preprint is an update to SituationCO v1.1 (Situation Core Ontology), which represents its new version 1.2. It specifies and defines all the terms, properties, relationships and axioms of SituationCO v1.2, being an ontology for particular and generic Situations placed at the core level in the context of a four-layered ontological architecture called FCD-OntoArch (Foundational, Core, and… ▽ More The current preprint is an update to SituationCO v1.1 (Situation Core Ontology), which represents its new version 1.2. It specifies and defines all the terms, properties, relationships and axioms of SituationCO v1.2, being an ontology for particular and generic Situations placed at the core level in the context of a four-layered ontological architecture called FCD-OntoArch (Foundational, Core, and Domain Ontological Architecture for Sciences). This is a four-layered ontological architecture, which considers Foundational, Core, Domain and Instance levels. In turn, the domain level is split down in two sub-levels, namely: Top-domain and Low-domain ontological levels. So in fact, we can consider it to be a five-tier architecture. Ontologies at the same level can be related to each other, except for the foundational level where only ThingFO (Thing Foundational Ontology) is found. In addition, ontologies' terms and relationships at lower levels can be semantically enriched by ontologies' terms and relationships from the higher levels. Note that both ThingFO and ontologies at the core level such as SituationCO, ProcessCO, among others, are domain independent. SituationCO's terms and relationships are specialized primarily from ThingFO. It also completely reuses terms primarily from ProcessCO, ProjectCO and GoalCO ontologies. Stereotypes are the used mechanism for enriching SituationCO terms. Note that in the end of this document, we address the SituationCO vs. ThingFO non-taxonomic relationship verification matrix. △ Less

Submitted 21 July, 2021; originally announced July 2021.

Comments: 9 pgs

Multiple field-induced phases in the frustrated triangular magnet Cs$_3$Fe$_2$Br$_9$

Authors: D. Brüning, T. Fröhlich, D. Gorkov, I. Císařová, Y. Skourski, L. Rossi, B. Bryant, S. Wiedmann, M. Meven, A. Ushakov, S. V. Streltsov, D. Khomskii, P. Becker, L. Bohatý, M. Braden, T. Lorenz

Abstract: The recently discovered material Cs$_3$Fe$_2$Br$_9$ contains Fe$_2$Br$_9$ bi-octahedra forming triangular layers with hexagonal stacking along the $c$ axis. In contrast to isostructural Cr-based compounds, the zero-field ground state is not a nonmagnetic $S=0$ singlet-dimer state. Instead, the Fe$_2$Br$_9$ bi-octahedra host semiclassical $S=5/2$ Fe$^{3+}$ spins with a pronounced easy-axis anisotro… ▽ More The recently discovered material Cs$_3$Fe$_2$Br$_9$ contains Fe$_2$Br$_9$ bi-octahedra forming triangular layers with hexagonal stacking along the $c$ axis. In contrast to isostructural Cr-based compounds, the zero-field ground state is not a nonmagnetic $S=0$ singlet-dimer state. Instead, the Fe$_2$Br$_9$ bi-octahedra host semiclassical $S=5/2$ Fe$^{3+}$ spins with a pronounced easy-axis anisotropy along $c$ and interestingly, the intra-dimer spins are ordered ferromagnetically. The high degree of magnetic frustration due to (various) competing intra- and inter-dimer couplings leads to a surprisingly rich magnetic phase diagram. Already the zero-field ground state is reached via an intermediate phase, and the high-field magnetization and thermal expansion data for $H\parallel c$ identify ten different ordered phases. Among them are phases with constant magnetization of 1/3, respectively 1/2 of the saturation value, and several transitions are strongly hysteretic with pronounced length changes reflecting strong magnetoelastic coupling. △ Less

Submitted 18 August, 2021; v1 submitted 1 June, 2021; originally announced June 2021.

Comments: 11 pages, 12 figures (minor corrections and 3 additional have been included in V3)

Journal ref: Phys. Rev. B 104, 064418 (2021)

Chiral Order and Multiferroic Domain Relaxation in NaFeGe$_2$O$_6$

Authors: S. Biesenkamp, D. Gorkov, W. Schmidt, K. Schmalzl, Y. Sidis, P. Becker, L. Bohatý, M. Braden

Abstract: The magnetic structure and the multiferroic relaxation dynamics of NaFeGe$_2$O$_6$ were studied by neutron scattering on single crystals partially utilizing polarization analysis. In addition to the previously reported transitions, the incommensurate spiral ordering of Fe$^{3+}$ moments in the $ac$ plane develops an additional component along the crystallographic $b$ direction below… ▽ More The magnetic structure and the multiferroic relaxation dynamics of NaFeGe$_2$O$_6$ were studied by neutron scattering on single crystals partially utilizing polarization analysis. In addition to the previously reported transitions, the incommensurate spiral ordering of Fe$^{3+}$ moments in the $ac$ plane develops an additional component along the crystallographic $b$ direction below $T\approx 5\text{ K}$, which coincides with a lock-in of the incommensurate modulation. The quasistatic control of the spin-spiral handedness, respectively of the vector chirality, by external electric fields proves the invertibility of multiferroic domains down to the lowest temperature. Time-resolved measurements of the multiferroic domain inversion in NaFeGe$_2$O$_6$ reveal a simple temperature and electric-field dependence of the multiferroic relaxation that is well described by a combined Arrhenius-Merz relation, as it has been observed for TbMnO$_3$. The maximum speed of domain wall motion is comparable to the spin wave velocity. △ Less

Submitted 14 May, 2021; originally announced May 2021.

TestTDO's v1.3 Terms, Properties, Relationships and Axioms -- A Top-Domain Software Testing Ontology

Authors: Guido Tebes, Denis Peppino, Pablo Becker, Luis Olsina

Abstract: The present preprint specifies and defines all Terms, Properties, Relationships and Axioms of TestTDO (software Testing Top-Domain Ontology) v1.3, which is a slightly updated version of its predecessor, TestTDO v1.2. TestTDO is a top-domain ontology built in the context of a four-layered ontological architecture named FCD-OntoArch (Foundational, Core, and Domain Ontological Architecture for Scienc… ▽ More The present preprint specifies and defines all Terms, Properties, Relationships and Axioms of TestTDO (software Testing Top-Domain Ontology) v1.3, which is a slightly updated version of its predecessor, TestTDO v1.2. TestTDO is a top-domain ontology built in the context of a four-layered ontological architecture named FCD-OntoArch (Foundational, Core, and Domain Ontological Architecture for Sciences). This is a four-layered ontological architecture, which considers Foundational, Core, Domain and Instance levels. In turn, the domain level is split down in two sub-levels, namely: Top-domain and Low-domain ontological levels. Ontologies at the same level can be related to each other, except for the foundational level where only the ThingFO ontology is found. In addition, ontologies' terms and relationships at lower levels can be semantically enriched by ontologies' terms properties and relationships from the higher levels. Some TestTDO's terms and are extended primarily from SituationCO (Situation Core Ontology), and ProcessCO (Process Core Ontology) concepts. Stereotypes are the mechanism used for enriching TestTDO terms. Note that annotations of updates from the previous version (v1.2) to the current one (v1.3) can be found in Appendix A. △ Less

Submitted 10 February, 2022; v1 submitted 7 April, 2021; originally announced April 2021.

Solution of universal nonrelativistic nuclear DFT equations in the Cartesian deformed harmonic-oscillator basis. (IX) HFODD (v3.06h): a new version of the program

Authors: J. Dobaczewski, P. Bączyk, P. Becker, M. Bender, K. Bennaceur, J. Bonnard, Y. Gao, A. Idini, M. Konieczka, M. Kortelainen, L. Próchniak, A. M. Romero, W. Satuła, Y. Shi, T. R. Werner, L. F. Yu

Abstract: We describe the new version (v3.06h) of the code HFODD that solves the universal nonrelativistic nuclear DFT Hartree-Fock or Hartree-Fock-Bogolyubov problem by using the Cartesian deformed harmonic-oscillator basis. In the new version, we implemented the following new features: (i) zero-range three- and four-body central terms, (ii) zero-range three-body gradient terms, (iii) zero-range tensor ter… ▽ More We describe the new version (v3.06h) of the code HFODD that solves the universal nonrelativistic nuclear DFT Hartree-Fock or Hartree-Fock-Bogolyubov problem by using the Cartesian deformed harmonic-oscillator basis. In the new version, we implemented the following new features: (i) zero-range three- and four-body central terms, (ii) zero-range three-body gradient terms, (iii) zero-range tensor terms, (iv) zero-range isospin-breaking terms, (v) finite-range higher-order regularized terms, (vi) finite-range separable terms, (vii) zero-range two-body pairing terms, (viii) multi-quasiparticle blocking, (ix) Pfaffian overlaps, (x) particle-number and parity symmetry restoration, (xi) axialization, (xii) Wigner functions, (xiii) choice of the harmonic-oscillator basis, (xiv) fixed Omega partitions, (xv) consistency formula between energy and fields, and we corrected several errors of the previous versions. △ Less

Submitted 4 August, 2021; v1 submitted 16 April, 2021; originally announced April 2021.

Comments: 52 LaTeX pages, revised version

Journal ref: J. Phys. G: Nucl. Part. Phys. 48 (2021) 102001