Magnons are the elementary excitations of the magnetic order that carry spin, momentum, and energy. Here we compare the magnon with the ferron, i.e. the elementary excitation of the electric dipolar order that transports polarization and heat in ferroelectrics.
Recently proposed neural architecture search (NAS) algorithms adopt neural predictors to accelerate the architecture search. The capability of neural predictors to accurately predict the performance metrics of neural architecture is critical to NAS,
and the acquisition of training datasets for neural predictors is time-consuming. How to obtain a neural predictor with high prediction accuracy using a small amount of training data is a central problem to neural predictor-based NAS. Here, we firstly design a new architecture encoding scheme that overcomes the drawbacks of existing vector-based architecture encoding schemes to calculate the graph edit distance of neural architectures. To enhance the predictive performance of neural predictors, we devise two self-supervised learning methods from different perspectives to pre-train the architecture embedding part of neural predictors to generate a meaningful representation of neural architectures. The first one is to train a carefully designed two branch graph neural network model to predict the graph edit distance of two input neural architectures. The second method is inspired by the prevalently contrastive learning, and we present a new contrastive learning algorithm that utilizes a central feature vector as a proxy to contrast positive pairs against negative pairs. Experimental results illustrate that the pre-trained neural predictors can achieve comparable or superior performance compared with their supervised counterparts with several times less training samples. We achieve state-of-the-art performance on the NASBench-101 and NASBench201 benchmarks when integrating the pre-trained neural predictors with an evolutionary NAS algorithm.
The Central Molecular Zone (CMZ) of our Galaxy hosts an extreme environment analogous to that found in typical starburst galaxies in the distant universe. In order to understand dust properties in environments like our CMZ, we present results from a
joint SED analysis of our AzTEC/Large Millimeter Telescope survey, together with existing textit{Herschel} far-IR data on the CMZ, from a wavelength range of $160$ $mu m$ to $1.1$ $mm$. We include global foreground and background contributions in a novel Bayesian modeling that incorporates the Point Spread Functions (PSFs) of the different maps, which enables the full utilization of our high resolution ($10.5$) map at 1.1 $mm$ and reveals unprecedentedly detailed information on the spatial distribution of dusty gas across the CMZ. There is a remarkable trend of increasing dust spectral index $beta$, from $2.0-2.4$, toward dense peaks in the CMZ, indicating a deficiency of large grains or a fundamental change in dust optical properties. This environmental dependence of $beta$ could have a significant impact on the determination of dust temperature in other studies. Depending on how the optical properties of dust deviate from the conventional model, dust temperatures could be underestimated by $10-50%$ in particularly dense regions.
We present a large-scale survey of the central molecular zone (CMZ) of our Galaxy, as well as a monitoring program of Sgr A*, with the AzTEC/Large Millimeter Telescope (LMT) in the 1.1 mm continuum. Our 1.1 mm map covers the main body of the CMZ over
a field of $1.6 times 1.1$ deg$^2$ with an angular resolution of $10.5$ and a depth of 15 mJy/beam. To account for the intensity loss due to the background removal process, we combine this map with lower resolution CSO/Bolocam and textit{Planck}/HFI data to produce an effective full intensity 1.1 mm continuum map. With this map and existing textit{Herschel} surveys, we have carried out a comprehensive analysis of the spectral energy distribution (SED) of dust in the CMZ. A key component of this analysis is the implementation of a model-based deconvolution approach, incorporating the Point Spread Functions (PSFs) of the different instruments, and hence recovering a significant amount of spatial information on angular scales larger than $10.5$. The monitoring of Sgr A* was carried out as part of a worldwide, multi-wavelength campaign when the so-called G2 object was undergoing the pericenter passage around the massive black hole (MBH). Our preliminary results include 1) high-resolution maps of column density, temperature and dust spectral index across the CMZ; 2) a 1.1~mm light curve of Sgr A* showing an outburst of $140%$ maximum amplitude on 9th May, 2014 but otherwise only stochastic variations of $10%$ and no systematic long-term change, consistent with other observations.
We show that periodic multilayered structures allow to drastically enhance near-field radiative heat transfer between nanoparticles. In particular, when the two nanoparticles are placed on each side of the multilayered structure, at the same interpar
ticle distance the resulting heat transfer is more than five orders of magnitude higher than that in the absence of the multilayered structure. This enhancement takes place in a broad range of distances and is due to the fact that the intermediate multilayered structure supports hyperbolic phonon polaritons with the key feature that the edge frequencies of the Type I and Type II Reststrahlen bands coincide with each other at a value extremely close to the particle resonance. This allow a very high-k evanescent modes resonating with the nanoparticles. Our predictions can be relevant for effective managing of energy at the nano-scale.
G21.5-0.9 is a plerionic supernova remnant (SNR) used as a calibration target for the Chandra X-ray telescope. The first observations found an extended halo surrounding the bright central pulsar wind nebula (PWN). A 2005 study discovered that this ha
lo is limb-brightened and suggested the halo to be the missing SNR shell. In 2010 the spectrum of the limb-brightened shell was found to be dominated by non-thermal X-rays. In this study, we combine 15 years of Chandra observations comprising over 1~Msec of exposure time (796.1~ks with the Advanced CCD Imaging Spectrometer (ACIS) and 306.1~ks with the High Resolution Camera (HRC)) to provide the deepest-to-date imaging and spectroscopic study. The emission from the limb is primarily non-thermal and is described by a power-law model with a photon index $Gamma = 2.22 , (2.04-2.34)$, plus a weak thermal component characterized by a temperature $kT = 0.37, (0.20-0.64)$ keV and a low ionization timescale of $n_{e}t < 2.95 times 10^{10}$ cm$^{-3}$s. The northern knot located in the halo is best fitted with a two-component power-law + non-equilibrium ionization thermal model characterized by a temperature of 0.14 keV and an enhanced abundance of silicon, confirming its nature as ejecta. We revisit the spatially resolved spectral study of the PWN and find that its radial spectral profile can be explained by diffusion models. The best fit diffusion coefficient is $D sim 2.1times 10^{27}rm cm^2/s$ assuming a magnetic field $B =130 mu G$, which is consistent with recent 3D MHD simulation results.
