Search

The precipitous rise of consumer network applications reiterates the urgency to redefine computing hardware with low power footprint. Neuromorphic computing utilizing correlated oxides offers an energy-efficient solution. By designing anisotropic functional properties in LSMO on a twinned LAO substrate and driving it out of thermodynamic equilibrium, we demonstrate two distinct negative differential resistance states in such volatile memristors. These were harnessed to exhibit oscillatory dynamics in LSMO at different frequencies and an artificial neuron with leaky integrate-and-fire dynamics. A material based modelling incorporating bond angle distortions in neighboring perovskites and capturing the inhomogeneity of domain distribution and propagation explains both the NDR regimes. Our findings establish LSMO as an important material for neuromorphic computing hardware.

The modulation of channel conductance in field-effect transistors (FETs) via metal-oxide-semiconductor (MOS) structures has revolutionized information processing and storage. However, the limitations of silicon-based FETs in electrical switching have driven the search for new materials capable of overcoming these constraints. Electrostatic gating of competing electronic phases in a Mott material near its metal to insulator transition (MIT) offers prospects of substantial modulation of the free carriers and electrical resistivity through small changes in band filling. While electrostatic control of the MIT has been previously reported, the advancement of Mott materials towards novel Mott transistors requires the realization of their charge gain prospects in a solid-state device. In this study, we present gate-control of electron correlation using a solid-state device utilizing the oxide Mott system $La_{1-x}Sr_xVO_3$ as a correlated FET channel. We report on a gate resistance response that cannot be explained in a purely electrostatic framework, suggesting at least $\times100$ charge gain originating from the correlated behavior. These preliminary results pave the way towards the development of highly efficient, low-power electronic devices that could surpass the performance bottlenecks of conventional FETs by leveraging the electronic phase transitions of correlated electron systems.

We give further evidence that the matrix-tensor model studied in \cite{belin2023} is dual to AdS$_{3}$ gravity including the sum over topologies. This provides a 3D version of the duality between JT gravity and an ensemble of random Hamiltonians, in which the matrix and tensor provide random CFT$_2$ data subject to a potential that incorporates the bootstrap constraints. We show how the Feynman rules of the ensemble produce a sum over all three-manifolds and how surgery is implemented by the matrix integral. The partition functions of the resulting 3d gravity theory agree with Virasoro TQFT (VTQFT) on a fixed, hyperbolic manifold. However, on non-hyperbolic geometries, our 3d gravity theory differs from VTQFT, leading to a difference in the eigenvalue statistics of the associated ensemble. As explained in \cite{belin2023}, the Schwinger-Dyson (SD) equations of the matrix-tensor integral play a crucial role in understanding how gravity emerges in the limit that the ensemble localizes to exact CFT's. We show how the SD equations can be translated into a combinatorial problem about three-manifolds., Comment: Equations 1.1 and 2.1 corrected. Section 3.1 revised. Typos fixed and references added. 75 pages

Machine learning approaches to Structure-Based Drug Design (SBDD) have proven quite fertile over the last few years. In particular, diffusion-based approaches to SBDD have shown great promise. We present a technique which expands on this diffusion approach in two crucial ways. First, we address the size disparity between the drug molecule and the target/receptor, which makes learning more challenging and inference slower. We do so through the notion of a Virtual Receptor, which is a compressed version of the receptor; it is learned so as to preserve key aspects of the structural information of the original receptor, while respecting the relevant group equivariance. Second, we incorporate a protein language embedding used originally in the context of protein folding. We experimentally demonstrate the contributions of both the virtual receptors and the protein embeddings: in practice, they lead to both better performance, as well as significantly faster computations.

A central problem in quantum mechanics involves solving the Electronic Schrodinger Equation for a molecule or material. The Variational Monte Carlo approach to this problem approximates a particular variational objective via sampling, and then optimizes this approximated objective over a chosen parameterized family of wavefunctions, known as the ansatz. Recently neural networks have been used as the ansatz, with accompanying success. However, sampling from such wavefunctions has required the use of a Markov Chain Monte Carlo approach, which is inherently inefficient. In this work, we propose a solution to this problem via an ansatz which is cheap to sample from, yet satisfies the requisite quantum mechanical properties. We prove that a normalizing flow using the following two essential ingredients satisfies our requirements: (a) a base distribution which is constructed from Determinantal Point Processes; (b) flow layers which are equivariant to a particular subgroup of the permutation group. We then show how to construct both continuous and discrete normalizing flows which satisfy the requisite equivariance. We further demonstrate the manner in which the non-smooth nature ("cusps") of the wavefunction may be captured, and how the framework may be generalized to provide induction across multiple molecules. The resulting theoretical framework entails an efficient approach to solving the Electronic Schrodinger Equation., Comment: AAAI 2025 Camera Ready Version

Graph generation is a fundamental problem in various domains, including chemistry and social networks. Recent work has shown that molecular graph generation using recurrent neural networks (RNNs) is advantageous compared to traditional generative approaches which require converting continuous latent representations into graphs. One issue which arises when treating graph generation as sequential generation is the arbitrary order of the sequence which results from a particular choice of graph flattening method. In this work we propose using RNNs, taking into account the non-sequential nature of graphs by adding an Orderless Regularization (OLR) term that encourages the hidden state of the recurrent model to be invariant to different valid orderings present under the training distribution. We demonstrate that sequential graph generation models benefit from our proposed regularization scheme, especially when data is scarce. Our findings contribute to the growing body of research on graph generation and provide a valuable tool for various applications requiring the synthesis of realistic and diverse graph structures., Comment: 16 pages, 3 figures

We study the effect of coupling an electronic reservoir to a Hubbard model and to a Dimer Hubbard Model. This is motivated by recent experiments on the effect of illumination on the insulator-metal transition in a vanadium oxides and photo-conductive cadmium sulfide heterostructure. We model the system as an electronic reservoir hybridized to the correlated system. We assume that the light intensity controls the hybridization coupling strength. We find that the light intensity acts similarly as the temperature in the weak interaction regime. This is consistent with the role played by electronic reservoirs in out-of-equilibrium systems. In contrast, qualitative differences appear at strong coupling. We show that modeling the V$_2$O$_3$ compound with a Hubbard model, our results describe qualitatively well the observed illumination-driven suppression of the insulator-metal transition. In contrast, in the DHM results fail to capture the mild suppression observed in the case of VO$_2$. This indicates that the lattice may play an important role in this case.

This report introduces LLaMandement, a state-of-the-art Large Language Model, fine-tuned by the French government and designed to enhance the efficiency and efficacy of processing parliamentary sessions (including the production of bench memoranda and documents required for interministerial meetings) by generating neutral summaries of legislative proposals. Addressing the administrative challenges of manually processing a growing volume of legislative amendments, LLaMandement stands as a significant legal technological milestone, providing a solution that exceeds the scalability of traditional human efforts while matching the robustness of a specialized legal drafter. We release all our fine-tuned models and training data to the community., Comment: 21 pages, 9 figures

There is growing interest in strongly correlated insulator thin films because the intricate interplay of their intrinsic and extrinsic state variables causes memristive behavior that might be used for bio-mimetic devices in the emerging field of neuromorphic computing. In this study we find that laser irradiation tends to drive V$_2$O$_3$ from supercooled/superheated metastable states towards thermodynamic equilibrium, most likely in a non-thermal way. We study thin films of the prototypical Mott-insulator V$_2$O$_3$, which show spontaneous phase separation into metal-insulator herringbone domains during the Mott transition. Here, we use low-temperature microscopy to investigate how these metal-insulator domains can be modified by scanning a focused laser beam across the thin film surface. We find that the response depends on the thermal history: When the thin film is heated from below the Mott transition temperature, the laser beam predominantly induces metallic domains. On the contrary, when the thin film is cooled from a temperature above the transition, the laser beam predominantly induces insulating domains. Very likely, the V$_2$O$_3$ thin film is in a superheated or supercooled state, respectively, during the first-order phase transition, and the perturbation by a laser beam drives these metastable states into stable ones. This way, the thermal history is locally erased. Our findings are supported by a phenomenological model with a laser-induced lowering of the energy barrier between the metastable and equilibrium states., Comment: 14 pages, 10 figures

The integrated Berry curvature is a geometric property that has dramatic implications for material properties. This study investigates the integrated Berry curvature and other contributions to the anomalous Hall effect in CrGeTe$_3$ as a function of pressure. The anomalous Hall effect is absent in the insulating phase of CrGeTe$_3$ and evolves with pressure in a dome-like fashion as pressure is applied. The dome's edges are characterized by Fermi surface deformations, manifested as mixed electron and hole transport. We corroborate the presence of bipolar transport by ab-initio calculations which also predict a nonmonotonic behavior of the Berry curvature as a function of pressure. Quantitative discrepancies between our calculations and experimental results indicate that additional scattering mechanisms, which are also strongly tuned by pressure, contribute to the anomalous Hall effect in CrGeTe$_3$.

Phase separation naturally occurs in a variety of magnetic materials and it often has a major impact on both electric and magnetotransport properties. In resistive switching systems, phase separation can be created on demand by inducing local switching, which provides an opportunity to tune the electronic and magnetic state of the device by applying voltage. Here we explore the magnetotransport properties in the ferromagnetic oxide (La,Sr)MnO3 (LSMO) during the electrical triggering of an intrinsic metal-insulator transition (MIT) that produces volatile resistive switching. This switching occurs in a characteristic spatial pattern, i.e., the formation of an insulating barrier perpendicular to the current flow, enabling an electrically actuated ferromagnetic-paramagnetic-ferromagnetic phase separation. At the threshold voltage of the MIT triggering, both anisotropic and colossal magnetoresistances exhibit anomalies including a large increase in magnitude and a sign flip. Computational analysis revealed that these anomalies originate from the coupling between the switching-induced phase separation state and the intrinsic magnetoresistance of LSMO. This work demonstrates that driving the MIT material into an out-of-equilibrium resistive switching state provides the means to electrically control of the magnetotransport phenomena.

We introduce AudioPaLM, a large language model for speech understanding and generation. AudioPaLM fuses text-based and speech-based language models, PaLM-2 [Anil et al., 2023] and AudioLM [Borsos et al., 2022], into a unified multimodal architecture that can process and generate text and speech with applications including speech recognition and speech-to-speech translation. AudioPaLM inherits the capability to preserve paralinguistic information such as speaker identity and intonation from AudioLM and the linguistic knowledge present only in text large language models such as PaLM-2. We demonstrate that initializing AudioPaLM with the weights of a text-only large language model improves speech processing, successfully leveraging the larger quantity of text training data used in pretraining to assist with the speech tasks. The resulting model significantly outperforms existing systems for speech translation tasks and has the ability to perform zero-shot speech-to-text translation for many languages for which input/target language combinations were not seen in training. AudioPaLM also demonstrates features of audio language models, such as transferring a voice across languages based on a short spoken prompt. We release examples of our method at https://google-research.github.io/seanet/audiopalm/examples, Comment: Technical report

Background: Injection drug use (IDU) is a dangerous health behavior that increases mortality and morbidity. Identifying IDU early and initiating harm reduction interventions can benefit individuals at risk. However, extracting IDU behaviors from patients' electronic health records (EHR) is difficult because there is no International Classification of Disease (ICD) code and the only place IDU information can be indicated is unstructured free-text clinical notes. Although natural language processing can efficiently extract this information from unstructured data, there are no validated tools. Methods: To address this gap in clinical information, we design and demonstrate a question-answering (QA) framework to extract information on IDU from clinical notes. Our framework involves two main steps: (1) generating a gold-standard QA dataset and (2) developing and testing the QA model. We utilize 2323 clinical notes of 1145 patients sourced from the VA Corporate Data Warehouse to construct the gold-standard dataset for developing and evaluating the QA model. We also demonstrate the QA model's ability to extract IDU-related information on temporally out-of-distribution data. Results: Here we show that for a strict match between gold-standard and predicted answers, the QA model achieves 51.65% F1 score. For a relaxed match between the gold-standard and predicted answers, the QA model obtains 78.03% F1 score, along with 85.38% Precision and 79.02% Recall scores. Moreover, the QA model demonstrates consistent performance when subjected to temporally out-of-distribution data. Conclusions: Our study introduces a QA framework designed to extract IDU information from clinical notes, aiming to enhance the accurate and efficient detection of people who inject drugs, extract relevant information, and ultimately facilitate informed patient care., Comment: 31 pages, 11 tables, 7 figures

We study the problem of learning conditional distributions of the form $p(G | \hat G)$, where $G$ and $\hat G$ are two 3D graphs, using continuous normalizing flows. We derive a semi-equivariance condition on the flow which ensures that conditional invariance to rigid motions holds. We demonstrate the effectiveness of the technique in the molecular setting of receptor-aware ligand generation., Comment: ICLR Physics for Machine Learning (Physics4ML) Workshop 2023. arXiv admin note: substantial text overlap with arXiv:2211.04754

We propose a novel, physically-constrained and differentiable approach for the generation of D-dimensional qudit states via spontaneous parametric down-conversion (SPDC) in quantum optics. We circumvent any limitations imposed by the inherently stochastic nature of the physical process and incorporate a set of stochastic dynamical equations governing its evolution under the SPDC Hamiltonian. We demonstrate the effectiveness of our model through the design of structured nonlinear photonic crystals (NLPCs) and shaped pump beams; and show, theoretically and experimentally, how to generate maximally entangled states in the spatial degree of freedom. The learning of NLPC structures offers a promising new avenue for shaping and controlling arbitrary quantum states and enables all-optical coherent control of the generated states. We believe that this approach can readily be extended from bulky crystals to thin Metasurfaces and potentially applied to other quantum systems sharing a similar Hamiltonian structures, such as superfluids and superconductors., Comment: ICLR Machine Learning for Materials (ML4Materials) Workshop 2023. arXiv admin note: text overlap with arXiv:2112.05934

The strongly correlated material La0.7Sr0.3MnO3 (LSMO) exhibits metal-to-insulator and magnetic transition near room temperature. Although the physical properties of LSMO can be manipulated by strain, chemical doping, temperature, or magnetic field, they often require large external stimuli. To include additional flexibility and tunability, we developed a hybrid optoelectronic heterostructure that uses photocarrier injection from cadmium sulfide (CdS) to an LSMO layer to change its electrical conductivity. LSMO exhibits no significant optical response, however, the CdS/LSMO heterostructures show an enhanced conductivity, with ~ 37 % resistance drop, at the transition temperature under light stimuli. This enhanced conductivity in response to light is comparable to the effect of a 9 T magnetic field in pure LSMO. Surprisingly, the optical and magnetic responses of CdS/LSMO heterostructures are decoupled and exhibit different effects when both stimuli are applied. This unexpected behavior shows that heterostructuring strongly correlated oxides may require a new understanding of the coupling of physical properties across the transitions and provide the means to implement new functionalities.

We aim at building a bridge between to {\it a priori} disconnected fields: Neuroscience and Material Science. We construct an analogy based on identifying spikes events in time with the positions of particles of matter. We show that one may think of the dynamical states of spiking neurons and spiking neural networks as {\it time-matter}. Namely, a structure of spike-events in time having analogue properties to that of ordinary matter. We can define for neural systems notions equivalent to the equations of state, phase diagrams and their phase transitions. For instance, the familiar Ideal Gas Law relation (P$v$ = constant) emerges as analogue of the Ideal Integrate and Fire neuron model relation ($I_{in}$ISI = constant). We define the neural analogue of the spatial structure correlation function, that can characterize spiking states with temporal long-range order, such as regular tonic spiking. We also define the ``neuro-compressibility'' response function in analogy to the lattice compressibility. We show that similarly to the case of ordinary matter, the anomalous behavior of the neuro-compressibility is a precursor effect that signals the onset of changes in spiking states. We propose that the notion of neuro-compressibility may open the way to develop novel medical tools for the early diagnose of diseases. It may allow to predict impending anomalous neural states, such as Parkinson's tremors, epileptic seizures, electric cardiopathies, and perhaps may even serve as a predictor of the likelihood of regaining consciousness., Comment: 8 pages, 8 figures

In Mott materials strong electron correlation yields a spectrum of complex electronic structures. Recent synthesis advancements open realistic opportunities for harnessing Mott physics to design transformative devices. However, a major bottleneck in realizing such devices remains the lack of control over the electron correlation strength. This stems from the complexity of the electronic structure, which often veils the basic mechanisms underlying the correlation strength. Here, we present control of the correlation strength by tuning the degree of orbital overlap using picometer-scale lattice engineering. We illustrate how bandwidth control and concurrent symmetry breaking can govern the electronic structure of a correlated $SrVO_3$ model system. We show how tensile and compressive biaxial strain oppositely affect the $SrVO_3$ in-plane and out-of-plane orbital occupancy, resulting in the partial alleviation of the orbital degeneracy. We derive and explain the spectral weight redistribution under strain and illustrate how high tensile strain drives the system towards a Mott insulating state. Implementation of such concepts will drive correlated electron phenomena closer towards new solid state devices and circuits. These findings therefore pave the way for understanding and controlling electron correlation in a broad range of functional materials, driving this powerful resource for novel electronics closer towards practical realization.

Gold is one of the most inert metals, forming very few compounds, some with rather interesting properties, and only two of them currently known to be superconducting under certain conditions (AuTe$_2$ and SrAuSi$_3$). Compounds of another noble element, Ag, are also relatively rare, and very few of them are superconducting. Finding new superconducting materials containing gold (and silver) is a challenge - especially having in mind that the best high-$T_c$ superconductors at normal conditions are based upon their rather close ''relative'', Cu. Here we report combined X-ray diffraction, Raman, and resistivity measurements, as well as first-principles calculations, to explore the effect of hydrostatic pressure on the properties of the sylvanite mineral, AuAgTe$_4$. Our experimental results, supported by density functional theory, reveal a structural phase transition at $\sim$5 GPa from a monoclinic $P2/c$ to $P2/m$ phase, resulting in almost identical coordinations of Au and Ag ions, with rather uniform interatomic distances. Further, resistivity measurements show the onset of superconductivity at $\sim$1.5 GPa in the $P2/c$ phase, followed by a linear increase of $T_c$ up to the phase transition, with a maximum in the $P2/m$ phase, and a gradual decrease afterwards. Our calculations indicate phonon-mediated superconductivity, with the electron-phonon coupling coming predominantly from the low-energy phonon modes. Thus, along with the discovery of a new superconducting compound of gold/silver, our results advance understanding of the mechanism of the superconductivity in Au-containing compounds, which may pave the way to the discovery of novel ones.

Sr2IrO4 and Sr3Ir2O7 form two families of spin-orbit Mott insulators with quite different charge gaps and an antiferromagnetic (AF) ground state. This offers a unique opportunity to study the impact of long-range magnetic order in Mott insulators. It appears to play a different role in the two families, as there is almost no change of the resistivity at the magnetic transition TN in Sr2IrO4 and a large one in Sr3Ir2O7. We use angle-resolved photoemission to study the evolution of the spectral lineshape through the magnetic transition. We use Ru and La substitutions to tune TN and discriminate changes due to temperature from those due to magnetic order. We evidence a shift and a transfer of spectral weight in the gap at TN in Sr3Ir2O7, which is absent in Sr2IrO4. We assign this behavior to a significantly larger coherent contribution to the spectral lineshape in Sr3Ir2O7, which evolves strongly at TN. On the contrary, the Sr2IrO4 lineshape is dominated by the incoherent part, which is insensitive to TN. We compare these findings to theoretical expections of the Slater vs Mott antiferromagnetism within Dynamical Mean Field Theory.

Some correlated materials display an insulator-to-metal transition as the temperature is increased. In most cases this transition can also be induced electrically, resulting in volatile resistive switching due to the formation of a conducting filament. While this phenomenon has attracted much attention due to potential applications, many fundamental questions remain unaddressed. One of them is its characteristic lengths: what sets the size of these filaments, and how does this impact resistive switching properties. Here we use a combination of wide-field and scattering-type scanning near-field optical microscopies to characterize filament formation in NdNiO3 and SmNiO3 thin films. We find a clear trend: smaller filaments increase the current density, yielding sharper switching and a larger resistive drop. With the aid of numerical simulations, we discuss the parameters controlling the filament width and, hence, the switching properties.

We propose an algorithm for learning a conditional generative model of a molecule given a target. Specifically, given a receptor molecule that one wishes to bind to, the conditional model generates candidate ligand molecules that may bind to it. The distribution should be invariant to rigid body transformations that act $\textit{jointly}$ on the ligand and the receptor; it should also be invariant to permutations of either the ligand or receptor atoms. Our learning algorithm is based on a continuous normalizing flow. We establish semi-equivariance conditions on the flow which guarantee the aforementioned invariance conditions on the conditional distribution. We propose a graph neural network architecture which implements this flow, and which is designed to learn effectively despite the vast differences in size between the ligand and receptor. We evaluate our method on the CrossDocked2020 dataset, attaining a significant improvement in binding affinity over competing methods.

In solids, strong repulsion between electrons can inhibit their movement and result in a "Mott" metal-to-insulator transition (MIT), a fundamental phenomenon whose understanding has remained a challenge for over 50 years. A key issue is how the wave-like itinerant electrons change into a localized-like state due to increased interactions. However, observing the MIT in terms of the energy- and momentum-resolved electronic structure of the system, the only direct way to probe both itinerant and localized states, has been elusive. Here we show, using angle-resolved photoemission spectroscopy (ARPES), that in V$_2$O$_3$ the temperature-induced MIT is characterized by the progressive disappearance of its itinerant conduction band, without any change in its energy-momentum dispersion, and the simultaneous shift to larger binding energies of a quasi-localized state initially located near the Fermi level., Comment: Main Text (4 figures) and Supplementary Information (12 figures)

We study correlation functions for extremal supersymmetric black holes. It is necessary to take into account the strongly coupled nature of the boundary supergraviton mode. We consider the case with ${\cal N}=2$ supercharges which is the minimal amount of supersymmetry needed to give a large ground state degeneracy, separated from the continuum. Using the exact solution for this theory we derive formulas for the two point function and we also give integral expressions for any $n$-point correlator. These correlators are time independent at large times and approach constant values that depend on the masses and couplings of the bulk theory. We also explain that in the non-supersymmetric case, the correlators develop a universal time dependence at long times. This paper is the longer companion paper of arXiv:2207.00407., Comment: 61 pages, 19 figures; v2: fixed typos, expanded comments on random matrix behavior of projected operators, added Appendix K, v3: fixed minus signs, v4: minor improvements, v5: correction to Fig 12

We study the gravitational description of extremal supersymmetric black holes. We point out that the $AdS_2$ near horizon geometry can be used to compute interesting observables, such as correlation functions of operators. In this limit, the Hamiltonian is zero and correlation functions are time independent. We discuss some possible implications for the gravity description of black hole microstates. We also compare with numerical results in a supersymmetric version of SYK. These results can also be interpreted as providing a construction of wormholes joining two extremal black holes. This is the short version of a longer and more technical companion paper arXiv:2207.00408., Comment: 21 pages, 9 figures; v2: slightly expanded discussion, v3: minor improvements