Search

Increasing penetration of renewable energy introduces significant uncertainty into power systems. Traditional simulation-based verification methods may not be applicable due to the unknown-but-bounded feature of the uncertainty sets. Emerging set-theoretic methods have been intensively investigated to tackle this challenge. The paper comprehensively reviews these methods categorized by underlying mathematical principles, that is, set operation-based methods and passivity-based methods. Set operation-based methods are more computationally efficient, while passivity-based methods provide semi-analytical expression of reachable sets, which can be readily employed for control. Other features between different methods are also discussed and illustrated by numerical examples. A benchmark example is presented and solved by different methods to verify consistency.

Massive protostars generate strong radiation feedback, which may help set the mass they achieve by the end of the accretion process. Studying such feedback is therefore crucial for understanding the formation of massive stars. We report the discovery of a photoionized bipolar outflow towards the massive protostar G45.47+0.05 using high-resolution observations at 1.3 mm with the Atacama Large Millimeter/Submillimeter Array (ALMA) and at 7 mm with the Karl G. Jansky Very Large Array (VLA). By modeling the free-free continuum, the ionized outflow is found to be a photoevaporation flow with an electron temperature of 10,000 K and an electron number density of ~1.5x10^7 cm^-3 at the center, launched from a disk of radius of 110 au. H30alpha hydrogen recombination line emission shows strong maser amplification, with G45 being one of very few sources to show such millimeter recombination line masers. The mass of the driving source is estimated to be 30-50 Msun based on the derived ionizing photon rate, or 30-40 Msun based on the H30alpha kinematics. The kinematics of the photoevaporated material is dominated by rotation close to the disk plane, while accelerated to outflowing motion above the disk plane. The mass loss rate of the photoevaporation outflow is estimated to be ~(2-3.5)x10^-5 Msun/yr. We also found hints of a possible jet embedded inside the wide-angle ionized outflow with non-thermal emissions. The possible co-existence of a jet and a massive photoevaporation outflow suggests that, in spite of the strong photoionization feedback, accretion is still on-going., Comment: Accepted to ApJL. 16 pages, 5 figures, 3 appendix figures

The large memory and computation consumption in convolutional neural networks (CNNs) has been one of the main barriers for deploying them on resource-limited systems. To this end, most cheap convolutions (e.g., group convolution, depth-wise convolution, and shift convolution) have recently been used for memory and computation reduction but with the specific architecture designing. Furthermore, it results in a low discriminability of the compressed networks by directly replacing the standard convolution with these cheap ones. In this paper, we propose to use knowledge distillation to improve the performance of the compact student networks with cheap convolutions. In our case, the teacher is a network with the standard convolution, while the student is a simple transformation of the teacher architecture without complicated redesigning. In particular, we propose a novel online distillation method, which online constructs the teacher network without pre-training and conducts mutual learning between the teacher and student network, to improve the performance of the student model. Extensive experiments demonstrate that the proposed approach achieves superior performance to simultaneously reduce memory and computation overhead of cutting-edge CNNs on different datasets, including CIFAR-10/100 and ImageNet ILSVRC 2012, compared to the state-of-the-art CNN compression and acceleration methods. The codes are publicly available at https://github.com/EthanZhangYC/OD-cheap-convolution.

During star formation, the accretion disk drives fast MHD winds which usually contain two components, a collimated jet and a radially distributed wide-angle wind. These winds entrain the surrounding ambient gas producing molecular outflows. We report recent observation of 12CO (2-1) emission of the HH 46/47 molecular outflow by the Atacama Large Millimeter/sub-millimeter Array, in which we identify multiple wide-angle outflowing shell structures in both the blue and red-shifted outflow lobes. These shells are highly coherent in position-position-velocity space, extending to >40-50 km/s in velocity and 10^4 au in space with well defined morphology and kinematics. We suggest these outflowing shells are the result of the entrainment of ambient gas by a series of outbursts from an intermittent wide-angle wind. Episodic outbursts in collimated jets are commonly observed, yet detection of a similar behavior in wide-angle winds has been elusive. Here we show clear evidence that the wide-angle component of the HH 46/47 protostellar outflows experiences similar variability seen in the collimated component., Comment: 13 pages, 5 figures, accepted to ApJ

We conduct a census of the high-mass protostellar population of the $\sim70,000\:M_\odot$ Infrared Dark Cloud (IRDC) G028.37+00.07, identifying 35 sources based on their $70\:\mu$m emission, as reported in the {\it Herschel} Hi-GAL catalog of Molinari et al. (2016). We perform aperture photometry to construct spectral energy distributions (SEDs), which are then fit with the massive protostar models of Zhang & Tan (2018). We find that the sources span a range of isotropic luminosities from $\sim$20 to 4,500$\:L_\odot$. The most luminous sources are predicted to have current protostellar masses of $m_{*}\sim10\:M_\odot$ forming from cores of mass $M_{c}\sim40$ to $400\:M_\odot$. The least luminous sources in our sample are predicted to be protostars with masses as low as $\sim 0.5\:M_\odot$ forming from cores with $M_{c}\sim10\:M_\odot$, which are the minimum values explored in the protostellar model grid. The detected protostellar population has a total estimated protostellar mass of $M_{*}\sim 100\:M_\odot$. Allowing for completeness corrections, which are constrained by comparison with an ALMA study in part of the cloud, we estimate a star formation efficiency per free-fall time of $\sim3\%$ in the IRDC. Finally, analyzing the spatial distribution of the sources, we find relatively low degrees of central concentration of the protostars. The protostars, including the most massive ones, do not appear to be especially centrally concentrated in the protocluster as defined by the IRDC boundary., Comment: 18 pages, accepted to ApJ, comments welcome

Neural coding is one of the central questions in systems neuroscience for understanding how the brain processes stimulus from the environment, moreover, it is also a cornerstone for designing algorithms of brain-machine interface, where decoding incoming stimulus is highly demanded for better performance of physical devices. Traditionally researchers have focused on functional magnetic resonance imaging (fMRI) data as the neural signals of interest for decoding visual scenes. However, our visual perception operates in a fast time scale of millisecond in terms of an event termed neural spike. There are few studies of decoding by using spikes. Here we fulfill this aim by developing a novel decoding framework based on deep neural networks, named spike-image decoder (SID), for reconstructing natural visual scenes, including static images and dynamic videos, from experimentally recorded spikes of a population of retinal ganglion cells. The SID is an end-to-end decoder with one end as neural spikes and the other end as images, which can be trained directly such that visual scenes are reconstructed from spikes in a highly accurate fashion. Our SID also outperforms on the reconstruction of visual stimulus compared to existing fMRI decoding models. In addition, with the aid of a spike encoder, we show that SID can be generalized to arbitrary visual scenes by using the image datasets of MNIST, CIFAR10, and CIFAR100. Furthermore, with a pre-trained SID, one can decode any dynamic videos to achieve real-time encoding and decoding of visual scenes by spikes. Altogether, our results shed new light on neuromorphic computing for artificial visual systems, such as event-based visual cameras and visual neuroprostheses., Comment: 35 pages, 10 figures

Almost all massive stars have bound stellar companions, existing in binaries or higher-order multiples. While binarity is theorized to be an essential feature of how massive stars form, essentially all information about such properties is derived from observations of already formed stars, whose orbital properties may have evolved since birth. Little is known about binarity during formation stages. Here we report high angular resolution observations of 1.3 mm continuum and H30alpha recombination line emission, which reveal a massive protobinary with apparent separation of 180 au at the center of the massive star-forming region IRAS07299-1651. From the line-of-sight velocity difference of 9.5 km/s of the two protostars, the binary is estimated to have a minimum total mass of 18 solar masses, consistent with several other metrics, and maximum period of 570 years, assuming a circular orbit. The H30alpha line from the primary protostar shows kinematics consistent with rotation along a ring of radius of 12 au. The observations indicate that disk fragmentation at several hundred au may have formed the binary, and much smaller disks are feeding the individual protostars., Comment: Published in Nature Astronomy. This is author's version. Full article is available here (https://rdcu.be/brENk). 47 pages, 10 figures, including methods and supplementary information

We present multi-wavelength images observed with SOFIA-FORCAST from $\sim$10 to 40 $\mu$m of seven high luminosity massive protostars, as part of the SOFIA Massive (SOMA) Star Formation Survey. Source morphologies at these wavelengths appear to be influenced by outflow cavities and extinction from dense gas surrounding the protostars. Using these images, we build spectral energy distributions (SEDs) of the protostars, also including archival data from Spitzer, Herschel and other facilities. Radiative transfer (RT) models of Zhang & Tan (2018), based on Turbulent Core Accretion theory, are then fit to the SEDs to estimate key properties of the protostars. Considering the best five models fit to each source, the protostars have masses $m_{*} \sim 12-64 \: M_{\odot}$ accreting at rates of $\dot{m}_{*} \sim 10^{-4}-10^{-3} \: M_{\odot} \: \rm yr^{-1}$ inside cores of initial masses $M_{c} \sim 100-500 \: M_{\odot}$ embedded in clumps with mass surface densities $\Sigma_{\rm cl} \sim 0.1-3 \: \rm g \: cm^{-2}$ and span a luminosity range of $10^{4} -10^{6} \: L_{\odot}$. Compared with the first eight protostars in Paper I, the sources analyzed here are more luminous, and thus likely to be more massive protostars. They are often in a clustered environment or have a companion protostar relatively nearby. From the range of parameter space of the models, we do not see any evidence that $\Sigma_{\rm cl}$ needs to be high to form these massive stars. For most sources the RT models provide reasonable fits to the SEDs, though the cold clump material often influences the long wavelength fitting. However, for sources in very clustered environments, the model SEDs may not be such a good description of the data, indicating potential limitations of the models for these regions., Comment: 30 pages, 19 figures, Accepted for publication in ApJ

This paper studies distributed estimation and inference for a general statistical problem with a convex loss that could be non-differentiable. For the purpose of efficient computation, we restrict ourselves to stochastic first-order optimization, which enjoys low per-iteration complexity. To motivate the proposed method, we first investigate the theoretical properties of a straightforward Divide-and-Conquer Stochastic Gradient Descent (DC-SGD) approach. Our theory shows that there is a restriction on the number of machines and this restriction becomes more stringent when the dimension $p$ is large. To overcome this limitation, this paper proposes a new multi-round distributed estimation procedure that approximates the Newton step only using stochastic subgradient. The key component in our method is the proposal of a computationally efficient estimator of $\Sigma^{-1} w$, where $\Sigma$ is the population Hessian matrix and $w$ is any given vector. Instead of estimating $\Sigma$ (or $\Sigma^{-1}$) that usually requires the second-order differentiability of the loss, the proposed First-Order Newton-type Estimator (FONE) directly estimates the vector of interest $\Sigma^{-1} w$ as a whole and is applicable to non-differentiable losses. Our estimator also facilitates the inference for the empirical risk minimizer. It turns out that the key term in the limiting covariance has the form of $\Sigma^{-1} w$, which can be estimated by FONE., Comment: 60 pages

We report molecular line observations of the massive protostellar source G339.88-1.26 with the Atacama Large Millimeter/Submillimeter Array. The observations reveal a highly collimated SiO jet extending from the 1.3 mm continuum source, which connects to a slightly wider but still highly collimated CO outflow. Rotational features perpendicular to the outflow axis are detected in various molecular emissions, including SiO, SO2, H2S, CH3OH, and H2CO emissions. Based on their spatial distributions and kinematics, we find that they trace different parts of the envelope-disk system. The SiO emission traces the disk and inner envelope in addition to the jet. The CH3OH and H2CO emissions mostly trace the infalling-rotating envelope, and are enhanced around the transition region between envelope and disk, i.e., the centrifugal barrier. The SO2 and H2S emissions are enhanced around the centrifugal barrier, and also trace the outer part of the disk. Envelope kinematics are consistent with rotating-infalling motion, while those of the disk are consistent with Keplerian rotation. The radius and velocity of the centrifugal barrier are estimated to be about 530 au and 6 km/s, leading to a central mass of about 11 solar masses, consistent with estimates based on spectral energy distribution fitting. These results indicate that an ordered transition from an infalling-rotating envelope to a Keplerian disk through a centrifugal barrier, accompanied by changes of types of molecular line emissions, is a valid description of this massive protostellar source. This implies that at least some massive stars form in a similar way as low-mass stars via Core Accretion., Comment: Accepted to ApJ

Deep convolutional neural networks (CNNs) have demonstrated impressive performance on many visual tasks. Recently, they became useful models for the visual system in neuroscience. However, it is still not clear what are learned by CNNs in terms of neuronal circuits. When a deep CNN with many layers is used for the visual system, it is not easy to compare the structure components of CNNs with possible neuroscience underpinnings due to highly complex circuits from the retina to higher visual cortex. Here we address this issue by focusing on single retinal ganglion cells with biophysical models and recording data from animals. By training CNNs with white noise images to predict neuronal responses, we found that fine structures of the retinal receptive field can be revealed. Specifically, convolutional filters learned are resembling biological components of the retinal circuit. This suggests that a CNN learning from one single retinal cell reveals a minimal neural network carried out in this cell. Furthermore, when CNNs learned from different cells are transferred between cells, there is a diversity of transfer learning performance, which indicates that CNNs are cell-specific. Moreover, when CNNs are transferred between different types of input images, here white noise v.s. natural images, transfer learning shows a good performance, which implies that CNNs indeed capture the full computational ability of a single retinal cell for different inputs. Taken together, these results suggest that CNNs could be used to reveal structure components of neuronal circuits, and provide a powerful model for neural system identification., Comment: updated version

This paper studies the inference problem in quantile regression (QR) for a large sample size $n$ but under a limited memory constraint, where the memory can only store a small batch of data of size $m$. A natural method is the na\"ive divide-and-conquer approach, which splits data into batches of size $m$, computes the local QR estimator for each batch, and then aggregates the estimators via averaging. However, this method only works when $n=o(m^2)$ and is computationally expensive. This paper proposes a computationally efficient method, which only requires an initial QR estimator on a small batch of data and then successively refines the estimator via multiple rounds of aggregations. Theoretically, as long as $n$ grows polynomially in $m$, we establish the asymptotic normality for the obtained estimator and show that our estimator with only a few rounds of aggregations achieves the same efficiency as the QR estimator computed on all the data. Moreover, our result allows the case that the dimensionality $p$ goes to infinity. The proposed method can also be applied to address the QR problem under distributed computing environment (e.g., in a large-scale sensor network) or for real-time streaming data.

We report molecular line observations of the NGC 1333 IRAS 4C outflow in the Perseus Molecular Cloud with the Atacama Large Millimeter/Submillimeter Array. The CCH and CS emission reveal an outflow cavity structure with clear signatures of rotation with respect to the outflow axis. The rotation is detected from about 120 au up to about 1400 au above the envelope/disk mid-plane. As the distance to the central source increases, the rotation velocity of the outflow decreases while the outflow radius increases, which gives a flat specific angular momentum distribution along the outflow. The mean specific angular momentum of the outflow is about 100 au km/s. Based on reasonable assumptions on the outward velocity of the outflow and the protostar mass, we estimate the range of outflow launching radii to be 5-15 au. Such a launching radius rules out that this outflow is launched as an X-wind, but rather, it is more consistent to be a slow disk wind launched from relatively large radii on the disk. The radius of the centrifugal barrier is roughly estimated, and the role of the centrifugal barrier in the outflow launching is discussed., Comment: Accepted to ApJ. 29 pages, 8 figures

Chemical diversity of the gas in low-mass protostellar cores is widely recognized. In order to explore its origin, a survey of chemical composition toward 36 Class 0/I protostars in the Perseus molecular cloud complex, which are selected in an unbiased way under certain physical conditions, has been conducted with IRAM 30 m and NRO 45 m telescope. Multiple lines of C2H, c-C3H2 and CH3OH have been observed to characterize the chemical composition averaged over a 1000 au scale around the protostar. The derived beam-averaged column densities show significant chemical diversity among the sources, where the column density ratios of C2H/CH3OH are spread out by 2 orders of magnitude. From previous studies, the hot corino sources have abundant CH3OH but deficient C2H, their C2H/CH3OH column density ratios being relatively low. In contrast, the warm-carbon-chain chemistry (WCCC) sources are found to reveal the high C2H/CH3OH column density ratios. We find that the majority of the sources have intermediate characters between these two distinct chemistry types. A possible trend is seen between the C2H/CH3OH ratio and the distance of the source from the edge of a molecular cloud. The sources located near cloud edges or in isolated clouds tend to have a high C2H/CH3OH ratio. On the other hand, the sources having a low C2H/CH3OH ratio tend to be located in inner regions of the molecular cloud complex. This result gives an important clue to an understanding of the origin of the chemical diversity of protostellar cores in terms of environmental effects., Comment: Accepted for publication in ApJS

Grid supportive (GS) modes integrated within distributed energy resources (DERs) can improve the frequency response. However, synthesis of GS modes for guaranteed performance is challenging. Moreover, a tool is needed to handle sophisticated specifications from grid codes and protection relays. This paper proposes a model predictive control (MPC)-based mode synthesis methodology, which can accommodate the temporal logic specifications (TLSs). The TLSs allow richer descriptions of control specifications addressing both magnitude and time at the same time. The proposed controller will compute a series of Boolean control signals to synthesize the GS mode of DERs by solving the MPC problem under the normal condition, where the frequency response predicted by a reduced-order model satisfies the defined specifications. Once a sizable disturbance is detected, the pre-calculated signals are applied to the DERs. The proposed synthesis methodology is verified on the full nonlinear model in Simulink. A robust factor is imposed on the specifications to compensate the response mismatch between the reduce-order model and nonlinear model so that the nonlinear response satisfies the required TLS., Comment: 2018 IEEE International Symposium on Power Electronics for Distributed Generation Systems (PEDG)

Frequency performance has been a crucial issue for islanded microgrids. On the one hand, most distributed energy resources (DER) are converter-interfaced and do not inherently respond to frequency variations. On the other hand, current inertia emulation approach cannot provide guaranteed response. In this paper, a model reference control based inertia emulation strategy is proposed for diesel-wind systems. Desired inertia can be precisely emulated through the proposed strategy. A typical frequency response model with parametric inertia is set to be the reference model. A measurement at a specific location delivers the information about the disturbance acting on the diesel-wind system to the reference model. The objective is for the speed of the diesel generator to track the reference so that the desired inertial response is realized. In addition, polytopic parameter uncertainty will be considered. The control strategy is configured in different ways according to different operating points. The parameters of the reference model are scheduled to ensure adequate frequency response under a pre-defined worst case. The controller is implemented in a nonlinear three-phase diesel-wind system fed microgrid using the Simulink software platform. The results show that exact synthetic inertia can be emulated and adequate frequency response is achieved.

We conduct a theoretical study of the formation of massive stars over a wide range of metallicities from 1e-5 to 1Zsun and evaluate the star formation efficiencies (SFEs) from prestellar cloud cores taking into account multiple feedback processes. Unlike for simple spherical accretion, in the case of disk accretion feedback processes do not set upper limits on stellar masses. At solar metallicity, launching of magneto-centrifugally-driven outflows is the dominant feedback process to set SFEs, while radiation pressure, which has been regarded to be pivotal, has only minor contribution even in the formation of over-100Msun stars. Photoevaporation becomes significant in over-20Msun star formation at low metallicities of <1e-2Zsun, where dust absorption of ionizing photons is inefficient. We conclude that if initial prestellar core properties are similar, then massive stars are rarer in extremely metal-poor environments of 1e-5 - 1e-3Zsun. Our results give new insight into the high-mass end of the initial mass function and its potential variation with galactic and cosmological environments., Comment: 13 pages, 9 figures, accepted for publication in The Astrophysical Journal

Inadequate frequency response can arise due to a high penetration of wind turbine generators (WTGs) and requires a frequency support function to be integrated in the WTG. The appropriate design for these controllers to ensure adequate response has not been investigated thoroughly. In this paper, a safety supervisory control (SSC) is proposed to synthesize the supportive modes in WTGs to guarantee performance. The concept, region of safety (ROS), is stated for safe switching synthesis. An optimization formula is proposed to calculate the largest ROS. By assuming a polynomial structure, the problem can be solved by a sum of squares program. A feasible result will generate a polynomial, the zero sublevel set of which represents the ROS and is employed as the safety supervisor. A decentralized communication architecture is proposed for small-scale systems. Moreover, a scheduling loop is suggested so that the supervisor updates its boundary with respect to the renewable penetration level to be robust with respect to variations in system inertia. The proposed controller is first verified on a single-machine three-phase nonlinear microgrid, and then implemented on the IEEE 39-bus system. Both results indicate that the proposed framework and control configuration can guarantee adequate response without excessive conservativeness.

Generation portfolio can be significantly altered due to the deployment of distributed energy resources (DER) in distribution networks and the concept of microgrid. Generally, distribution networks can operate in a more resilient and economic fashion through proper coordination of DER. However, due to the partially uncontrollable and stochastic nature of some DER, the variance of net load of distribution systems increases, which raises the operational cost and complicates operation for transmission companies. This motivates peak shaving and valley filling using energy storage units deployed in distribution systems. This paper aims at theoretical formulation of optimal load variance minimization, where the infinity norm of net load is minimized. Then, the problem is reformulated equivalently as a linear program. A case study is performed with capacity-limited battery energy storage model and the simplified power flow model of a radial distribution network. The influence of capacity limit and deployment location are studied., Comment: 2018 IEEE PES Innovative Smart Grid Technologies Conference (ISGT)

Converter-interfaced power sources (CIPS) are hybrid control systems as they may switch between multiple operating modes. Due to increasing penetration, the hybrid behavior of CIPS, such as, wind turbine generators (WTG), may have significant impact on power system dynamics. In this paper, the frequency dynamics under inertia emulation and primary support from WTG is studied. A mode switching for WTG to ensure adequate frequency response is proposed. The switching instants are determined by our proposed concept of a region of safety (ROS), which is the initial set of safe trajectories. The barrier certificate methodology is employed to derive a new algorithm to obtain and enlarge the ROS for the given desired safe limits and the worst-case disturbance scenarios. Then critical switching instants and a safe recovery procedure are found. In addition, the emulated inertia and load-damping effect is derived in the time frame of inertia and primary frequency response, respectively. The theoretical results under critical cases are consistent with simulations and can be used as guidance for practical control design.

In this paper, a model reference control based inertia emulation strategy is proposed. Desired inertia can be precisely emulated through this control strategy so that guaranteed performance is ensured. A typical frequency response model with parametrical inertia is set to be the reference model. A measurement at a specific location delivers the information of disturbance acting on the diesel-wind system to the reference model. The objective is for the speed of the diesel-wind system to track the reference model. Since active power variation is dominantly governed by mechanical dynamics and modes, only mechanical dynamics and states, i.e., a swing-engine-governor system plus a reduced-order wind turbine generator, are involved in the feedback control design. The controller is implemented in a three-phase diesel-wind system feed microgrid. The results show exact synthetic inertia is emulated, leading to guaranteed performance and safety bounds., Comment: 2017 IEEE PES Innovative Smart Grid Technologies Conference

As a fundamental phenomenon in nature, randomness has a wide range of applications in the fields of science and engineering. Among different types of random number generators (RNG), quantum random number generator (QRNG) is a kind of promising RNG as it can provide provable true random numbers based on the inherent randomness of fundamental quantum processes. Nevertheless, the randomness from a QRNG can be diminished (or even destroyed) if the devices (especially the entropy source devices) are not perfect or ill-characterized. To eliminate the practical security loopholes from the source, source-independent QRNGs, which allow the source to have arbitrary and unknown dimensions, have been introduced and become one of the most important semi-device-independent QRNGs. Herein a method that enables ultra-fast unpredictable quantum random number generation from quadrature fluctuations of quantum optical field without any assumptions on the input states is proposed. Particularly, to estimate a lower bound on the extractable randomness that is independent from side information held by an eavesdropper, a new security analysis framework is established based on the extremality of Gaussian states, which can be easily extended to design and analyze new semi-device-independent continuous variable QRNG protocols. Moreover, the practical imperfections of the QRNG including the effects of excess noise, finite sampling range, finite resolution and asymmetric conjugate quadratures are taken into account and quantitatively analyzed. Finally, the proposed method is experimentally demonstrated to obtain high secure random number generation rates of 15.07 Gbits/s in off-line configuration and can potentially achieve 6 Gbits/s by real-time post-processing., Comment: 24 pages, 16 figures