arXiv subjects

Hua Gao

Publications and source records attributed to Hua Gao.

48 records · Page 3Linked to original sources

arXiv2020

The Carnegie-Irvine Galaxy Survey. IX. Classification of Bulge Types and Statistical Properties of Pseudo Bulges

We study the statistical properties of 320 bulges of disk galaxies in the Carnegie-Irvine Galaxy Survey, using robust structural parameters of galaxies derived from image fitting. We apply the Kormendy relation to classify classical and pseudo bulges and characterize bulge dichotomy with respect to bulge structural properties and physical properties of host galaxies. We confirm previous findings that pseudo bulges on average have smaller Sérsic indices, smaller bulge-to-total ratios, and fainter surface brightnesses when compared with classical bulges. Our sizable sample statistically shows that pseudo bulges are more intrinsically flattened than classical bulges. Pseudo bulges are most frequent (incidence $\gtrsim 80\%$) in late-type spirals (later than Sc). Our measurements support the picture in which pseudo bulges arose from star formation induced by inflowing gas, while classical bulges were born out of violent processes such as mergers and coalescence of clumps. We reveal differences with the literature that warrant attention: (1) the bimodal distribution of Sérsic indices presented by previous studies is not reproduced in our study; (2) classical and pseudo bulges have similar relative bulge sizes; and (3) the pseudo bulge fraction is considerably smaller in early-type disks compared with previous studies based on one-dimensional surface brightness profile fitting. We attribute the above differences to our improved image quality, more robust bulge-to-disk decomposition technique, and different classification criteria applied. Moreover, we find that barred galaxies do not host more pseudo bulges or more prominent pseudo bulges than unbarred galaxies. Various implications of these findings are discussed.

astro-ph.GA↗

arXiv2019

Analyzing Learned Molecular Representations for Property Prediction

Advancements in neural machinery have led to a wide range of algorithmic solutions for molecular property prediction. Two classes of models in particular have yielded promising results: neural networks applied to computed molecular fingerprints or expert-crafted descriptors, and graph convolutional neural networks that construct a learned molecular representation by operating on the graph structure of the molecule. However, recent literature has yet to clearly determine which of these two methods is superior when generalizing to new chemical space. Furthermore, prior research has rarely examined these new models in industry research settings in comparison to existing employed models. In this paper, we benchmark models extensively on 19 public and 16 proprietary industrial datasets spanning a wide variety of chemical endpoints. In addition, we introduce a graph convolutional model that consistently matches or outperforms models using fixed molecular descriptors as well as previous graph neural architectures on both public and proprietary datasets. Our empirical findings indicate that while approaches based on these representations have yet to reach the level of experimental reproducibility, our proposed model nevertheless offers significant improvements over models currently used in industrial workflows.

cs.LG↗

arXiv2019

The Carnegie-Irvine Galaxy Survey. VIII. Demographics of Bulges along the Hubble Sequence

We present multi-component decomposition of high-quality $R$-band images of 312 disk galaxies from the Carnegie-Irvine Galaxy Survey. In addition to bulges and disks, we successfully model nuclei, bars, disk breaks, nuclear/inner lenses, and inner rings. Our modeling strategy treats nuclear rings and nuclear bars as part of the bulge component, while other features such as spiral arms, outer lenses, and outer rings are omitted from the fits because they are not crucial for accurate bulge measurements. The error budget of bulge parameters includes the uncertainties from sky level measurements and model assumptions. Comparison with multi-component decomposition from the \textit{Spitzer} Survey of Stellar Structure in Galaxies reveals broad agreement for the majority of the overlapping galaxies, but for a considerable fraction of galaxies there are significant differences in bulge parameters caused by different strategies in model construction. We confirm that on average bulge prominence decreases from early to late-type disk galaxies, although the large scatter of bulge-to-total ratios in each morphological bin limits the application of Hubble type as an accurate predictor of bulge-to-total ratio. In contrast with previous studies claiming that barred galaxies host weaker bulges, we find that barred and unbarred spiral galaxies have similar bulge prominence.

astro-ph.GA↗

arXiv2018

The Carnegie-Irvine Galaxy Survey. VII. Constraints on the Origin of S0 Galaxies from Their Photometric Structure

Using high-quality optical images from the Carnegie-Irvine Galaxy Survey, we perform multi-component decompositions of S0 galaxies to derive accurate structural parameters to constrain their physical origin. Many S0s do not host prominent bulges. S0 galaxies have a broad distribution of bulge-to-total ratios ($B/T$) and Sérsic indices ($n$), with average values of $B/T=0.34\pm0.15$ and $n=2.62\pm1.02$, qualitatively consistent with the notion that S0s define a parallel sequence with and may have evolved from spiral galaxies. This is further reinforced by the incidence of bars and lenses in S0s, which when compared with the statistics in spirals, supports the idea that lenses are demised bars. However, despite their wide range of prominence, the bulges of S0s form a surprisingly uniform sequence on both the Kormendy and fundamental plane relations. There is no evidence for population dichotomy or other indications of differentiation into classical and pseudo bulges. Most of the S0s reside in the field and in groups; cluster environment is not a necessary condition for S0 production. The properties of S0 bulges show little correlation with environmental indicators, after the dependence of galaxy stellar mass on environment is taken into account. As the bulges of late-type spirals and S0s are intrinsically different, and environmental effects that may account for such evolution appear to be minimal, we conclude that late-type spirals are not plausible progenitors of S0s. The bulges of S0s likely formed at an early epoch, after which secular processes contributed little to their subsequent evolution.

astro-ph.GA↗

arXiv2018

Chemi-net: a graph convolutional network for accurate drug property prediction

Absorption, distribution, metabolism, and excretion (ADME) studies are critical for drug discovery. Conventionally, these tasks, together with other chemical property predictions, rely on domain-specific feature descriptors, or fingerprints. Following the recent success of neural networks, we developed Chemi-Net, a completely data-driven, domain knowledge-free, deep learning method for ADME property prediction. To compare the relative performance of Chemi-Net with Cubist, one of the popular machine learning programs used by Amgen, a large-scale ADME property prediction study was performed on-site at Amgen. The results showed that our deep neural network method improved current methods by a large margin. We foresee that the significantly increased accuracy of ADME prediction seen with Chemi-Net over Cubist will greatly accelerate drug discovery.

cs.LG↗

arXiv2017

An Optimal Strategy for Accurate Bulge-to-disk Decomposition of Disk Galaxies

The development of two-dimensional (2D) bulge-to-disk decomposition techniques has shown their advantages over traditional one-dimensional (1D) techniques, especially for galaxies with non-axisymmetric features. However, the full potential of 2D techniques has yet to be fully exploited. Secondary morphological features in nearby disk galaxies, such as bars, lenses, rings, disk breaks, and spiral arms, are seldom accounted for in 2D image decompositions, even though some image-fitting codes, such as GALFIT, are capable of handling them. We present detailed, 2D multi-model and multi-component decomposition of high-quality $R$-band images of a representative sample of nearby disk galaxies selected from the Carnegie-Irvine Galaxy Survey, using the latest version of GALFIT. The sample consists of five barred and five unbarred galaxies, spanning Hubble types from S0 to Sc. Traditional 1D decomposition is also presented for comparison. In detailed case studies of the 10 galaxies, we successfully model the secondary morphological features. Through a comparison of best-fit parameters obtained from different input surface brightness models, we identify morphological features that significantly impact bulge measurements. We show that nuclear and inner lenses/rings and disk breaks must be properly taken into account to obtain accurate bulge parameters, whereas outer lenses/rings and spiral arms have a negligible effect. We provide an optimal strategy to measure bulge parameters of typical disk galaxies, as well as prescriptions to estimate realistic uncertainties of them, which will benefit subsequent decomposition of a larger galaxy sample.

astro-ph.GA↗

arXiv2017

Data-driven Job Search Engine Using Skills and Company Attribute Filters

According to a report online, more than 200 million unique users search for jobs online every month. This incredibly large and fast growing demand has enticed software giants such as Google and Facebook to enter this space, which was previously dominated by companies such as LinkedIn, Indeed and CareerBuilder. Recently, Google released their "AI-powered Jobs Search Engine", "Google For Jobs" while Facebook released "Facebook Jobs" within their platform. These current job search engines and platforms allow users to search for jobs based on general narrow filters such as job title, date posted, experience level, company and salary. However, they have severely limited filters relating to skill sets such as C++, Python, and Java and company related attributes such as employee size, revenue, technographics and micro-industries. These specialized filters can help applicants and companies connect at a very personalized, relevant and deeper level. In this paper we present a framework that provides an end-to-end "Data-driven Jobs Search Engine". In addition, users can also receive potential contacts of recruiters and senior positions for connection and networking opportunities. The high level implementation of the framework is described as follows: 1) Collect job postings data in the United States, 2) Extract meaningful tokens from the postings data using ETL pipelines, 3) Normalize the data set to link company names to their specific company websites, 4) Extract and ranking the skill sets, 5) Link the company names and websites to their respective company level attributes with the EVERSTRING Company API, 6) Run user-specific search queries on the database to identify relevant job postings and 7) Rank the job search results. This framework offers a highly customizable and highly targeted search experience for end users.

cs.IR↗

arXiv2017

Modeling of hysteresis loop and its applications in ferroelectric materials

In order to understand the physical hysteresis loops clearly, we constructed a novel model, which is combined with the electric field, the temperature, and the stress as one synthetically parameter. This model revealed the shape of hysteresis loop was determined by few variables in ferroelectric materials: the saturation of polarization, the coercive field, the electric susceptibility and the equivalent field. Comparison with experimental results revealed the model can retrace polarization versus electric field and temperature. As a applications of this model, the calculate formula of energy storage efficiency, the electrocaloric effect, and the P(E,T) function have also been included in this article.

cond-mat.mtrl-sci↗

arXiv2017

Time delay signature elimination of chaos in a semiconductor laser by dispersive feedback from a chirped FBG

Time delay signature (TDS) of a semiconductor laser subject to dispersive optical feedback from a chirped fiber Bragg grating (CFBG) is investigated experimentally and numerically. Different from mirror, CFBG provides additional frequency-dependent delay caused by dispersion, and thus induces external-cavity modes with irregular mode separation rather than a fixed separation induced by mirror feedback. Compared with mirror feedback, the CFBG feedback can greatly depress and even eliminate the TDS, although it leads to a similar quasi-period route to chaos with increases of feedback. In experiments, by using a CFBG with dispersion of 2000ps/nm, the TDS is decreased by 90% to about 0.04 compared with mirror feedback. Furthermore, both numerical and experimental results show that the TDS evolution is quite different: the TDS decreases more quickly down to a lower plateau (even background noise level of autocorrelation function) and never rises again. This evolution tendency is also different from that of FBG feedback, of which the TDS first decreases to a minimal value and then increases again as feedback strength increases. In addition, the CFBG feedback has no filtering effects and does not require amplification for feedback light.

physics.optics↗

arXiv2016

Mathematical modeling of hysteresis loops in ferroelectric materials

In published papers, the Gibbs free energy of ferroelectric materials has usually been quantified by the retention of 6th or 8th order polarization terms. In this paper, a newly analytical model of Gibbs free energy, thereout, a new model of polarization-electric field hysteresis loops in ferroelectric materials has been derived mathematically. As a model validation, four patterns of polarization-electric field hysteresis loops of ferroelectric materials have been depicted by using the model. The calculated results indicated that the self-similar model can characterize the various patterns of hysteresis loops in ferroelectric materials through adjusting the external excitation or the synthetically parameter (e.g., electric, temperature, and stress, etc.) employed in the model.

cond-mat.mtrl-sci↗

arXiv2015

Validation of LAMOST Stellar Parameters with the PASTEL Catalog

Recently the Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST) published its first data release (DR1), which is ranked as the largest stellar spectra dataset in the world so far. We combine the PASTEL catalog and SIMBAD radial velocities as a testing standard to validate the DR1 stellar parameters (effective temperature $T_{\mathrm{eff}}$, surface gravity $\log g$, metallicity $\mathrm{[Fe/H]}$ and radial velocity $V_{\mathrm{r}}$). Through cross-identification of the DR1 catalogs and the PASTEL catalog, we obtain a preliminary sample of 422 stars. After removal of stellar parameter measurements from problematic spectra and applying effective temperature constraints to the sample, we compare the stellar parameters from DR1 with those from PASTEL and SIMBAD to prove that the DR1 results are reliable in restricted $T_{\mathrm{eff}}$ ranges. We derive standard deviations of 110 K, 0.19 dex, 0.11 dex and 4.91 $\mathrm{km\,s^{-1}}$ , for $T_{\mathrm{eff}}$, $\log g$, $[\mathrm{Fe/H}]$ when $T_{\mathrm{eff}}<8000\,\mathrm{K}$, and for $V_{\mathrm{r}}$ when $T_{\mathrm{eff}}<10000\,\mathrm{K}$, respectively. Systematic errors are negligible except for that of $V_{\mathrm{r}}$. Besides, metallicities in DR1 are systematically higher than those in PASTEL, in the range of PASTEL $[\mathrm{Fe/H}]<-1.5$.

astro-ph.SR↗

arXiv2015

Three-Dimensional Phase Field Simulations of Hysteresis and Butterfly Loops by Finite Volume Method

Three-dimensional (3D) simulations of ferroelectric hysteresis and butterfly loops have been carried out based on solving the time dependent Ginsburg-Landau equations by using a finite volume method. The influence of externally mechanical loadings with a tensile strain and a compressive strain on the hysteresis and butterfly loops has been studied numerically. The 3D ferroelectric domain formation and its evolution have also been presented in the paper. The simulation results successfully reveal the macroscopically nonlinear response to the applied stresses and electric field.

cond-mat.mtrl-sci↗