• Publicationsclose
• Research softwareclose
• Other research productsclose
• 2018-2022close
• Open Access close
• Preprint close
• AE close
• arXiv.org e-Print Archive close

Creating compelling captions for data visualizations has been a longstanding challenge. Visualization researchers are typically untrained in journalistic reporting and hence the captions that are placed below data visualizations tend to be not overly engaging and rather just stick to basic observations about the data. In this work we explore the opportunities offered by the newly emerging crop of large language models (LLM) which use sophisticated deep learning technology to produce human-like prose. We ask, can these powerful software devices be purposed to produce engaging captions for generic data visualizations like a scatterplot. It turns out that the key challenge lies in designing the most effective prompt for the LLM, a task called prompt engineering. We report on first experiments using the popular LLM GPT-3 and deliver some promising results.

Laser wakefield acceleration is paving the way for the next generation of electron accelerators, for their own sake and as radiation sources. A controllable dual-wake injection scheme is put forward here to generate an ultrashort triplet electron bunch with high brightness and high polarization, employing a radially polarized laser as a driver. We find that the dual wakes can be driven by both transverse and longitudinal components of the laser field in the quasi-blowout regime, sustaining the laser-modulated wakefield which facilitates the sub-cycle and transversely-split injection of the triplet bunch. {Polarization of the triplet bunch can be highly preserved due to the laser-assisted collective spin precession and the non-canceled transverse spins. In our three-dimensional particle-in-cell simulations, the triplet electron bunch, with duration about $500$ as, six-dimensional brightness exceeding $10^{14}$ A/m$^2$/0.1$\%$ and polarization over $80\%$, can be generated using a few-terawatt laser}. Such an electron bunch could play an essential role in many applications, such as ultrafast imaging, nuclear structure and high-energy physics studies, and the operation of coherent radiation sources.

With over 100 years of solar observations, the Kodaikanal Solar Observatory (KoSO) is a one-of-a-kind solar data repository in the world. Among its many data catalogues, the `suncharts' at KoSO are of particular interest. These Suncharts (1904-2020) are coloured drawings of different solar features, such as sunspots, plages, filaments, and prominences, made on papers with a Stonyhurst latitude-longitude grid etched on them. In this paper, we analyze this unique data by first digitizing each suncharts using an industry-standard scanner and saving those digital images in high-resolution `.tif' format. We then examine the Cycle~19 and Cycle~20 data (two of the strongest cycles of the last century) with the aim of detecting filaments. To this end, we employed `k-means clustering' method and obtained different filament parameters such as position, tilt angle, length, and area. Our results show that filament length (and area) increases with latitude and the pole-ward migration is clearly dominated by a particular tilt sign. Lastly, we cross-verified our findings with results from KoSO digitized photographic plate database for the overlapping time period and obtained a good agreement between them. This work, acting as a proof-of-the-concept, will kick-start new efforts to effectively use the entire hand-drawn series of multi-feature, full-disk solar data and enable researchers to extract new sciences, such as the generation of pseudo magnetograms for the last 100 years. Comment: 12 pages, 7 Figures, Accepted for publication in ApJ

Highly excited Rydberg atoms are a powerful platform for quantum simulation and information processing. Here, we propose atomic ring networks to study chiral currents of Rydberg excitations. The currents are controlled by a phase pattern imprinted via a Raman scheme and can persist even in the presence of dephasing. Depending on the interplay between the Rabi coupling of Rydberg states and the dipole-dipole atom interaction, the current shows markedly different features. The excitations propagate with a velocity displaying a characteristic peak in time, reflecting the chiral nature of the current. We find that the time-averaged current in a quench behaves similarly to the ground-state current. This analysis paves the way for the development of new methods to transport information in atomic networks. Comment: 6 pages main, 6 figures + appendices

In this work, we investigate the viability of a light Higgs ($\eta$) scenario in the Georgi-Machacek (GM) model, where we consider all theoretical and experimental constraints such as the perturbativity, vacuum stability, unitarity, electroweak precision tests, the Higgs di-photon and undetermined decays and the Higgs total decay width. In addition, we consider more recent experimental bounds from the searches for doubly-charged Higgs bosons in the VBF channel $H_{5}^{++}\rightarrow W^{+}W^{+}$, Drell-Yan production of a neutral Higgs boson $pp\rightarrow H_{5}^{0}(\gamma\gamma)H_{5}^{+}$, and for the light scalars at LEP $e^{-}e^{+}\rightarrow Z\eta$, and at ATLAS and CMS in different final states such as $pp\rightarrow\eta\rightarrow2\gamma$ and $pp\rightarrow h\rightarrow\eta\eta\rightarrow4\gamma,2\mu2\tau,2\mu2b,2\tau2b$. By combining these bounds together, we found a parameter space region that is significant as the case of the SM-like Higgs to be the light CP-even eigenstate, and this part of the parameter space would be tightened by the coming analyses. Comment: 13 pages, 3 figures, version includes modifications due to the erratum in https://journals.aps.org/prd/abstract/10.1103/PhysRevD.108.019902

We identify a class of potentials for which the scattering of flat-top solitons and thin-top solitons of the nonlinear Schr\"odinger equation with dual nonlinearity can be reflectionless. The scattering is characterized by sharp resonances between regimes of full transmission and full quantum reflection. Perturbative expansion in terms of the magnitude of radiation losses leads to the general form of reflectionless potentials. Simulating the scattering of flat-top solitons and thin-top solitons confirms the reflectionless feature of these potentials.

Pre-trained language models (LMs) have shown remarkable reasoning performance using explanations (or ``chain-of-thought'' (CoT)) for in-context learning. On the other hand, these reasoning tasks are usually presumed to be more approachable for symbolic programming. To make progress towards understanding in-context learning, we curate synthetic datasets containing equivalent (natural, symbolic) data pairs, where symbolic examples contain first-order logic rules and predicates from knowledge bases (KBs). Then we revisit neuro-symbolic approaches and use Language Models as Logic Programmer (LMLP) that learns from demonstrations containing logic rules and corresponding examples to iteratively reason over KBs, recovering Prolog's backward chaining algorithm. Comprehensive experiments are included to systematically compare LMLP with CoT in deductive reasoning settings, showing that LMLP enjoys more than 25% higher accuracy than CoT on length generalization benchmarks even with fewer parameters. Comment: NeurIPS Neuro Causal and Symbolic AI Workshop, 2022

In the superconducting state, the presence of a finite gap in the excitation spectrum implies that the number of excitations (quasiparticles) is exponentially small at temperatures well below the critical one. Conversely, minute perturbations can significantly impact both the distribution in energy and number of quasiparticles. Typically, the interaction with the electromagnetic environment is the main perturbation source driving quasiparticles out of thermal equilibrium, while a phonon bath is responsible for restoration of equilibrium. Here we derive approximate analytical solutions for the quasiparticle distribution function in superconducting resonators and explore the impact of nonequilibrium on two measurable quantities: the resonator's quality factor and its resonant frequency. Applying our results to experimental data, we conclude that while at intermediate temperatures there is clear evidence for the nonequilibrium effects due to heating of the quasiparticles by photons, the low-temperature measurements are not explained by this mechanism. Comment: 22 pages, 8 figures

Quantum state tomography is an essential component of modern quantum technology. In application to continuous-variable harmonic-oscilator systems, such as the electromagnetic field, existing tomography methods typically reconstruct the state in discrete bases, and are hence limited to states with relatively low amplitudes and energies. Here we overcome this limitation by utilizing a feed-forward neural network to obtain the density matrix directly in the continuous position basis. An important benefit of our approach is the ability to choose specific regions in the phase space for detailed reconstruction. This results in relatively slow scaling of the amount of resources required for the reconstruction with the state amplitude, and hence allows us to dramatically increase the range of amplitudes accessible with our method. Comment: 8 pages, 4 figures

Traffic forecasting has attracted widespread attention recently. In reality, traffic data usually contains missing values due to sensor or communication errors. The Spatio-temporal feature in traffic data brings more challenges for processing such missing values, for which the classic techniques (e.g., data imputations) are limited: 1) in temporal axis, the values can be randomly or consecutively missing; 2) in spatial axis, the missing values can happen on one single sensor or on multiple sensors simultaneously. Recent models powered by Graph Neural Networks achieved satisfying performance on traffic forecasting tasks. However, few of them are applicable to such a complex missing-value context. To this end, we propose GCN-M, a Graph Convolutional Network model with the ability to handle the complex missing values in the Spatio-temporal context. Particularly, we jointly model the missing value processing and traffic forecasting tasks, considering both local Spatio-temporal features and global historical patterns in an attention-based memory network. We propose as well a dynamic graph learning module based on the learned local-global features. The experimental results on real-life datasets show the reliability of our proposed method. Comment: To appear in Data Mining and Knowledge Discovery (DMKD), Springer