Search

Altermagnetism defies conventional classifications of collinear magnetic phases, standing apart from ferromagnetism and antiferromagnetism with its unique combination of spin-dependent symmetries, net-zero magnetization, and anomalous Hall transport. Although altermagnetic states have been realized experimentally, their integration into functional devices has been hindered by the structural rigidity and poor tunability of existing materials. First, through cobalt intercalation of the superconducting 2H-NbSe$_2$ polymorph, we induce and stabilize a robust altermagnetic phase and using both theory and experiment, we directly observe the lifting of Kramers degeneracy. Then, using ultrafast laser pulses, we demonstrate how the low temperature phase of this system can be quenched, realizing the first example of an optical altermagnetic switch. While shedding light on overlooked aspects of altermagnetism, our findings open pathways to spin-based technologies and lay a foundation for advancing the emerging field of altertronics.

Atomically thin layered materials are systems with zero limit bulk-to-surface ratio. Their physical properties are determined by two-dimensionality and strongly affected by interfacing with other systems. Therefore, they represent an accessible platform for the abundance of quantum effects that can be engineered by combining them into vertical stacks. Two types of layered systems are considered here - artificially prepared (exfoliated) van der Waals nanostructures, and naturally layered systems showing quasi 2D behaviour already in a bulk form. A special class of naturally layered materials is misfit structures combining atomic layers of hexagonal transition metal dichalcogenides and slabs of tetragonal ionic rare-earth monochalcogenides in the same superlattice. Both types of layered systems feature a new state of quantum matter, the Ising superconductivity extremely resilient to external magnetic field. A giant electron doping, natural to the misfit structures, can lead to topological superconductivity. Both systems can also be assembled into heterostructures combining different constituents. Layered 2D heterostructures have a large number of implications for many potential applications in solid-state devices., Comment: 2 figures

Learning policies in simulation and transferring them to the real world has become a promising approach in dexterous manipulation. However, bridging the sim-to-real gap for each new task requires substantial human effort, such as careful reward engineering, hyperparameter tuning, and system identification. In this work, we present a system that leverages low-level skills to address these challenges for more complex tasks. Specifically, we introduce a hierarchical policy for in-hand object reorientation based on previously acquired rotation skills. This hierarchical policy learns to select which low-level skill to execute based on feedback from both the environment and the low-level skill policies themselves. Compared to learning from scratch, the hierarchical policy is more robust to out-of-distribution changes and transfers easily from simulation to real-world environments. Additionally, we propose a generalizable object pose estimator that uses proprioceptive information, low-level skill predictions, and control errors as inputs to estimate the object pose over time. We demonstrate that our system can reorient objects, including symmetrical and textureless ones, to a desired pose., Comment: website: https://dexhier.github.io

This article introduces the ManiSkill-ViTac Challenge 2025, which focuses on learning contact-rich manipulation skills using both tactile and visual sensing. Expanding upon the 2024 challenge, ManiSkill-ViTac 2025 includes 3 independent tracks: tactile manipulation, tactile-vision fusion manipulation, and tactile sensor structure design. The challenge aims to push the boundaries of robotic manipulation skills, emphasizing the integration of tactile and visual data to enhance performance in complex, real-world tasks. Participants will be evaluated using standardized metrics across both simulated and real-world environments, spurring innovations in sensor design and significantly advancing the field of vision-tactile fusion in robotics., Comment: Challenge webpage: https://ai-workshops.github.io/maniskill-vitac-challenge-2025/

Touch is a crucial sensing modality that provides rich information about object properties and interactions with the physical environment. Humans and robots both benefit from using touch to perceive and interact with the surrounding environment (Johansson and Flanagan, 2009; Li et al., 2020; Calandra et al., 2017). However, no existing systems provide rich, multi-modal digital touch-sensing capabilities through a hemispherical compliant embodiment. Here, we describe several conceptual and technological innovations to improve the digitization of touch. These advances are embodied in an artificial finger-shaped sensor with advanced sensing capabilities. Significantly, this fingertip contains high-resolution sensors (~8.3 million taxels) that respond to omnidirectional touch, capture multi-modal signals, and use on-device artificial intelligence to process the data in real time. Evaluations show that the artificial fingertip can resolve spatial features as small as 7 um, sense normal and shear forces with a resolution of 1.01 mN and 1.27 mN, respectively, perceive vibrations up to 10 kHz, sense heat, and even sense odor. Furthermore, it embeds an on-device AI neural network accelerator that acts as a peripheral nervous system on a robot and mimics the reflex arc found in humans. These results demonstrate the possibility of digitizing touch with superhuman performance. The implications are profound, and we anticipate potential applications in robotics (industrial, medical, agricultural, and consumer-level), virtual reality and telepresence, prosthetics, and e-commerce. Toward digitizing touch at scale, we open-source a modular platform to facilitate future research on the nature of touch., Comment: 28 pages

While confinement is understood to increase the power and thrust coefficients of cross-flow turbines, how the optimal rotor geometry changes with the blockage ratio -- defined as the ratio between the array projected area and the channel cross-sectional area -- has not been systematically explored. Here, the interplay between rotor geometry and the blockage ratio on turbine performance is investigated experimentally with an array of two identical cross-flow turbines at blockage ratios from 35% to 55%. Three geometric parameters are varied -- the number of blades, the chord-to-radius ratio, and the preset pitch angle -- resulting in 180 unique combinations of rotor geometry and blockage ratio. While the optimal chord-to-radius ratio and preset pitch angle do not depend on the blockage ratio, the optimal blade count increases with the blockage ratio -- an inversion of the relationship between efficiency and blade count typically observed at lower blockage. To explore the combined effects of rotor geometry, rotation rate, and the blockage ratio on array performance, we utilize two bluff-body models: dynamic solidity (which relates thrust to the rotor geometry and kinematics) and Maskell-inspired linear momentum theory (which describes the array-channel interaction as a function of the blockage ratio and thrust). By combining these models, we demonstrate that the array time-average thrust coefficient increases with dynamic solidity in a manner that is self-similar across blockage ratios. Overall, these results highlight key design principles for cross-flow turbines in confined flow and provide insights into the similarities between the dynamics of cross-flow turbines and bluff bodies at high blockage., Comment: SUBMITTED to the Journal of Renewable and Sustainable Energy. To access the supporting dataset prior to publication, please contact Aidan Hunt

Maximally localized Wannier functions use the gauge freedom of Bloch wavefunctions to define the optimally smooth subspace with matrix elements that depend smoothly on crystal momentum. The associated Wannier functions are real-space localized, a feature often used to Fourier interpolate periodic observables in reciprocal space on ultradense momentum grids. However, Fourier interpolation cannot handle non-periodic quantities in reciprocal space, such as the oscillator strength matrix elements, which are crucial for the evaluation of optical properties. We show that a direct multidimensional interpolation in the optimally smooth subspace yields comparable accuracy with respect to Fourier interpolation at a similar or lower computational cost. This approach can also interpolate and extrapolate non-periodic observables, enabling the calculation of optical properties on ultradense momentum grids. Finally, we underline that direct interpolation in the optimally smooth subspace can be employed for periodic and non-periodic tensors of any order without any information on the position of the Wannier centers in real space.

Simulation has enabled unprecedented compute-scalable approaches to robot learning. However, many existing simulation frameworks typically support a narrow range of scenes/tasks and lack features critical for scaling generalizable robotics and sim2real. We introduce and open source ManiSkill3, the fastest state-visual GPU parallelized robotics simulator with contact-rich physics targeting generalizable manipulation. ManiSkill3 supports GPU parallelization of many aspects including simulation+rendering, heterogeneous simulation, pointclouds/voxels visual input, and more. Simulation with rendering on ManiSkill3 can run 10-1000x faster with 2-3x less GPU memory usage than other platforms, achieving up to 30,000+ FPS in benchmarked environments due to minimal python/pytorch overhead in the system, simulation on the GPU, and the use of the SAPIEN parallel rendering system. Tasks that used to take hours to train can now take minutes. We further provide the most comprehensive range of GPU parallelized environments/tasks spanning 12 distinct domains including but not limited to mobile manipulation for tasks such as drawing, humanoids, and dextrous manipulation in realistic scenes designed by artists or real-world digital twins. In addition, millions of demonstration frames are provided from motion planning, RL, and teleoperation. ManiSkill3 also provides a comprehensive set of baselines that span popular RL and learning-from-demonstrations algorithms., Comment: Project website: http://maniskill.ai/

A charge-density-wave (CDW) is characterized by a dynamical order parameter consisting of a time-dependent amplitude and phase, which manifest as optically-active collective modes of the CDW phase. Studying the behaviour of such collective modes in the time-domain, and their coupling with electronic and lattice order, provides important insight into the underlying mechanisms behind CDW formation. In this work, we report on femtosecond broadband transient reflectivity experiments on bulk 1$T$-VSe$_2$ using near-infrared excitation. At low temperature, we observe coherent oscillations associated with the CDW amplitude mode and phonons of the distorted lattice. Across the expected transition temperature at 110 K, we confirm signatures of a rearrangement of the electronic structure evident in the quasiparticle dynamics. However, we find that the amplitude mode instead softens to zero frequency at 80 K, possibly indicating an additional phase transition at this temperature. In addition, we demonstrate photoinduced CDW melting, associated with a collapse of the electronic and lattice order, which occurs at moderate excitation densities, consistent with a dominant electron-phonon CDW mechanism., Comment: 13 pages, 6 figures, supporting information

Screening, a ubiquitous phenomenon associated with the shielding of electric fields by surrounding charges, has been widely adopted as a means to modify a material's properties. While so far most studies have relied on static changes of screening through doping or gating, here we demonstrate that screening can also drive the onset of distinct quantum states on the ultrafast timescale. By using time and angle-resolved photoemission spectroscopy we show that intense optical excitation can drive 1T-TiSe$_2$, a prototypical charge density wave material, almost instantly from a gapped into a semimetallic state. By systematically comparing changes in bandstructure over time and excitation strength with theoretical calculations we find that the appearance of this state is likely caused by a dramatic reduction of the screening length. In summary, this work showcases how optical excitation enables the screening driven design of a non-equilibrium semimetallic phase in TiSe$_2$, possibly providing a general pathway into highly screened phases in other strongly correlated materials.

Misfit layer compounds, heterostructures composed by a regular alternating stacking of rocksalt monochalcogenides bilayers and few-layer transition metal dichalchogenides, are an emergent platform to investigate highly doped transition metal dichalcogenides. Among them, (LaSe)$_{1.14}$(NbSe$_2$)$_2$ displays Ising superconductivity, while the presence of a charge density wave (CDW) in the material is still under debate. Here, by using polarized Raman spectroscopy and first-principles calculations, we show that NbSe$_2$ undergoes a doping-driven collapse of the CDW ordering within the misfit, and no signature of the CDW is detected down to 8~K. We provide a complete experimental and theoretical description of the lattice dynamics of this misfit compound. We show that the vibrational properties are obtained from those of the two subunits, namely the LaSe unit and the NbSe$_2$ bilayer, in the presence of a suitable field-effect doping, and then highlight the 2D nature of the lattice dynamics of NbSe$_2$ within the (LaSe)$_{1.14}$(NbSe$_2$)$_2$ 3D structure., Comment: First version of the manuscript (before revisions)

Photoexcited GeTe undergoes a non-thermal phase transition from a rhombohedral to a rocksalt crystalline phase. The microscopic mechanism and the nature of the transition are unclear. By using constrained density functional perturbation theory and by accounting for quantum anharmonicity within the stochastic self-consistent harmonic approximation, we show that the non-thermal phase transition is strongly first order and does not involve phonon softening, at odd with the thermal one. The transition is driven by the closure of the single particle gap in the photoexcited rhombohedral phase. Finally, we show that ultrafast X-ray diffraction data are consistent with a coexistence of the two phases, as expected in a first order transition. Our results are relevant for the understanding of phase transitions and bonding in phase change materials., Comment: 5 pages, 6 figures To be published in Phys. Rev. Lett

Vision-based tactile sensors have recently become popular due to their combination of low cost, very high spatial resolution, and ease of integration using widely available miniature cameras. The associated field of view and focal length, however, are difficult to package in a human-sized finger. In this paper we employ optical fiber bundles to achieve a form factor that, at 15 mm diameter, is smaller than an average human fingertip. The electronics and camera are also located remotely, further reducing package size. The sensor achieves a spatial resolution of 0.22 mm and a minimum force resolution 5 mN for normal and shear contact forces. With these attributes, the DIGIT Pinki sensor is suitable for applications such as robotic and teleoperated digital palpation. We demonstrate its utility for palpation of the prostate gland and show that it can achieve clinically relevant discrimination of prostate stiffness for phantom and ex vivo tissue., Comment: This work has been submitted to the IEEE for possible publication. The CAD design files of DIGIT Pinki are available at https://github.com/facebookresearch/digit-design

Touch is an important sensing modality for humans, but it has not yet been incorporated into a multimodal generative language model. This is partially due to the difficulty of obtaining natural language labels for tactile data and the complexity of aligning tactile readings with both visual observations and language descriptions. As a step towards bridging that gap, this work introduces a new dataset of 44K in-the-wild vision-touch pairs, with English language labels annotated by humans (10%) and textual pseudo-labels from GPT-4V (90%). We use this dataset to train a vision-language-aligned tactile encoder for open-vocabulary classification and a touch-vision-language (TVL) model for text generation using the trained encoder. Results suggest that by incorporating touch, the TVL model improves (+29% classification accuracy) touch-vision-language alignment over existing models trained on any pair of those modalities. Although only a small fraction of the dataset is human-labeled, the TVL model demonstrates improved visual-tactile understanding over GPT-4V (+12%) and open-source vision-language models (+32%) on a new touch-vision understanding benchmark. Code and data: https://tactile-vlm.github.io.

The photoluminescence spectrum of a single-layer boron nitride remains elusive, marked by enigmatic satellites that hint at a significant but unidentified exciton-phonon coupling.Here, by employing a first principles approach based on the many-body cumulant expansion of the charge response, we calculate the optical absorption and photoluminescence of a single layer boron nitride. We identify the specific exciton-phonon scattering channels and unravel their impact on the optical absorption and photoluminescence spectra thereby providing an interpretation of the experimental features. Finally, we show that, even in a strongly polar material such as h-BN monolayer, the electron-hole interaction responsible for the excitonic effect results in the cancellation of the Fr\"olich interaction at small phonon momenta. This effect is captured only if the invariance of the exciton-phonon matrix elements under unitary transformations in the Bloch functions manifold is preserved in the calculation.

To achieve human-level dexterity, robots must infer spatial awareness from multimodal sensing to reason over contact interactions. During in-hand manipulation of novel objects, such spatial awareness involves estimating the object's pose and shape. The status quo for in-hand perception primarily employs vision, and restricts to tracking a priori known objects. Moreover, visual occlusion of objects in-hand is imminent during manipulation, preventing current systems to push beyond tasks without occlusion. We combine vision and touch sensing on a multi-fingered hand to estimate an object's pose and shape during in-hand manipulation. Our method, NeuralFeels, encodes object geometry by learning a neural field online and jointly tracks it by optimizing a pose graph problem. We study multimodal in-hand perception in simulation and the real-world, interacting with different objects via a proprioception-driven policy. Our experiments show final reconstruction F-scores of $81$% and average pose drifts of $4.7\,\text{mm}$, further reduced to $2.3\,\text{mm}$ with known CAD models. Additionally, we observe that under heavy visual occlusion we can achieve up to $94$% improvements in tracking compared to vision-only methods. Our results demonstrate that touch, at the very least, refines and, at the very best, disambiguates visual estimates during in-hand manipulation. We release our evaluation dataset of 70 experiments, FeelSight, as a step towards benchmarking in this domain. Our neural representation driven by multimodal sensing can serve as a perception backbone towards advancing robot dexterity. Videos can be found on our project website https://suddhu.github.io/neural-feels/, Comment: 43 pages, 20 figures, 1 table; https://suddhu.github.io/neural-feels/

Motivated by growing interest in atomically-thin van der Waals magnetic materials, we present an {\it ab initio} theoretical study of the dependence of their magnetic properties on the electron/hole density $\rho$ induced via the electrical field effect. By focusing on the case of monolayer ${\rm Cr}_2{\rm Ge}_2{\rm Te}_6$ (a prototypical 2D Ising ferromagnet) and employing a hybrid functional, we first study the dependence of the gap and effective mass on the carrier concentration $\rho$. We then investigate the robustness of magnetism by studying the dependencies of the exchange couplings and magneto-crystalline anisotropy energy (MAE) on $\rho$. In agreement with experimental results, we find that magnetism displays a bipolar electrically-tunable character, which is, however, much more robust for hole ($\rho>0$) rather than electron ($\rho<0$) doping. Indeed, the MAE vanishes for an electron density $\rho\approx - 7.5 \times 10^{13}~{\rm e} \times {\rm cm}^{-2}$, signalling the failure of a localized description based on a Heisenberg-type anisotropic spin Hamiltonian. This is in agreement with the rapid increase of the coupling between fourth-neighbor atoms with increasing electron density., Comment: 13 pages, 11 figures

Optical tactile sensors have recently become popular. They provide high spatial resolution, but struggle to offer fine temporal resolutions. To overcome this shortcoming, we study the idea of replacing the RGB camera with an event-based camera and introduce a new event-based optical tactile sensor called Evetac. Along with hardware design, we develop touch processing algorithms to process its measurements online at 1000 Hz. We devise an efficient algorithm to track the elastomer's deformation through the imprinted markers despite the sensor's sparse output. Benchmarking experiments demonstrate Evetac's capabilities of sensing vibrations up to 498 Hz, reconstructing shear forces, and significantly reducing data rates compared to RGB optical tactile sensors. Moreover, Evetac's output and the marker tracking provide meaningful features for learning data-driven slip detection and prediction models. The learned models form the basis for a robust and adaptive closed-loop grasp controller capable of handling a wide range of objects. We believe that fast and efficient event-based tactile sensors like Evetac will be essential for bringing human-like manipulation capabilities to robotics. The sensor design is open-sourced at https://sites.google.com/view/evetac ., Comment: Accepted at IEEE Transactions On Robotics. Project Website: https://sites.google.com/view/evetac

Reinforcement learning from human feedback (RLHF) has emerged as a powerful technique to make large language models (LLMs) more capable in complex settings. RLHF proceeds as collecting human preference data, training a reward model on said data, and optimizing a base ML model with respect to said reward for extrinsic evaluation metrics (e.g. MMLU, GSM8k). RLHF relies on many assumptions about how the various pieces fit together, such as a reward model capturing human preferences and an RL optimizer extracting the right signal from a reward model. As the RLHF process involves many distinct design decisions, it is easy to assume that multiple processes are correlated and therefore numerically linked. This apparent correlation is often not true, where reward models are easily overoptimized or RL optimizers can reduce performance on tasks not modeled in the data. Notable manifestations of models trained with imperfect RLHF systems are those that are prone to refusing basic requests for safety reasons or appearing lazy in generations. As chat model evaluation becomes increasingly nuanced, the reliance on a perceived link between reward model training, RL scores, and downstream performance drives these issues, which we describe as an objective mismatch. In this paper, we illustrate the causes of this issue, reviewing relevant literature from model-based reinforcement learning, and argue for solutions. By solving objective mismatch in RLHF, the ML models of the future will be more precisely aligned to user instructions for both safety and helpfulness., Comment: 11 pages, 5 figures

We theoretically investigate the double-resonance Raman spectrum of monolayer graphene down to infrared laser excitation energies. By using first-principles density functional theory calculations, we improve upon previous theoretical predictions based on conical models or tight-binding approximations, and rigorously justify the evaluation of the electron-phonon enhancement found in Ref. [Venanzi, T., Graziotto, L. et al., Phys. Rev. Lett. 130, 256901 (2023)]. We proceed to discuss the effects of such enhancement on the room temperature graphene resistivity, hinting towards a possible reconciliation of theoretical and experimental discrepancies., Comment: 19 pages, 18 figures

Femtosecond pulses have been used to reveal hidden broken symmetry states and induce transitions to metastable states. However, these states are mostly transient and disappear after laser removal. Photoinduced phase transitions towards crystalline metastable states with a change of topological order are rare and difficult to predict and realize experimentally. Here, by using constrained density functional perturbation theory and accounting for light-induced quantum anharmonicity, we show that ultra-fast lasers can permanently transform the topologically-trivial orthorhombic structure of SnSe into the topological crystalline insulating rocksalt phase via a first-order non-thermal phase transition. We describe the reaction path and evaluate the critical fluence and the possible decay channels after photoexcitation. Our simulations of the photoexcited structural and vibrational properties are in excellent agreement with recent pump-probe data in the intermediate fluence regime below the transition with an error on the curvature of the quantum free energy of the photoexcited state that is smaller than 2%., Comment: 17 pages, 4 figures, The data underlying this study are openly available in Zenodo at https://doi.org/10.5281/zenodo.8413390

Model-based Reinforcement Learning (MBRL) aims to make agents more sample-efficient, adaptive, and explainable by learning an explicit model of the environment. While the capabilities of MBRL agents have significantly improved in recent years, how to best learn the model is still an unresolved question. The majority of MBRL algorithms aim at training the model to make accurate predictions about the environment and subsequently using the model to determine the most rewarding actions. However, recent research has shown that model predictive accuracy is often not correlated with action quality, tracing the root cause to the objective mismatch between accurate dynamics model learning and policy optimization of rewards. A number of interrelated solution categories to the objective mismatch problem have emerged as MBRL continues to mature as a research area. In this work, we provide an in-depth survey of these solution categories and propose a taxonomy to foster future research.

Few-layer graphene possesses low-energy carriers which behave as massive fermions, exhibiting intriguing properties in both transport and light scattering experiments. Lowering the excitation energy of resonance Raman spectroscopy down to 1.17 eV we target these massive quasiparticles in the split bands close to the K point. The low excitation energy weakens some of the Raman processes which are resonant in the visible, and induces a clearer frequency-separation of the sub-structures of the resonance 2D peak in bi- and trilayer samples. We follow the excitation-energy dependence of the intensity of each sub-structure and, comparing experimental measurements on bilayer graphene with ab initio theoretical calculations, we trace back such modifications on the joint effects of probing the electronic dispersion close to the band splitting and enhancement of electron-phonon matrix elements., Comment: 22 pages, 12 figures

We introduce RotateIt, a system that enables fingertip-based object rotation along multiple axes by leveraging multimodal sensory inputs. Our system is trained in simulation, where it has access to ground-truth object shapes and physical properties. Then we distill it to operate on realistic yet noisy simulated visuotactile and proprioceptive sensory inputs. These multimodal inputs are fused via a visuotactile transformer, enabling online inference of object shapes and physical properties during deployment. We show significant performance improvements over prior methods and the importance of visual and tactile sensing., Comment: CoRL 2023; Website: https://haozhi.io/rotateit/

Traffic congestion in dense urban centers presents an economical and environmental burden. In recent years, the availability of vehicle-to-anything communication allows for the transmission of detailed vehicle states to the infrastructure that can be used for intelligent traffic light control. The other way around, the infrastructure can provide vehicles with advice on driving behavior, such as appropriate velocities, which can improve the efficacy of the traffic system. Several research works applied deep reinforcement learning to either traffic light control or vehicle speed advice. In this work, we propose a first attempt to jointly learn the control of both. We show this to improve the efficacy of traffic systems. In our experiments, the joint control approach reduces average vehicle trip delays, w.r.t. controlling only traffic lights, in eight out of eleven benchmark scenarios. Analyzing the qualitative behavior of the vehicle speed advice policy, we observe that this is achieved by smoothing out the velocity profile of vehicles nearby a traffic light. Learning joint control of traffic signaling and speed advice in the real world could help to reduce congestion and mitigate the economical and environmental repercussions of today's traffic systems., Comment: 6 pages, 2 figures, accepted for publication at IEEE ICMLA 2023

In order to characterize in detail the charge density wave (CDW) transition of 1$T$-VSe$_2$, its electronic structure and lattice dynamics are comprehensively studied by means of x-ray diffraction, angle resolved photoemission (ARPES), diffuse and inelastic x-ray scattering (IXS), and state-of-the-art first principles density functional theory calculations. Resonant elastic x-ray scattering (REXS) does not show any resonant enhancement at either V or Se K-edges, indicating that the CDW peak describes a purely structural modulation of the electronic ordering. ARPES identifies (i) a pseudogap at T$>$T$_{CDW}$, which leads to a depletion of the density of states in the $ML-M'L'$ plane at T$<$T$_{CDW}$, and (ii) anomalies in the electronic dispersion reflecting a sizable impact of phonons on it. A diffuse scattering precursor, characteristic of soft phonons, is observed at room temperature (RT) and leads to the full collapse of the low-energy phonon ($\omega_1$) with propagation vector (0.25 0 -0.3) r.l.u. We show that the frequency and linewidth of this mode are anisotropic in momentum space, reflecting the momentum dependence of the electron-phonon interaction (EPI), hence demonstrating that the origin of the CDW is, to a much larger extent, due to the momentum dependence EPI with a small contribution from nesting. The pressure dependence of the $\omega_1$ soft mode remains nearly constant up to 13 GPa at RT, with only a modest softening before the transition to the high-pressure monoclinic $C2/m$ phase. The wide set of experimental data are well captured by our state-of-the art first-principles anharmonic calculations with the inclusion of van der Waals (vdW) corrections in the exchange-correlation functional. The description of the electronics and dynamics of VSe$_2$ reported here adds important pieces of information to the understanding of the electronic modulations of TMDs.

The compound CuTe (vulcanite) undergoes a quasi one dimensional charge density wave (CDW) at $T< T_{\mathrm{CDW}}=335$ K with a $5\times1\times2$ periodicity. The mechanism at its origin is debated. Several theoretical works claimed that semilocal functionals are unable to describe its occurrence and ascribed its formation only to strong electron-electron interaction. Moreover, the possible role of quantum anharmonicity has not been addressed. Here, by performing quantum anharmonic calculations, we show that semilocal functionals correctly describe the occurrence of a CDW in CuTe if ultradense electron momentum grids allowing for small electronic temperatures are used. The distortion is driven by the perfect nesting among 1D Fermi surface sheets extending in the $k_y$ direction. Quantum anharmonic effects are important and tend to suppress both the distortion and $T_{\mathrm{CDW}}$. The quantum anharmonic structural minimization of the CDW phase in the generalized gradient approximation leads, however, to distorted Te-Te bond lengths in the low temperature phase that are $21\%$ of the experimental ones at $T=20$ K. This suggests that, even if the electron-electron interaction is not crucial for the mechanism of CDW formation, it is relevant to accurately describe the structural data for the low-T phase. We assess the effect of correlation on the CDW by using the DFT+U+V approximation with parameters calculated from first principles. We find that correlation enhances the Te-Te distortion, T$_{CDW}$ and the total energy gain by the distortion., Comment: 9 pages, 8 figures, to appear on Phys. Rev. B

Misfit layer compounds are heterostructures composed of rocksalt units stacked with few layers transition metal dichalcogenides. They host Ising superconductivity, charge density waves and good thermoelectricity. The design of misfits emergent properties is, however, hindered by the lack of a global understanding of the electronic transfer among the constituents. Here, by performing first principles calculations, we unveil the mechanism controlling the charge transfer and demonstrate that rocksalt units are always donor and dichalcogenides acceptors. We show that misfits behave as a periodic arrangement of ultra-tunable field effect transistors where a charging as large as 6\times10^{14} e^-cm^{-2} can be reached and controlled efficiently by the La-Pb alloying in the rocksalt. Finally, we identify a strategy to design emergent superconductivity and demonstrate its applicability in (LaSe)_{1.27}(SnSe_2)_2. Our work paves the way to the design synthesis of misfit compounds with tailored physical properties., Comment: First version of the manuscript (before revisions)

EPIq (Electron-Phonon wannier Interpolation over k and q-points) is an open-source software for the calculation of electron-phonon interaction related properties from first principles.Acting as a post-processing tool for a density-functional perturbation theory code ( Quantum ESPRESSO ) and wannier90, EPIq exploits the localization of the deformation potential in the Wannier function basis and the stationary properties of a force-constant functional with respect to the first-order perturbation of the electronic charge density to calculate many electron-phonon related properties with high accuracy and free from convergence issues related to Brillouin zone sampling. EPIq features includes: the adiabatic and non-adiabatic phonon dispersion, superconducting properties (including the superconducting band gap in the Migdal-Eliashberg formulation), double-resonant Raman spectra and lifetime of excited carriers. The possibility to customize most of its input makes EPIq a versatile and interoperable tool. Particularly relevant is the interaction with the Stochastic Self-Consistent Harmonic Approximation (SSCHA) allowing anharmonic effects to be included in the calculation of electron-properties. The scalability offered by the Wannier representation combined with a straightforward workflow and easy-to-read input and output files make EPIq accessible to the wide condensed matter and material science communities.

Low-dimensional materials have remarkable properties that are distinct from their bulk counterparts. A paradigmatic example is Ising superconductivity that occurs in monolayer materials such as NbSe2 which show a strong violation of the Pauli limit. In monolayers, this occurs due to a combination of broken inversion symmetry and spin-orbit coupling that locks the spins of the electrons out-of-plane. Bulk NbSe2 is centrosymmetric and is therefore not an Ising superconductor. We show that bulk misfit compound superconductors, (LaSe)1.14(NbSe2) and (LaSe)1.14(NbSe2)2, comprised of monolayers and bilayers of NbSe2, exhibit unexpected Ising protection with a Pauli-limit violation comparable to monolayer NbSe2, despite formally having inversion symmetry. We study these misfit compounds using complementary experimental methods in combination with first-principles calculations. We propose theoretical mechanisms of how the Ising protection can survive in bulk materials. We show how some of these mechanisms operate in these bulk compounds due to a concerted effect of charge-transfer, defects, reduction of interlayer hopping, and stacking. This highlights how Ising superconductivity can, unexpectedly, arise in bulk materials, and possibly enable the design of bulk superconductors that are resilient to magnetic fields., Comment: 14 pages, 4 figures, 4 extended data items

Interactions mediated by electron-phonon coupling are responsible for important cooperative phenomena in metals such as superconductivity and charge-density waves. The same interaction mechanisms produce strong collision rates in the normal phase of correlated metals, causing sizeable reductions of the dc conductivity and reflectivity. As a consequence, low-energy excitations like phonons, which are crucial for materials characterization, become visible in optical infrared spectra. A quantitative assessment of vibrational resonances requires the evaluation of dynamical Born effective charges, which quantify the coupling between macroscopic electric fields and lattice deformations. We show that the Born effective charges of metals crucially depend on the collision regime of conducting electrons. In particular, we describe, within a first principles framework, the impact of electron scattering on the infrared vibrational resonances, from the undamped, collisionless regime to the overdamped, collision-dominated limit. Our approach enables the interpretation of vibrational reflectance measurements of both super and bad conducting metals, as we illustrate for the case of strongly electron-phonon coupled superhydride H$_3$S.

We report on resonance Raman spectroscopy measurements with excitation photon energy down to 1.16 eV on graphene, to study how low-energy carriers interact with lattice vibrations. Thanks to the excitation energy close to the Dirac point at $\mathbf{K}$, we unveil a giant increase of the intensity ratio between the double-resonant 2D and 2D$^\prime$ peaks with respect to that measured in graphite. Comparing with fully \textit{ab initio} theoretical calculations, we conclude that the observation is explained by an enhanced, momentum-dependent coupling between electrons and Brillouin zone-boundary optical phonons. This finding applies to two dimensional Dirac systems and has important consequences for the modeling of transport in graphene devices operating at room temperature., Comment: 6 pages, 3 figures

Generalized in-hand manipulation has long been an unsolved challenge of robotics. As a small step towards this grand goal, we demonstrate how to design and learn a simple adaptive controller to achieve in-hand object rotation using only fingertips. The controller is trained entirely in simulation on only cylindrical objects, which then - without any fine-tuning - can be directly deployed to a real robot hand to rotate dozens of objects with diverse sizes, shapes, and weights over the z-axis. This is achieved via rapid online adaptation of the controller to the object properties using only proprioception history. Furthermore, natural and stable finger gaits automatically emerge from training the control policy via reinforcement learning. Code and more videos are available at https://haozhi.io/hora, Comment: CoRL 2022. Code and Website: https://haozhi.io/hora

Humans make extensive use of vision and touch as complementary senses, with vision providing global information about the scene and touch measuring local information during manipulation without suffering from occlusions. While prior work demonstrates the efficacy of tactile sensing for precise manipulation of deformables, they typically rely on supervised, human-labeled datasets. We propose Self-Supervised Visuo-Tactile Pretraining (SSVTP), a framework for learning multi-task visuo-tactile representations in a self-supervised manner through cross-modal supervision. We design a mechanism that enables a robot to autonomously collect precisely spatially-aligned visual and tactile image pairs, then train visual and tactile encoders to embed these pairs into a shared latent space using cross-modal contrastive loss. We apply this latent space to downstream perception and control of deformable garments on flat surfaces, and evaluate the flexibility of the learned representations without fine-tuning on 5 tasks: feature classification, contact localization, anomaly detection, feature search from a visual query (e.g., garment feature localization under occlusion), and edge following along cloth edges. The pretrained representations achieve a 73-100% success rate on these 5 tasks., Comment: RSS 2023, site: https://sites.google.com/berkeley.edu/ssvtp

Understanding collective phenomena in quantum materials from first principles is a promising route toward engineering materials properties on demand and designing new functionalities. This work examines the quantum paraelectric state, an elusive state of matter characterized by the smooth saturation of the ferroelectric instability at low temperature due to quantum fluctuations associated with anharmonic phonon effects. The temperature-dependent evolution of the soft ferroelectric phonon mode in the quantum paraelectric KTaO$_3$ in the range 0-300 K is modelled by combining density functional theory (DFT) calculations with the stochastic self-consistent harmonic approximation assisted by an on-the-fly machine-learned force field. The calculated data show that including anharmonic terms is essential to stabilize the spurious imaginary ferroelectric phonon predicted by DFT, in agreement with experiments. Augmenting the DFT workflow with machine-learned force fields allows for efficient stochastic sampling of the configurational space using large supercells in a broad and dense temperature range, inaccessible by conventional ab initio protocols. This work proposes a robust computational workflow capable of accounting for collective behaviors involving different degrees of freedom and occurring at large time/length scales, paving the way for precise modeling and control of quantum effects in materials., Comment: 10 pages, 6 figures. To be published in Advanced Quantum Technologies. Editor: Luigi Ranalli

We show how to properly extract the local charge order in two-dimensional materials from scanning tunneling microscopy/spectroscopy (STM/STS) measurements. When the charge order presents spatial variations at the atomic scale inside the unit cell and is energy dependent, particular care should be taken. In such cases the use of the lock-in technique, while acquiring an STM topography in closed feedback loop, leads to systematically incorrect dI/dV measurements giving a false local charge order. A correct method is either to perform a constant height measurement or to perform a full grid of dI/dV(V) spectroscopies, using a bias voltage setpoint outside the material bandwidth where the local density-of-states (LDOS) is spatially homogeneous. We take as a paradigmatic example of two-dimensional material the 1/3 single-layer Pb/Si(111). As large areas of this phase cannot be grown, charge ordering in this system is not accessible to angular resolved photoemission or grazing x-ray diffraction. Previous investigations by STM/STS supplemented by {\it ab initio} Density Functional Theory (DFT) calculations concluded that this material undergoes a phase transition to a low-temperature $3\times 3$ reconstruction where one Pb atom moves up, the two remaining Pb atoms shifting down. A third STM/STS study by Adler {\it et al.} [PRL 123, 086401 (2019)] came to the opposite conclusion, i.e. that two Pb atoms move up, while one Pb atom shifts down. This latter erroneous conclusion comes from a misuse of the lock-in technique. In contrast, using a full grid of dI/dV(V) spectroscopy measurements, we show that the energy-dependent LDOS maps agree very well with state-of-the-art DFT calculations confirming the one-up two-down charge ordering. This structural and charge re-ordering in the $3\times 3$ unit cell is equally driven by electron-electron interactions and the coupling to the substrate., Comment: 11 pages, 3 figures

Superconductivity occurs in electrochemically doped molybdenum dichalcogenides samples thicker than four layers. While the critical temperature (Tc) strongly depends on the field effect geometry (single or double gate) and on the sample (MoS2 or MoSe2), Tc always saturates at high doping. The pairing mechanism and the complicate dependence of Tc on doping, samples and field-effect geometry are currently not understood. Previous theoretical works assumed homogeneous doping of a single layer and attributed the Tc saturation to a charge density wave instability, however the calculated values of the electron-phonon coupling in the harmonic approximation were one order of magnitude larger than the experimental estimates based on transport data. Here, by performing fully relativistic first principles calculations accounting for the sample thickness, the field-effect geometry and anharmonicity, we rule out the occurrence of charge density waves in the experimental doping range and demonstrate a suppression of one order of magnitude in the electron-phonon coupling, now in excellent agreement with transport data. By solving the anisotropic Migdal-Eliashberg equations, we explain the behaviour of Tc in different systems and geometries. From an analysis of mobility data, we propose that the Tc saturation is due to carriers localization and disorder.

Magnetic layered materials have emerged recently as promising systems to introduce magnetism in structures based on two-dimensional (2D) materials and to investigate exotic magnetic ground states in the 2D limit. In this work, we apply high hydrostatic pressures up to P = 8.7 GPa to the bulk layered antiferromagnet FePS$_3$ to tune the collective lattice excitations (phonons) in resonance with magnetic excitations (magnons). Close to P = 4 GPa, the magnon-phonon resonance is achieved and the strong coupling between these collective modes leads to the formation of new quasi-particles, the magnon-polarons, evidenced in our low temperature Raman scattering experiments by a particular avoided crossing behavior between the phonon and the doubly degenerate antiferromagnetic magnon. At the pressure-induced magnon-phonon resonance, three distinct coupled modes emerge. As it is mainly defined by intralayer properties, we show that the energy of the magnon is nearly pressure independent. We additionally apply high magnetic fields up to B = 30 T to fully identify and characterize the magnon excitations, and to explore the different magnon-polaron regimes for which the phonon has an energy lower-, equal to-, or higher- than the magnon energy. The description of our experimental data requires introducing a phonon-phonon coupling not taken into account in actual calculations., Comment: 11 pages + 6 pages of supplementary materials

The magnetization of a non-magnetic semiconductor by femtosecond light pulses is crucial to achieve an all-optical control of the spin dynamics in materials and to develop faster memory devices. However, the conditions for its detection are largely unknown. In this work we identify the criteria for the observation of ultrafast magnetization and critically discuss the difficulties hindering its experimental detection. We show that ultrafast magnetization of a non magnetic semiconductor can be observed in compounds with very localized conduction band states and more delocalized valence bands, such as in the case of a p-d charge transfer gap. By using constrained and time dependent density functional theory simulations, we demonstrate that a transient ferrimagnetic state can be induced in diamagnetic semiconductor V2O5 via ultrafast pulses at realistic fluences. The ferrimagnetic state has opposite magnetic moments on vanadium (conduction) and oxygen (valence) states. Our methodology outruns the case of V2O5 as it identifies the key requirements for a computational screening of ultrafast magnetism in non-magnetic semiconductors., Comment: https://journals.aps.org/prb/accepted/d907dYc6Zfb16a70e88534261f79f72c0bb0e0d75

Controllable topological phase transitions are appealing as they allow for tunable single particle electronic properties. Here, by using state-of-the-art manybody perturbation theory techniques, we show that the topological $Z_2$ phase transition occurring in the single particle spectrum of the recently synthetized ethynylene bridged polyacene polymers is accompanied by a topological excitonic phase transition: the band inversion in the non-trivial phase yields real-space exciton wave functions in which electrons and holes exchange orbital characters with respect to the trivial phase. The topological excitonic phase transition results in a broad tunability of the singlet-triplet splittings, opening appealing perspectives for the occurrence of singlet fission. Finally, the flatness of the single-particle electronic structure in the topological non trivial phase leads to negatively dispersing triplet excitons in a large portion of the Brillouin zone, opening a route for spontaneously coherent energy transport at room temperature.

Accurately predicting the consequences of agents' actions is a key prerequisite for planning in robotic control. Model-based reinforcement learning (MBRL) is one paradigm which relies on the iterative learning and prediction of state-action transitions to solve a task. Deep MBRL has become a popular candidate, using a neural network to learn a dynamics model that predicts with each pass from high-dimensional states to actions. These "one-step" predictions are known to become inaccurate over longer horizons of composed prediction - called the compounding error problem. Given the prevalence of the compounding error problem in MBRL and related fields of data-driven control, we set out to understand the properties of and conditions causing these long-horizon errors. In this paper, we explore the effects of subcomponents of a control problem on long term prediction error: including choosing a system, collecting data, and training a model. These detailed quantitative studies on simulated and real-world data show that the underlying dynamics of a system are the strongest factor determining the shape and magnitude of prediction error. Given a clearer understanding of compounding prediction error, researchers can implement new types of models beyond "one-step" that are more useful for control., Comment: 25 pages, 19 figures

The combination of Reinforcement Learning (RL) with deep learning has led to a series of impressive feats, with many believing (deep) RL provides a path towards generally capable agents. However, the success of RL agents is often highly sensitive to design choices in the training process, which may require tedious and error-prone manual tuning. This makes it challenging to use RL for new problems, while also limits its full potential. In many other areas of machine learning, AutoML has shown it is possible to automate such design choices and has also yielded promising initial results when applied to RL. However, Automated Reinforcement Learning (AutoRL) involves not only standard applications of AutoML but also includes additional challenges unique to RL, that naturally produce a different set of methods. As such, AutoRL has been emerging as an important area of research in RL, providing promise in a variety of applications from RNA design to playing games such as Go. Given the diversity of methods and environments considered in RL, much of the research has been conducted in distinct subfields, ranging from meta-learning to evolution. In this survey we seek to unify the field of AutoRL, we provide a common taxonomy, discuss each area in detail and pose open problems which would be of interest to researchers going forward., Comment: Published in JAIR. Co-first authors and co-last authors are listed in alphabetical order

Despite being the oldest known superconductor, solid mercury is mysteriously absent from all current computational databases of superconductors. In this work, we present a critical study of its superconducting properties based on state-of-the-art superconducting density-functional theory. Our calculations reveal numerous anomalies in electronic and lattice properties, which can mostly be handled, with due care, by modern ab-initio techniques. In particular, we highlight an anomalous role of (i) electron-electron correlations on structural properties (ii) spin-orbit coupling on the dynamical stability, and (iii) semicore $d$ levels on the effective Coulomb interaction and, ultimately, the critical temperature.

By using constrained density functional theory modeling, we demonstrate that ultrafast optical pumping unveils hidden charge orders in group VI monolayer transition metal ditellurides. We show that irradiation of the insulating 2H phases stabilizes multiple transient charge density wave orders with light-tunable distortion, periodicity, electronic structure and bandgap. Moreover, optical pumping of the semimetallic 1T' phases generates a transient charge ordered metallic phase composed of 2D diamond clusters. For each transient phase we identify the critical fluence at which it is observed and the specific optical and Raman fingerprints to directly compare with future ultrafast pump-probe experiments. Our work demonstrates that it is possible to stabilize charge density waves even in insulating 2D transition metal dichalcogenides by ultrafast irradiation., Comment: Accepted for publication in Physical Review Letters

The co-adaptation of robot morphology and behaviour becomes increasingly important with the advent of fast 3D-manufacturing methods and efficient deep reinforcement learning algorithms. A major challenge for the application of co-adaptation methods to the real world is the simulation-to-reality-gap due to model and simulation inaccuracies. However, prior work focuses primarily on the study of evolutionary adaptation of morphologies exploiting analytical models and (differentiable) simulators with large population sizes, neglecting the existence of the simulation-to-reality-gap and the cost of manufacturing cycles in the real world. This paper presents a new approach combining classic high-frequency deep neural networks with computational expensive Graph Neural Networks for the data-efficient co-adaptation of agents with varying numbers of degrees-of-freedom. Evaluations in simulation show that the new method can co-adapt agents within such a limited number of production cycles by efficiently combining design optimization with offline reinforcement learning, that it allows for the direct application to real-world co-adaptation tasks in future work

Humans build 3D understandings of the world through active object exploration, using jointly their senses of vision and touch. However, in 3D shape reconstruction, most recent progress has relied on static datasets of limited sensory data such as RGB images, depth maps or haptic readings, leaving the active exploration of the shape largely unexplored. Inactive touch sensing for 3D reconstruction, the goal is to actively select the tactile readings that maximize the improvement in shape reconstruction accuracy. However, the development of deep learning-based active touch models is largely limited by the lack of frameworks for shape exploration. In this paper, we focus on this problem and introduce a system composed of: 1) a haptic simulator leveraging high spatial resolution vision-based tactile sensors for active touching of 3D objects; 2)a mesh-based 3D shape reconstruction model that relies on tactile or visuotactile signals; and 3) a set of data-driven solutions with either tactile or visuotactile priors to guide the shape exploration. Our framework enables the development of the first fully data-driven solutions to active touch on top of learned models for object understanding. Our experiments show the benefits of such solutions in the task of 3D shape understanding where our models consistently outperform natural baselines. We provide our framework as a tool to foster future research in this direction.

This paper discusses the perspective of the H2020 TEACHING project on the next generation of autonomous applications running in a distributed and highly heterogeneous environment comprising both virtual and physical resources spanning the edge-cloud continuum. TEACHING puts forward a human-centred vision leveraging the physiological, emotional, and cognitive state of the users as a driver for the adaptation and optimization of the autonomous applications. It does so by building a distributed, embedded and federated learning system complemented by methods and tools to enforce its dependability, security and privacy preservation. The paper discusses the main concepts of the TEACHING approach and singles out the main AI-related research challenges associated with it. Further, we provide a discussion of the design choices for the TEACHING system to tackle the aforementioned challenges

Single layer Pb on top of (111) surfaces of group IV semiconductors hosts charge density wave and superconductivity depending on the coverage and on the substrate. These systems are normally considered to be experimental realizations of single band Hubbard models and their properties are mostly investigated using lattice models with frozen structural degrees of freedom, although the reliability of this approximation is unclear. Here, we consider the case of Pb/Ge(111) at 1/3 coverage, for which surface X-ray diffraction and ARPES data are available. By performing first principles calculations, we demonstrate that the non-local exchange between Pb and the substrate drives the system into a $3\times 3$ charge density wave. The electronic structure of this charge ordered phase is mainly determined by two effects: the magnitude of the Pb distortion and the large spin-orbit coupling. Finally, we show that the effect applies also to the $3\times 3$ phase of Pb/Si(111) where the Pb-substrate exchange interaction increases the bandwidth by more than a factor 1.5 with respect to DFT+U, in better agreement with STS data. The delicate interplay between substrate, structural and electronic degrees of freedom invalidates the widespread interpretation available in literature considering these compounds as physical realizations of single band Hubbard models., Comment: To appear in Phys. Rev. B

We present a constrained density functional perturbation theory scheme for the calculation of structural and harmonic vibrational properties of insulators in the presence of an excited and thermalized electron-hole plasma. The method is ideal to tame ultrafast light induced structural transitions in the regime where the photocarriers thermalize faster than the lattice, the electron-hole recombination time is longer than the phonon period and the photocarrier concentration is large enough to be approximated by an electron-hole plasma. The complete derivation presented here includes total energy, forces and stress tensor, variable cell structural optimization, harmonic vibrational properties and the electron-phonon interaction. We discuss in detail the case of zone center optical phonons not conserving the number of electrons and inducing a Fermi shift in the photo-electron and hole distributions. We validate our implementation by comparing with finite differences in Te and VSe2. By calculating the evolution of the phonon spectrum of Te, Si and GaAs as a function of the fluence of the incoming laser light, we demonstrate that even at low fluences, corresponding to approximately 0.1 photocarriers per cell, the phonon spectrum is substantially modified with respect to the ground state one with new Kohn anomalies appearing and a substantial softening of zone center optical phonons. Our implementation can be efficiently used to detect reversible transient phases and irreversible structural transition induced by ultrafast light absorption.

The virtuoso plays the piano with passion, poetry and extraordinary technical ability. As Liszt said (a virtuoso)must call up scent and blossom, and breathe the breath of life. The strongest robots that can play a piano are based on a combination of specialized robot hands/piano and hardcoded planning algorithms. In contrast to that, in this paper, we demonstrate how an agent can learn directly from machine-readable music score to play the piano with dexterous hands on a simulated piano using reinforcement learning (RL) from scratch. We demonstrate the RL agents can not only find the correct key position but also deal with various rhythmic, volume and fingering, requirements. We achieve this by using a touch-augmented reward and a novel curriculum of tasks. We conclude by carefully studying the important aspects to enable such learning algorithms and that can potentially shed light on future research in this direction.