arXiv subjects

Hongyu Zhu

Publications and source records attributed to Hongyu Zhu.

49 records · Page 3Linked to original sources

arXiv2021

Modeling of cellular response after FLASH irradiation: a quantitative analysis based on the radiolytic oxygen depletion hypothesis

Purpose: Recent studies suggest ultra-high dose rate (FLASH) irradiation can spare normal tissues from radiotoxicity, while efficiently controlling the tumor, and this is known as the FLASH effect. This study performed theoretical analyses about the impact of radiolytic oxygen depletion (ROD) on the cellular responses after FLASH irradiation. Methods: Monte Carlo simulation was used to model the ROD process, determine the DNA damage, and calculate the amount of oxygen depleted (LROD) during FLASH exposure. A mathematical model was applied to analyze oxygen tension (pO2) distribution in human tissues and the recovery of pO2 after FLASH irradiation. DNA damage and cell survival fractions (SFs) after FLASH irradiation were calculated. The impact of initial cellular pO2, FLASH pulse number, pulse interval, and radiation quality of the source particles on ROD and subsequent cellular responses were systematically evaluated. Results: The simulated electron LROD range was 0.38-0.43 μM/Gy when pO2 ranged from 7.5-160 mmHg. The calculated DNA damage and SFs show that radioprotective effect is only evident in cells with a lower pO2. Different irradiation setups alter the cellular responses by modifying the pO2. Single pulse delivery or multi-pulse delivery with pulse intervals shorter than 10-50 ms resulted in fewer DNA damages and higher SFs. Source particles with a low radiation quality have a higher capacity to deplete oxygen, and thus, lead to a more conspicuous radioprotective effect. Conclusions: The FLASH radioprotective effect due to ROD may only be observed in cells with a low pO2. Single pulse delivery or multi-pulse delivery with short pulse intervals are suggested for FLASH irradiation to avoid oxygen tension recovery during pulse intervals. Source particles with low radiation quality are preferred for their conspicuous radioprotective effects.

physics.med-ph↗

arXiv2021

The Impact of COVID-19 on Urban Energy Consumption of the Commercial Tourism City

In 2020, the COVID-19 pandemic spreads all over the world. In order to alleviate the spread of the epidemic, various blockade policies have been implemented in many areas. In order to formulate a better epidemic prevention policy for urban energy consumption of the commercial tourism cities, this paper first analyses the energy characteristics of Macao during the epidemic period from two aspects, based on the energy consumption data of Macao. On this basis, the power consumption characteristics of commercial tourism cities during the epidemic were analyzed. Then, this paper provides analysis of the characteristics of the energy consumption in different fields of commercial tourism cities from the aspects of hotel, transportation, tourism culture and public utilities. Finally, a detailed analysis of the energy consumption characteristics of commercial tourism cities represented by Macao during the epidemic period is provided, by comparing with some typical countries.

physics.soc-ph↗

arXiv2021

Dual MINE-based Neural Secure Communications under Gaussian Wiretap Channel

Recently, some researches are devoted to the topic of end-to-end learning a physical layer secure communication system based on autoencoder under Gaussian wiretap channel. However, in those works, the reliability and security of the encoder model were learned through necessary decoding outputs of not only legitimate receiver but also the eavesdropper. In fact, the assumption of known eavesdropper's decoder or its output is not practical. To address this issue, in this paper we propose a dual mutual information neural estimation (MINE) based neural secure communications model. The security constraints of this method is constructed only with the input and output signal samples of the legal and eavesdropper channels and benefit that training the encoder is completely independent of the decoder. Moreover, since the design of secure coding does not rely on the eavesdropper's decoding results, the security performance would not be affected by the eavesdropper's decoding means. Numerical results show that the performance of our model is guaranteed whether the eavesdropper learns the decoder himself or uses the legal decoder.

cs.IT↗

arXiv2020

Daydream: Accurately Estimating the Efficacy of Optimizations for DNN Training

Modern deep neural network (DNN) training jobs use complex and heterogeneous software/hardware stacks. The efficacy of software-level optimizations can vary significantly when used in different deployment configurations. It is onerous and error-prone for ML practitioners and system developers to implement each optimization separately, and determine which ones will improve performance in their own configurations. Unfortunately, existing profiling tools do not aim to answer predictive questions such as "How will optimization X affect the performance of my model?". We address this critical limitation, and proposes a new profiling tool, Daydream, to help programmers efficiently explore the efficacy of DNN optimizations. Daydream models DNN execution with a fine-grained dependency graph based on low-level traces collected by CUPTI, and predicts runtime by simulating execution based on the dependency graph. Daydream maps the low-level traces using DNN domain-specific knowledge, and introduces a set of graph-transformation primitives that can easily model a wide variety of optimizations. We show that Daydream is able to model most mainstream DNN optimization techniques, and accurately predict the efficacy of optimizations that will result in significant performance improvements.

cs.DC↗

arXiv2019

Memory In Memory: A Predictive Neural Network for Learning Higher-Order Non-Stationarity from Spatiotemporal Dynamics

Natural spatiotemporal processes can be highly non-stationary in many ways, e.g. the low-level non-stationarity such as spatial correlations or temporal dependencies of local pixel values; and the high-level variations such as the accumulation, deformation or dissipation of radar echoes in precipitation forecasting. From Cramer's Decomposition, any non-stationary process can be decomposed into deterministic, time-variant polynomials, plus a zero-mean stochastic term. By applying differencing operations appropriately, we may turn time-variant polynomials into a constant, making the deterministic component predictable. However, most previous recurrent neural networks for spatiotemporal prediction do not use the differential signals effectively, and their relatively simple state transition functions prevent them from learning too complicated variations in spacetime. We propose the Memory In Memory (MIM) networks and corresponding recurrent blocks for this purpose. The MIM blocks exploit the differential signals between adjacent recurrent states to model the non-stationary and approximately stationary properties in spatiotemporal dynamics with two cascaded, self-renewed memory modules. By stacking multiple MIM blocks, we could potentially handle higher-order non-stationarity. The MIM networks achieve the state-of-the-art results on four spatiotemporal prediction tasks across both synthetic and real-world datasets. We believe that the general idea of this work can be potentially applied to other time-series forecasting tasks.

cs.LG↗

arXiv2019

Gravitational redshift profiles of MaNGA BCGs

The gravitational potential well of an $M>10^{13}$ $M_\odot$ galaxy will lead to a gravitational redshift differential of order 1 km/s between the galaxy core and its outskirts. Current surveys of massive galaxies with spatially resolved spectroscopy have reached a size which makes feasible attempts to detect gravitational redshifts within galaxies. We use spectra from the Mapping Nearby Galaxies at Apache Point Observatory (MaNGA) experiment to attempt a measurement of the averaged stellar redshift profile of large elliptical galaxies. We find that systematic effects (possibly related to charge transfer or wavelength calibration accuracy) make the standard MaNGA data pipeline unsuitable for measuring the relevant sub km/s wavelength shifts. We therefore develop a cross-correlation technique to mitigate these effects, but caution that we are working beyond the design accuracy of the MaNGA experiment. With a sample of $272$ galaxies in halos with $\log (M/M_\odot)>13$, we attempt a measurement of the gravitational redshift profile, achieving $1 σ$ errors of size $\sim 0.5$ km/s, but are unable to make a significant detection of the signal. Even without a detection, our measurement can be used to limit the dark matter mass within the half light radius of elliptical galaxies to $1.2$ times the stellar mass, at the 68% confidence level. We also perform weighting scheme tests and split sample tests, and address target selection issues and other relativistic effects, including the transverse Doppler effect and relativistic beaming of stars. Future detections could lead to new constraints on the galaxy mass distribution that are different from kinematic and lensing determinations and open a window on galaxy properties and tests of gravity.

astro-ph.GA↗

arXiv2019

Effects of baryons on the gravitational redshift profile of ΛCDM halos

Gravitational redshifts and other relativistic effects are beginning to be studied in the context of galaxy clustering. Distortions consistent with those expected in General Relativity have been measured in galaxy cluster redshift profiles by Wojtak et al. and others and in the the cross-correlation function of galaxy populations by Alam et al. On scales below ~20 Mpc/$h$ simulations have shown that gravitational redshift dominates over other effects. However, this signal is related to the shape and depth of gravitational potentials, and therefore the matter density in galaxies and galaxy clusters that is responsible for them. We investigate the effects of baryonic physics on the gravitational redshift profiles of massive (group and cluster-sized) halos. We compare the profiles of different components in halos taken from the MassiveBlack-II cosmological hydrodynamic simulation and a dark matter-only version of the same simulation. We find that inclusion of baryons, cooling, star formation and feedback significantly alters the relevant inner density profiles. These baryonic effects lead to overall increases in both gravitational redshifts and the transverse relativistic Doppler effects by up to ~50% for group size halos. We show how modified Navarro Frenk White halo profiles can be used to parametrize these differences, and provide relevant halo profile fits.

astro-ph.CO↗

arXiv2019

A Discussion on the Algorithm Design of Electrical Impedance Tomography for Biomedical Applications

In this paper, we present a discussion on the algorithms design of Electrical Impedance Tomography (EIT) for biomedical applications. Based on the Maxwell differential equations and the derived the finite element(FE) linear equations, we first investigate the possibility to estimate the matrix that contains the impedance values based on Singular Value Decomposition(SVD) approximations. Secondly based on the biomedical properties we further explore the possibility to recover the impedance values uniquely by injecting various different types of currents with multi-frequency. Injecting various types of multi-frequency currents lead to a set of different measured voltages configurations, thus enhance the possibility of uniquely recovering the impedance values in a stable way under the assumption that the biological cells respond to the different type of injecting currents in a different way.

eess.SP↗

arXiv2018

TBD: Benchmarking and Analyzing Deep Neural Network Training

The recent popularity of deep neural networks (DNNs) has generated a lot of research interest in performing DNN-related computation efficiently. However, the primary focus is usually very narrow and limited to (i) inference -- i.e. how to efficiently execute already trained models and (ii) image classification networks as the primary benchmark for evaluation. Our primary goal in this work is to break this myopic view by (i) proposing a new benchmark for DNN training, called TBD (TBD is short for Training Benchmark for DNNs), that uses a representative set of DNN models that cover a wide range of machine learning applications: image classification, machine translation, speech recognition, object detection, adversarial networks, reinforcement learning, and (ii) by performing an extensive performance analysis of training these different applications on three major deep learning frameworks (TensorFlow, MXNet, CNTK) across different hardware configurations (single-GPU, multi-GPU, and multi-machine). TBD currently covers six major application domains and eight different state-of-the-art models. We present a new toolchain for performance analysis for these models that combines the targeted usage of existing performance analysis tools, careful selection of new and existing metrics and methodologies to analyze the results, and utilization of domain specific characteristics of DNN training. We also build a new set of tools for memory profiling in all three major frameworks; much needed tools that can finally shed some light on precisely how much memory is consumed by different data structures (weights, activations, gradients, workspace) in DNN training. By using our tools and methodologies, we make several important observations and recommendations on where the future research and optimization of DNN training should be focused.

cs.LG↗

arXiv2017

Relativistic asymmetries in the galaxy cross-correlation function

We study the asymmetry in the two-point cross-correlation function of two populations of galaxies focusing in particular on the relativistic effects that include the gravitational redshift. We derive the cross-correlation function on small and large scales using two different approaches: General Relativistic and Newtonian perturbation theory. Following recent work by Bonvin et al., Gaztanaga et al. and Croft, we calculate the dipole and the shell estimator with the two procedures and we compare our results. We find that while General Relativistic Perturbation Theory (GRPT) is able to make predictions of relativistic effects on very large, obviously linear scales (r > 50 Mpc/h), the presence of non-linearities physically occurring on much smaller scales (down to those describing galactic potential wells) can strongly affect the asymmetry estimators. These can lead to cancellations of the relativistic terms, and sign changes in the estimators on scales up to r ~ 50 Mpc/h. On the other hand, with an appropriate non-linear gravitational potential, the results obtained using Newtonian theory can successfully describe the asymmetry on smaller, non-linear scales (r < 20 Mpc/h) where gravitational redshift is the dominant term. On larger scales the asymmetry is much smaller in magnitude, and measurement is not within reach of current observations. This is in agreement with the observational results obtained by Gaztnaga et al. and the first detection of relativistic effects (on (r < 20 Mpc/h) scales) by Alam et al.

astro-ph.CO↗

arXiv2017

Relativistic distortions in the large-scale clustering of SDSS-III BOSS CMASS galaxies

General relativistic effects have long been predicted to subtly influence the observed large-scale structure of the universe. The current generation of galaxy redshift surveys have reached a size where detection of such effects is becoming feasible. In this paper, we report the first detection of the redshift asymmetry from the cross-correlation function of two galaxy populations which is consistent with relativistic effects. The dataset is taken from the Sloan Digital Sky Survey DR12 CMASS galaxy sample, and we detect the asymmetry at the $2.7σ$ level by applying a shell-averaged estimator to the cross-correlation function. Our measurement dominates at scales around $10$ h$^{-1}$Mpc, larger than those over which the gravitational redshift profile has been recently measured in galaxy clusters, but smaller than scales for which linear perturbation theory is likely to be accurate. The detection significance varies by 0.5$σ$ with the details of our measurement and tests for systematic effects. We have also devised two null tests to check for various survey systematics and show that both results are consistent with the null hypothesis. We measure the dipole moment of the cross-correlation function, and from this the asymmetry is also detected, at the $2.8 σ$ level. The amplitude and scale-dependence of the clustering asymmetries are approximately consistent with the expectations of General Relativity and a biased galaxy population, within large uncertainties. We explore theoretical predictions using numerical simulations in a companion paper.

astro-ph.CO↗

arXiv2017

N-body simulations of gravitational redshifts and other relativistic distortions of galaxy clustering

Large redshift surveys of galaxies and clusters are providing the first opportunities to search for distortions in the observed pattern of large-scale structure due to such effects as gravitational redshift. We focus on non-linear scales and apply a quasi-Newtonian approach using N-body simulations to predict the small asymmetries in the cross-correlation function of two galaxy different populations. Following recent work by Bonvin et al., Zhao and Peacock and Kaiser on galaxy clusters, we include effects which enter at the same order as gravitational redshift: the transverse Doppler effect, light-cone effects, relativistic beaming, luminosity distance perturbation and wide-angle effects. We find that all these effects cause asymmetries in the cross-correlation functions. Quantifying these asymmetries, we find that the total effect is dominated by the gravitational redshift and luminosity distance perturbation at small and large scales, respectively. By adding additional subresolution modelling of galaxy structure to the large-scale structure information, we find that the signal is significantly increased, indicating that structure on the smallest scales is important and should be included. We report on comparison of our simulation results with measurements from the SDSS/BOSS galaxy redshift survey in a companion paper.

astro-ph.CO↗

arXiv2017

Relativistic Effects on Galaxy Redshift Samples due to Target Selection

In a galaxy redshift survey the objects to be targeted for spectra are selected from a photometrically observed sample. The observed magnitudes and colours of galaxies in this parent sample will be affected by their peculiar velocities, through relativistic Doppler and relativistic beaming effects. In this paper we compute the resulting expected changes in galaxy photometry. The magnitudes of the relativistic effects are a function of redshift, stellar mass, galaxy velocity and velocity direction. We focus on the CMASS sample from the Sloan Digital Sky Survey (SDSS), Baryon Oscillation Spectroscopic Survey (BOSS), which is selected on the basis of colour and magnitude. We find that 0.10\% of the sample ($\sim 585$ galaxies) has been scattered into the targeted region of colour-magnitude space by relativistic effects, and conversely 0.09\% of the sample ($\sim 532$ galaxies) has been scattered out. Observational consequences of these effects include an asymmetry in clustering statistics, which we explore in a companion paper. Here we compute a set of weights which can be used to remove the effect of modulations introduced into the density field inferred from a galaxy sample. We conclude by investigating the possible effects of these relativistic modulation on large scale clustering of the galaxy sample.

astro-ph.CO↗