Search

We investigate the easy-axis Heisenberg model on the triangular lattice by numerically studying the ground-state dynamical spin structure factor $S^{\mu\mu}({\bf q},\omega)$ on lattices up to $N=36$ sites. Results are analyzed within the supersolid scenario, characterized by the translation-symmetry-breaking parameter $m_z$ and the rotational supersolid order parameter $m_\perp$. We find very robust $m_z > 0$ in the whole easy-axis anisotropy regime $\alpha = J_\perp/J_z > 0$, even enhanced by the magnetic field $h>0$. Results support also $m_\perp >0$ for intermediate $\alpha <1$ as well as for $h>0$. Still, at small $\alpha \ll 1$, relevant for recent experiments on the magnetic material K$_2$Co(SeO$_3$)$_2$, we find at $h=0$ rather vanishing $m_\perp \sim 0$, reflecting the gapped spin excitations., Comment: 4 pages, 4 figures

Deep learning methods in LiDAR-based archaeological research often leverage visualisation techniques derived from Digital Elevation Models to enhance characteristics of archaeological objects present in the images. This paper investigates the impact of visualisations on deep learning performance through a comprehensive testing framework. The study involves the use of eight semantic segmentation models to evaluate seven diverse visualisations across two study areas, encompassing five archaeological classes. Experimental results reveal that the choice of appropriate visualisations can influence performance by up to 8%. Yet, pinpointing one visualisation that outperforms the others in segmenting all archaeological classes proves challenging. The observed performance variation, reaching up to 25% across different model configurations, underscores the importance of thoughtfully selecting model configurations and LiDAR visualisations for successfully segmenting archaeological objects.

The bad metallic phase with resistivity above the Mott-Ioffe-Regel limit, which appears also in cuprate superconductors, was recently understood by cold atom and computer simulations of the Hubbard model via charge susceptibility and charge diffusion constant. However, since reliable simulations can be typically done only at temperatures above the experimental temperatures, the question for cuprate superconductors is still open. This paper addresses this question by resorting to heat transport, which allows for the estimate of electronic diffusion and it further combines it with the resistivity to estimate the charge susceptibility. The doping and temperature dependencies of diffusion constant and charge susceptibilities are shown and discussed for two samples of YBa$_2$Cu$_3$O$_{6+x}$. Results indicate strongly incoherent transport, mean free path corresponding to the Mott-Ioffe-Regel limit for the underdoped sample at temperatures above ~200 K and significant effect of the charge susceptibility on the resistivity., Comment: 5+1 pages, 2 figures

Next-generation cellular concepts rely on the processing of large quantities of radio-frequency (RF) samples. This includes Radio Access Networks (RAN) connecting the cellular front-end based on software defined radios (SDRs) and a framework for the AI processing of spectrum-related data. The RF data collected by the dense RAN radio units and spectrum sensors may need to be jointly processed for intelligent decision making. Moving large amounts of data to AI agents may result in significant bandwidth and latency costs. We propose a deep learned compression (DLC) model, HQARF, based on learned vector quantization (VQ), to compress the complex-valued samples of RF signals comprised of 6 modulation classes. We are assessing the effects of HQARF on the performance of an AI model trained to infer the modulation class of the RF signal. Compression of narrow-band RF samples for the training and off-the-site inference will allow for an efficient use of the bandwidth and storage for non-real-time analytics, and for a decreased delay in real-time applications. While exploring the effectiveness of the HQARF signal reconstructions in modulation classification tasks, we highlight the DLC optimization space and some open problems related to the training of the VQ embedded in HQARF.

We study charge and heat transport in the square lattice Hubbard model at strong coupling using the finite-temperature Lanczos method. We construct the diffusion matrix and estimate the effect of thermoelectric terms on diffusive and hydrodynamic time evolution. The thermoelectric terms prevent the interpretation of the diffusion in terms of a single time scale. We discuss our results in relation to cold-atom experiments and measurements of heat conductivity based on the measurements of heat diffusion., Comment: 13 pages

Motivated by recent experiments on a compound {displaying Ising-like short-range correlations on the triangular lattice, we study the anisotropic easy-axis spin-$1/2$ Heisenberg model on the triangular and kagome lattice} by performing numerical calculations of finite-temperature properties, in particular of static spin structure factor and of thermodynamic quantities, on systems with up to 36 sites. On the triangular lattice, the low-temperature spin structure factor {exhibits long-range} spin correlations in the whole range of anisotropies, whereas thermodynamic quantities reveal a crossover upon increasing the anisotropy, most pronounced in the vanishing generalized Wilson ratio in the easy-axis regime. In contrast, on the kagome lattice, the spin structure factor is short-range, and thermodynamic quantities evolve steadily between the easy-axis and the isotropic case, consistent with the interpretation in terms of {a} spin liquid., Comment: 9 pages

When applying deep learning to remote sensing data in archaeological research, a notable obstacle is the limited availability of suitable datasets for training models. The application of transfer learning is frequently employed to mitigate this drawback. However, there is still a need to explore its effectiveness when applied across different archaeological datasets. This paper compares the performance of various transfer learning configurations using two semantic segmentation deep neural networks on two LiDAR datasets. The experimental results indicate that transfer learning-based approaches in archaeology can lead to performance improvements, although a systematic enhancement has not yet been observed. We provide specific insights about the validity of such techniques that can serve as a baseline for future works., Comment: Accepted to IEEE International Geoscience and Remote Sensing Symposium 2023 (IGARSS 2023) @IEEE copyright

We study the electronic thermal conductivity $\kappa_\textrm{el}$ and the thermal diffusion constant $D_\textrm{Q,el}$ in the square lattice Hubbard model using the finite-temperature Lanczos method. We exploit the Nernst-Einstein relation for thermal transport and interpret the strong non-monotonous temperature dependence of $\kappa_\textrm{el}$ in terms of that of $D_\textrm{Q,el}$ and the electronic specific heat $c_\textrm{el}$. We present also the results for the Heisenberg model on a square lattice and ladder geometries. We study the effects of doping and consider the doped case also with the dynamical mean-field theory. We show that $\kappa_\textrm{el}$ is below the corresponding Mott-Ioffe-Regel value in almost all calculated regimes, while the mean free path is typically above or close to lattice spacing. We discuss the opposite effect of quasi-particle renormalization on charge and heat diffusion constants. We calculate the Lorenz ratio and show that it differs from the Sommerfeld value. We discuss our results in relation to experiments on cuprates. Additionally, we calculate the thermal conductivity of overdoped cuprates within the anisotropic marginal Fermi liquid phenomenological approach., Comment: 8+5 figures

The volume contains selected contributions from the Machine Learning Challenge "Discover the Mysteries of the Maya", presented at the Discovery Challenge Track of The European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML PKDD 2021). Remote sensing has greatly accelerated traditional archaeological landscape surveys in the forested regions of the ancient Maya. Typical exploration and discovery attempts, beside focusing on whole ancient cities, focus also on individual buildings and structures. Recently, there have been several successful attempts of utilizing machine learning for identifying ancient Maya settlements. These attempts, while relevant, focus on narrow areas and rely on high-quality aerial laser scanning (ALS) data which covers only a fraction of the region where ancient Maya were once settled. Satellite image data, on the other hand, produced by the European Space Agency's (ESA) Sentinel missions, is abundant and, more importantly, publicly available. The "Discover the Mysteries of the Maya" challenge aimed at locating and identifying ancient Maya architectures (buildings, aguadas, and platforms) by performing integrated image segmentation of different types of satellite imagery (from Sentinel-1 and Sentinel-2) data and ALS (lidar) data., Comment: Chapter authors. Chapter 1: Matthew Painter and Iris Kramer; Chapter 2: J\"urgen Landauer, Burkhard Hoppenstedt, and Johannes Allgaier; Chapter 3: Thorben Hellweg, Stefan Oehmcke, Ankit Kariryaa, Fabian Gieseke, and Christian Igel; Chapter 4: Christian Ayala, Carlos Aranda, and Mikel Galar

Existing datasets used to train deep learning models for narrowband radio frequency (RF) signal classification lack enough diversity in signal types and channel impairments to sufficiently assess model performance in the real world. We introduce the Sig53 dataset consisting of 5 million synthetically-generated samples from 53 different signal classes and expertly chosen impairments. We also introduce TorchSig, a signals processing machine learning toolkit that can be used to generate this dataset. TorchSig incorporates data handling principles that are common to the vision domain, and it is meant to serve as an open-source foundation for future signals machine learning research. Initial experiments using the Sig53 dataset are conducted using state of the art (SoTA) convolutional neural networks (ConvNets) and Transformers. These experiments reveal Transformers outperform ConvNets without the need for additional regularization or a ConvNet teacher, which is contrary to results from the vision domain. Additional experiments demonstrate that TorchSig's domain-specific data augmentations facilitate model training, which ultimately benefits model performance. Finally, TorchSig supports on-the-fly synthetic data creation at training time, thus enabling massive scale training sessions with virtually unlimited datasets.

We investigate the spin Seebeck coefficient $S_s$ in the square lattice Hubbard model at high temperatures of relevance to cold-atom measurements. We solve the model with the finite-temperature Lanczos and with the dynamical mean-field theory methods and find they give similar results in the considered regime. $S_s$ exceeds the atomic 'Heikes' estimates and the Kelvin entropic estimates drastically. We analyze the behavior in terms of a mapping onto the problem of a doped attractive model and derive an approximate expression that allows relating the enhancement of $S_s$ to distinct scattering of the spin-majority and the spin-minority excitations. Our analysis reveals the limitations of entropic interpretations of Seebeck coefficient even in the high-temperature regime. Large values of $S_s$ could be observed on optical lattices. We also calculate the full diffusion matrix. We quantify the spin-thermal diffusion, that is, the extent of the mixing between the spin and the thermal diffusion and discuss the results in the context of recent measurements of the spin-diffusion constant in cold atoms., Comment: 7 pages, 4 figures

We present a new RF fingerprinting technique for wireless emitters that is based on a simple, easily and efficiently retrainable Ridge Regression (RR) classifier. The RR learns to identify devices using bursts of waveform samples, conveniently transformed and preprocessed by delay-loop reservoirs. Deep delay Loop Reservoir Computing (DLR) is our processing architecture that supports general machine learning algorithms on resource-constrained devices by leveraging delay-loop reservoir computing (RC) and innovative architectures of loop trees. In prior work, we trained and evaluated DLR using high SNR device emissions in clean channels. We here demonstrate how to use DLR for IoT authentication by performing RF-based Specific Emitter Identification (SEI), even in the presence of fading channels and heavy in-band jamming by leveraging a matched filter (MF) extension, dubbed MF-DLR. We show that the MF processing improves the SEI performance of RR without the RC transformation (MF-RR), but the MF-DLR is more robust and applicable for addressing signatures beyond waveform transients (e.g. turn-on)., Comment: arXiv admin note: text overlap with arXiv:2104.00751

Current radio frequency (RF) sensors at the Edge lack the computational resources to support practical, in-situ training for intelligent spectrum monitoring, and sensor data classification in general. We propose a solution via Deep Delay Loop Reservoir Computing (DLR), a processing architecture that supports general machine learning algorithms on compact mobile devices by leveraging delay-loop reservoir computing in combination with innovative electrooptical hardware. With both digital and photonic realizations of our design of the loops, DLR delivers reductions in form factor, hardware complexity and latency, compared to the State-of-the-Art (SoA). The main impact of the reservoir is to project the input data into a higher dimensional space of reservoir state vectors in order to linearly separate the input classes. Once the classes are well separated, traditionally complex, power-hungry classification models are no longer needed for the learning process. Yet, even with simple classifiers based on Ridge regression (RR), the complexity grows at least quadratically with the input size. Hence, the hardware reduction required for training on compact devices is in contradiction with the large dimension of state vectors. DLR employs a RR-based classifier to exceed the SoA accuracy, while further reducing power consumption by leveraging the architecture of parallel (split) loops. We present DLR architectures composed of multiple smaller loops whose state vectors are linearly combined to create a lower dimensional input into Ridge regression. We demonstrate the advantages of using DLR for two distinct applications: RF Specific Emitter Identification (SEI) for IoT authentication, and wireless protocol recognition for IoT situational awareness., Comment: arXiv admin note: text overlap with arXiv:2104.00751

We introduce a novel design for in-situ training of machine learning algorithms built into smart sensors, and illustrate distributed training scenarios using radio frequency (RF) spectrum sensors. Current RF sensors at the Edge lack the computational resources to support practical, in-situ training for intelligent signal classification. We propose a solution using Deepdelay Loop Reservoir Computing (DLR), a processing architecture that supports machine learning algorithms on resource-constrained edge-devices by leveraging delayloop reservoir computing in combination with innovative hardware. DLR delivers reductions in form factor, hardware complexity and latency, compared to the State-ofthe- Art (SoA) neural nets. We demonstrate DLR for two applications: RF Specific Emitter Identification (SEI) and wireless protocol recognition. DLR enables mobile edge platforms to authenticate and then track emitters with fast SEI retraining. Once delay loops separate the data classes, traditionally complex, power-hungry classification models are no longer needed for the learning process. Yet, even with simple classifiers such as Ridge Regression (RR), the complexity grows at least quadratically with the input size. DLR with a RR classifier exceeds the SoA accuracy, while further reducing power consumption by leveraging the architecture of parallel (split) loops. To authenticate mobile devices across large regions, DLR can be trained in a distributed fashion with very little additional processing and a small communication cost, all while maintaining accuracy. We illustrate how to merge locally trained DLR classifiers in use cases of interest.

We study the spin diffusion and spin conductivity in the square lattice Hubbard model by using the finite-temperature Lanczos method. We show that the spin diffusion behaves differently from the charge diffusion and has a nonmonotonic $T$ dependence. This is due to a progressive liberation of charges that contribute to spin transport and enhance it beyond that active at low temperature due to the Heisenberg exchange. We further show that going away from half-filling and zero magnetization increases the spin diffusion, but that the increase is insufficient to reconcile the difference between the model calculations and the recent measurements on cold-atoms., Comment: 9 pages, 5+4 figures

Current AI systems at the tactical edge lack the computational resources to support in-situ training and inference for situational awareness, and it is not always practical to leverage backhaul resources due to security, bandwidth, and mission latency requirements. We propose a solution through Deep delay Loop Reservoir Computing (DLR), a processing architecture supporting general machine learning algorithms on compact mobile devices by leveraging delay-loop (DL) reservoir computing in combination with innovative photonic hardware exploiting the inherent speed, and spatial, temporal and wavelength-based processing diversity of signals in the optical domain. DLR delivers reductions in form factor, hardware complexity, power consumption and latency, compared to State-of-the-Art . DLR can be implemented with a single photonic DL and a few electro-optical components. In certain cases multiple DL layers increase learning capacity of the DLR with no added latency. We demonstrate the advantages of DLR on the application of RF Specific Emitter Identification., Comment: 6 pages, appeared in GOMACTech 2020, released for public

High-temperature bad-metal transport has been recently studied both theoretically and in experiments as one of the key signatures of strong electronic correlations. Here we use the dynamical mean field theory (DMFT) and its cluster extensions, as well as the finite-temperature Lanczos method (FTLM) to explore the influence of lattice frustration on the thermodynamic and transport properties of the Hubbard model at high temperatures. We consider the triangular and the square lattice at half-filling and at 15\% hole-doping. We find that for $T \gtrsim 1.5t$ the self-energy becomes practically local, while the finite-size effects become small at lattice-size $4 \times 4$ for both lattice types and doping levels. The vertex corrections to optical conductivity, which are significant on the square lattice even at high temperatures, contribute less on the triangular lattice. We find approximately linear temperature dependence of dc resistivity in doped Mott insulator for both types of lattices., Comment: 11 pages

Even something as conceptually simple as adsorption of electronegative adatoms on metal surfaces, where repulsive lateral interactions are expected for obvious reasons, can lead to unanticipated behavior. In this context, we explain the origin of surprising lateral interactions between electronegative adatoms observed on some metal surfaces by means of density functional theory calculations of four electronegative atoms (N, O, F, Cl) on 70 surfaces of 44 pristine metals. Four different scenarios for lateral interactions are identified, some of them being unexpected: (i) they are repulsive, which is the typical case and occurs on almost all transition metals. (ii,iii) They are atypical, being either attractive or negligible, which occurs on p-block metals and Mg, and (iv) surface reconstruction stabilizes the low-coverage configuration, preventing atypical lateral interactions. The last case occurs predominantly on s-block metals., Comment: 8 pages, 5 figures, and supplemental material

Derivation of a reduced effective model allows for a unified treatment and discussion of the $J_1$-$J_2$ Heisenberg model on a triangular and kagome lattice. Calculating thermodynamic quantities, i.e. the entropy $s(T)$ and uniform susceptibility $\chi_0(T)$, numerically on systems up to effectively $N=48$ sites we show by comparing to full-model results that low-$T$ properties are well represented within the reduced model. Moreover, we find in the spin-liquid regime similar variation of $s(T)$ as well as $\chi_0(T)$ in both models down to $T \ll J_1$. In particular, the spin liquid appears to be characterized by Wilson ratio vanishing at low $T$, indicating on the low-lying singlet dominating over the triplet excitations., Comment: 6 pages, 5 figures + supplementary

We show that compact fully connected (FC) deep learning networks trained to classify wireless protocols using a hierarchy of multiple denoising autoencoders (AEs) outperform reference FC networks trained in a typical way, i.e., with a stochastic gradient based optimization of a given FC architecture. Not only is the complexity of such FC network, measured in number of trainable parameters and scalar multiplications, much lower than the reference FC and residual models, its accuracy also outperforms both models for nearly all tested SNR values (0 dB to 50dB). Such AE-trained networks are suited for in-situ protocol inference performed by simple mobile devices based on noisy signal measurements. Training is based on the data transmitted by real devices, and collected in a controlled environment, and systematically augmented by a policy-based data synthesis process by adding to the signal any subset of impairments commonly seen in a wireless receiver.