Search

Large computer models are ubiquitous in the earth sciences. These models often have tens or hundreds of tuneable parameters and can take thousands of core-hours to run to completion while generating terabytes of output. It is becoming common practice to develop emulators as fast approximations, or surrogates, of these models in order to explore the relationships between these inputs and outputs, understand uncertainties and generate large ensembles datasets. While the purpose of these surrogates may differ, their development is often very similar. Here we introduce ESEm: an open-source tool providing a general workflow for emulating and validating a wide variety of models and outputs. It includes efficient routines for sampling these emulators for the purpose of uncertainty quantification and model calibration. It is built on well-established, high-performance libraries to ensure robustness, extensibility and scalability. We demonstrate the flexibility of ESEm through three case-studies using ESEm to reduce parametric uncertainty in a general circulation model, explore precipitation sensitivity in a cloud resolving model and scenario uncertainty in the CMIP6 multi-model ensemble.

Sparsity in the structure of Neural Networks can lead to less energy consumption, less memory usage, faster computation times on convenient hardware, and automated machine learning. If sparsity gives rise to certain kinds of structure, it can explain automatically obtained features during learning. We provide insights into experiments in which we show how sparsity can be achieved through prior initialization, pruning, and during learning, and answer questions on the relationship between the structure of Neural Networks and their performance. This includes the first work of inducing priors from network theory into Recurrent Neural Networks and an architectural performance prediction during a Neural Architecture Search. Within our experiments, we show how magnitude class blinded pruning achieves 97.5% on MNIST with 80% compression and re-training, which is 0.5 points more than without compression, that magnitude class uniform pruning is significantly inferior to it and how a genetic search enhanced with performance prediction achieves 82.4% on CIFAR10. Further, performance prediction for Recurrent Networks learning the Reber grammar shows an $R^2$ of up to 0.81 given only structural information.

Nanoplasmonic systems combined with optically-active two-dimensional materials provide intriguing opportunities to explore and control light-matter interactions at extreme sub-wavelength lengthscales approaching the exciton Bohr radius. Here, we present room- and cryogenic-temperature investigations of light-matter interactions between an MoSe$_2$ monolayer and individual lithographically defined gold dipole nanoantennas having sub-10 nm feed gaps. By progressively tuning the nanoantenna size, their dipolar resonance is tuned relative to the A-exciton transition in a proximal MoSe$_2$ monolayer achieving a total tuning of $\sim 130\;\mathrm{meV}$. Differential reflectance measurements performed on $> 100$ structures reveal an apparent avoided crossing between exciton and dipolar mode and an exciton-plasmon coupling constant of $g= 55\;\mathrm{meV}$, representing $g/(\hbar\omega_X)\geq3\%$ of the transition energy. This places our hybrid system in the intermediate-coupling regime where spectra exhibit a characteristic Fano-like shape, indicative of the interplay between pronounced light-matter coupling and significant damping. We also demonstrate active control of the optical response by varying the polarization of the excitation light to programmably suppress coupling to the dipole mode. We further study the emerging optical signatures of the monolayer localized at dipole nanoantennas at $10\;\mathrm{K}$. Our findings represent a key step towards realizing non-linear photonic devices based on 2D materials with potential for low-energy and ultrafast performance., Comment: 17 pages, 12 figures, https://pubs.acs.org/doi/10.1021/acs.nanolett.1c02676

Deep learning models have been shown to be vulnerable to adversarial attacks. This perception led to analyzing deep learning models not only from the perspective of their performance measures but also their robustness to certain types of adversarial attacks. We take another step forward in relating the architectural structure of neural networks from a graph theoretic perspective to their robustness. We aim to investigate any existing correlations between graph theoretic properties and the robustness of Sparse Neural Networks. Our hypothesis is, that graph theoretic properties as a prior of neural network structures are related to their robustness. To answer to this hypothesis, we designed an empirical study with neural network models obtained through random graphs used as sparse structural priors for the networks. We additionally investigated the evaluation of a randomly pruned fully connected network as a point of reference. We found that robustness measures are independent of initialization methods but show weak correlations with graph properties: higher graph densities correlate with lower robustness, but higher average path lengths and average node eccentricities show negative correlations with robustness measures. We hope to motivate further empirical and analytical research to tightening an answer to our hypothesis.

Atomically thin semiconductors can be readily integrated into a wide range of nanophotonic architectures for applications in quantum photonics and novel optoelectronic devices. We report the observation of nonlocal interactions of \textit{free} trions in pristine hBN/MoS$_2$/hBN heterostructures coupled to single mode (Q $>10^4$) quasi 0D nanocavities. The high excitonic and photonic quality of the interaction system stems from our integrated nanofabrication approach simultaneously with the hBN encapsulation and the maximized local cavity field amplitude within the MoS$_2$ monolayer. We observe a nonmonotonic temperature dependence of the cavity-trion interaction strength, consistent with the nonlocal light-matter interactions in which the extent of the center-of-mass wavefunction is comparable to the cavity mode volume in space. Our approach can be generalized to other optically active 2D materials, opening the way towards harnessing novel light-matter interaction regimes for applications in quantum photonics.

Manipulating electronic interlayer coupling in layered van der Waals (vdW) materials is essential for designing opto-electronic devices. Here, we control vibrational and electronic interlayer coupling in bi- and trilayer 2H-MoS$_2$ using large external electric fields in a micro-capacitor device. The electric field lifts Raman selection rules and activates phonon modes in excellent agreement with ab-initio calculations. Through polarization resolved photoluminescence spectroscopy in the same device, we observe a strongly tunable valley dichroism with maximum circular polarization degree of $\sim 60\%$ in bilayer and $\sim 35\%$ in trilayer MoS$_2$ that are fully consistent with a rate equation model which includes input from electronic band structure calculations. We identify the highly delocalized electron wave function between the layers close to the high symmetry $Q$ points as the origin of the tunable circular dichroism. Our results demonstrate the possibility of electric field tunable interlayer coupling for controlling emergent spin-valley physics and hybridization driven effects in vdW materials and their heterostructures., Comment: Main manuscript: 10 pages, 4 figures ; Supplemental material: 14 pages, 9 figures

Moderating content in social media platforms is a formidable challenge due to the unprecedented scale of such systems, which typically handle billions of posts per day. Some of the largest platforms such as Facebook blend machine learning with manual review of platform content by thousands of reviewers. Operating a large-scale human review system poses interesting and challenging methodological questions that can be addressed with operations research techniques. We investigate the problem of optimally operating such a review system at scale using ideas from queueing theory and simulation.

In a comprehensive study on several samples we demonstrate for our laboratory-based computed tomography system resolutions down to 150nm. The achieved resolution is validated by imaging com-mon test structures in 2D and Fourier Shell Correlation of 3D volumes. As representative application examples from nowadays material research, we show metallization processes in multilayer integrated circuits, ageing in lithium battery electrodes, and volumetric of metallic sub-micrometer fillers of com-posites. Thus, our laboratory system provides the unique possibility to image non-destructively struc-tures in the range of hundred nanometers, even for high density materials., Comment: 18 pages, 8 figures

Parzanchevski and Sarnak recently adapted an algorithm of Ross and Selinger for factorization of PU(2)-diagonal elements to within distance $\varepsilon$ into an efficient probabilistic algorithm for any PU(2)-element, using at most $3\log_p\frac{1}{\varepsilon^3}$ factors from certain well-chosen sets. The Clifford+$T$ gates are one such set arising from $p=2$. In that setting, we leverage recent work of Carvalho Pinto and Petit to improve this to $\frac{7}{3}\log_2\frac{1}{\varepsilon^3}$, and implement the algorithm in Haskell.

The cyclic sieving phenomenon of Reiner, Stanton, and White characterizes the stabilizers of cyclic group actions on finite sets using q-analogue polynomials. Eu and Fu demonstrated a cyclic sieving phenomenon on generalized cluster complexes of every type using the q-Catalan numbers. In this paper, we exhibit the dihedral sieving phenomenon, introduced for odd n by Rao and Suk, on clusters of every type. In the type A case, we show that the Raney numbers count both reflection-symmetric k-angulations of an n-gon and a particular evaluation of the (q,t)-Fuss--Catalan numbers. We also introduce a sieving phenomenon for the symmetric group, and discuss possibilities for dihedral sieving for even n., Comment: 26 pages, 13 figures

We study how impurities influence the current-induced dynamics of magnetic Skyrmions moving in a racetrack geometry. For this, we solve numerically the generalized Landau-Lifshitz-Gilbert equation extended by the current-induced spin transfer torque. In particular, we investigate two classes of impurities, non-conducting and magnetic impurities. The former are magnetically rigid objects and yield to an inhomogeneous current density over the racetrack which we determine separately by solving the fundamental electrostatic equations. In contrast, magnetic impurities leave the applied current density homogeneous throughout the stripe. Depending on parameters, we observe four different scenarios of Skyrmion motions in the presence of disorder, the Skyrmion decay, the pinning, the creation of additional Skyrmions, and ordinary Skyrmion passage. We calculate and discuss phase diagrams in dependence of the impurity concentration and radii of the impurities.

Monolayers of semiconducting transition metal dichalcogenides are a strongly emergent platform for exploring quantum phenomena in condensed matter, building novel opto-electronic devices with enhanced functionalities. Due to their atomic thickness, their excitonic optical response is highly sensitive to their dielectric environment. In this work, we explore the optical properties of monolayer thick MoSe$_2$ straddling domain wall boundaries in periodically poled LiNbO$_3$. Spatially-resolved photoluminescence experiments reveal spatial sorting of charge and photo-generated neutral and charged excitons across the boundary. Our results reveal evidence for extremely large in-plane electric fields of 3000\,kV/cm at the domain wall whose effect is manifested in exciton dissociation and routing of free charges and trions toward oppositely poled domains and a non-intuitive spatial intensity dependence. By modeling our result using drift-diffusion and continuity equations, we obtain excellent qualitative agreement with our observations and have explained the observed spatial luminescence modulation using realistic material parameters., Comment: 29 pages, 6 figures, submetted material

Materials combining the optoelectronic functionalities of semiconductors with control of the spin degree of freedom are highly sought after for the advancement of quantum technology devices. Here, we report the paramagnetic Ruddlesden-Popper hybrid perovskite Mn:(PEA)2PbI4 (PEA = phenethylammonium) in which the interaction of isolated Mn2+ ions with magnetically brightened excitons leads to circularly polarized photoluminescence. Using a combination of superconducting quantum interference device (SQUID) magnetometry and magneto-optical experiments, we find that the Brillouin-shaped polarization curve of the photoluminescence follows the magnetization of the material. This indicates coupling between localized manganese magnetic moments and exciton spins via a magnetic proximity effect. The saturation polarization of 15% at 4 K and 6 T indicates a highly imbalanced spin population and demonstrates that manganese doping enables efficient control of excitonic spin states in Ruddlesden-Popper perovskites. Our finding constitutes the first example of polarization control in magnetically doped hybrid perovskites and will stimulate research on this highly tuneable material platform that promises tailored interactions between magnetic moments and electronic states.

Rare earth ions typically exhibit larger magnetic moments than transition metal ions and thus promise the opening of a wider exchange gap in the Dirac surface states of topological insulators. Yet, in a recent photoemission study of Eu-doped Bi$_2$Te$_3$ films, the spectra remained gapless down to $T = 20\;\text{K}$. Here, we scrutinize whether the conditions for a substantial gap formation in this system are present by combining spectroscopic and bulk characterization methods with theoretical calculations. For all studied Eu doping concentrations, our atomic multiplet analysis of the $M_{4,5}$ x-ray absorption and magnetic circular dichroism spectra reveals a Eu$^{2+}$ valence and confirms a large magnetic moment, consistent with a $4f^7 \; {^8}S_{7/2}$ ground state. At temperatures below $10\;\text{K}$, bulk magnetometry indicates the onset of antiferromagnetic (AFM) ordering. This is in good agreement with density functional theory, which predicts AFM interactions between the Eu impurities. Our results support the notion that antiferromagnetism can coexist with topological surface states in rare-earth doped Bi$_2$Te$_3$ and call for spectroscopic studies in the kelvin range to look for novel quantum phenomena such as the quantum anomalous Hall effect., Comment: 15 pages, 9 figures

Topologically distinct magnetic structures like skyrmions, domain walls, and the uniformly magnetized state have multiple applications in logic devices, sensors, and as bits of information. One of the most promising concepts for applying these bits is the racetrack architecture controlled by electric currents or magnetic driving fields. In state-of-the-art racetracks, these fields or currents are applied to the whole circuit. Here, we employ micromagnetic and atomistic simulations to establish a concept for racetrack memories free of global driving forces. Surprisingly, we realize that mixed sequences of topologically distinct objects can be created and propagated over far distances exclusively by local rotation of magnetization at the sample boundaries. We reveal the dependence between the chirality of the rotation and the direction of propagation and define the phase space where the proposed procedure can be realized. The advantages of this approach are the exclusion of high current and field densities as well as its compatibility with an energy-efficient three-dimensional design., Comment: 23 pages, 7 figures

We report magneto-absorption spectroscopy of gated WSe$_2$ monolayers in high magnetic fields up to 60~T. When doped with a 2D Fermi sea of mobile holes, well-resolved sequences of optical transitions are observed in both $\sigma^\pm$ circular polarizations, which unambiguously and separately indicate the number of filled Landau levels (LLs) in both $K$ and $K'$ valleys. This reveals the interaction-enhanced valley Zeeman energy, which is found to be highly tunable with hole density $p$. We exploit this tunability to align the LLs in $K$ and $K'$, and find that the 2D hole gas becomes unstable against small changes in LL filling and can spontaneously valley-polarize. These results cannot be understood within a single-particle picture, highlighting the importance of exchange interactions in determining the ground state of 2D carriers in monolayer semiconductors., Comment: 8 pages, 4 figures + 3 supplementary figures

We report magneto-optical spectroscopy of gated monolayer MoS$_2$ in high magnetic fields up to 28T and obtain new insights on the many-body interaction of neutral and charged excitons with the resident charges of distinct spin and valley texture. For neutral excitons at low electron doping, we observe a nonlinear valley Zeeman shift due to dipolar spin-interactions that depends sensitively on the local carrier concentration. As the Fermi energy increases to dominate over the other relevant energy scales in the system, the magneto-optical response depends on the occupation of the fully spin-polarized Landau levels in both $K/K^{\prime}$ valleys. This manifests itself in a many-body state. Our experiments demonstrate that the exciton in monolayer semiconductors is only a single particle boson close to charge neutrality. We find that away from charge neutrality it smoothly transitions into polaronic states with a distinct spin-valley flavour that is defined by the Landau level quantized spin and valley texture., Comment: Main manuscript: 7 pages, 4 figures ; Supplemental material: 20 pages, 8 figures

Learning distributions of graphs can be used for automatic drug discovery, molecular design, complex network analysis, and much more. We present an improved framework for learning generative models of graphs based on the idea of deep state machines. To learn state transition decisions we use a set of graph and node embedding techniques as memory of the state machine. Our analysis is based on learning the distribution of random graph generators for which we provide statistical tests to determine which properties can be learned and how well the original distribution of graphs is represented. We show that the design of the state machine favors specific distributions. Models of graphs of size up to 150 vertices are learned. Code and parameters are publicly available to reproduce our results., Comment: 8 pages condensed preprint with github link, under review

The current outbreak of novel coronavirus disease 2019 (COVID-19) poses an unprecedented global health and economic threat to interconnected human societies. Until a vaccine is developed, strategies for controlling the outbreak rely on aggressive social distancing. These measures largely disconnect the social network fabric of human societies, especially in urban areas. Here, we estimate the growth rates and reproductive numbers of COVID-19 in US cities from March 14th through March 19th to reveal a power-law scaling relationship to city population size. This means that COVID-19 is spreading faster on average in larger cities with the additional implication that, in an uncontrolled outbreak, larger fractions of the population are expected to become infected in more populous urban areas. We discuss the implications of these observations for controlling the COVID-19 outbreak, emphasizing the need to implement more aggressive distancing policies in larger cities while also preserving socioeconomic activity.

Sarnak's golden mean conjecture states that $(m+1)d_\varphi(m)\le1+\frac{2}{\sqrt{5}}$ for all integers $m\ge1$, where $\varphi$ is the golden mean and $d_\theta$ is the discrepancy function for $m+1$ multiples of $\theta$ modulo 1. In this paper, we characterize the set $\mathcal{S}$ of values $\theta$ that share this property, as well as the set $\mathcal{T}$ of those with the property for some lower bound $m\ge M$. Remarkably, $\mathcal{S}\text{ mod }1$ has only 16 elements, whereas $\mathcal{T}$ is the set of $GL_2(\mathbb{Z})$-transformations of $\varphi$.

Since the discovery of the quantum anomalous Hall effect in the magnetically doped topological insulators (MTI) Cr:(Bi,Sb)$_2$Te$_3$ and V:(Bi,Sb)$_2$Te$_3$, the search for the exchange coupling mechanisms underlying the onset of ferromagnetism has been a central issue, and a variety of different scenarios have been put forward. By combining resonant photoemission, X-ray magnetic dichroism and multiplet ligand field theory, we determine the local electronic and magnetic configurations of V and Cr impurities in (Bi,Sb)$_2$Te$_3$. While strong pd hybridisation is found for both dopant types, their 3d densities of states show pronounced differences. State-of-the-art first-principles calculations show how these impurity states mediate characteristic short-range pd exchange interactions, whose strength sensitively varies with the position of the 3d states relative to the Fermi level. Measurements on films with varying host stoichiometry support this trend. Our results establish the essential role of impurity-state mediated exchange interactions in the magnetism of MTI.

We consider the question of whether, and in what sense, Wardrop equilibria provide a good approximation for Nash equilibria in atomic unsplittable congestion games with a large number of small players. We examine two different definitions of small players. In the first setting, we consider games where each player's weight is small. We prove that when the number of players goes to infinity and their weights to zero, the random flows in all (mixed) Nash equilibria for the finite games converge in distribution to the set of Wardrop equilibria of the corresponding nonatomic limit game. In the second setting, we consider an increasing number of players with a unit weight that participate in the game with a decreasingly small probability. In this case, the Nash equilibrium flows converge in total variation towards Poisson random variables whose expected values are Wardrop equilibria of a different nonatomic game with suitably-defined costs. The latter can be viewed as symmetric equilibria in a Poisson game in the sense of Myerson, establishing a plausible connection between the Wardrop model for routing games and the stochastic fluctuations observed in real traffic. In both settings we provide explicit approximation bounds, and we study the convergence of the price of anarchy. Beyond the case of congestion games, we prove a general result on the convergence of large games with random players towards Poisson games., Comment: Mathematics of Operations Research, Forthcoming, 2022