Searcharxiv⌕ Search

arXiv subjects

Yu Feng

Publications and source records attributed to Yu Feng.

At least 217 records · Page 12Linked to original sources
arXiv2019

Visualization by Example

While visualizations play a crucial role in gaining insights from data, generating useful visualizations from a complex dataset is far from an easy task. Besides understanding the functionality provided by existing visualization libraries, generating the desired visualization also requires reshaping and aggregating the underlying data as well as composing different visual elements to achieve the intended visual narrative. This paper aims to simplify visualization tasks by automatically synthesizing the required program from simple visual sketches provided by the user. Specifically, given an input data set and a visual sketch that demonstrates how to visualize a very small subset of this data, our technique automatically generates a program that can be used to visualize the entire data set. Automating visualization poses several challenges. First, because many visualization tasks require data wrangling in addition to generating plots, we need to decompose the end-to-end synthesis task into two separate sub-problems. Second, because the intermediate specification that results from the decomposition is necessarily imprecise, this makes the data wrangling task particularly challenging in our context. In this paper, we address these problems by developing a new compositional visualization-by-example technique that (a) decomposes the end-to-end task into two different synthesis problems over different DSLs and (b) leverages bi-directional program analysis to deal with the complexity that arises from having an imprecise intermediate specification. We implemented our visualization-by-example algorithm and evaluate it on 83 visualization tasks collected from on-line forums and tutorials. Viser can solve 84% of these benchmarks within a 600 second time limit, and, for those tasks that can be solved, the desired visualization is among the top-5 generated by Viser in 70% of the cases.

cs.PL↗
arXiv2019

Intensity mapping with neutral hydrogen and the Hidden Valley simulations

This paper introduces the Hidden Valley simulations, a set of trillion-particle N-body simulations in gigaparsec volumes aimed at intensity mapping science. We present details of the simulations and their convergence, then specialize to the study of 21-cm fluctuations between redshifts 2 and 6. Neutral hydrogen is assigned to halos using three prescriptions, and we investigate the clustering in real and redshift-space at the 2-point level. In common with earlier work we find the bias of HI increases from near 2 at z = 2 to 4 at z = 6, becoming more scale dependent at high z. The level of scale-dependence and decorrelation with the matter field are as predicted by perturbation theory. Due to the low mass of the hosting halos, the impact of fingers of god is small on the range relevant for proposed 21-cm instruments. We show that baryon acoustic oscillations and redshift-space distortions could be well measured by such instruments. Taking advantage of the large simulation volume, we assess the impact of fluctuations in the ultraviolet background, which change HI clustering primarily at large scales.

astro-ph.CO↗
arXiv2019

PES: Proactive Event Scheduling for Responsive and Energy-Efficient Mobile Web Computing

Web applications are gradually shifting toward resource-constrained mobile devices. As a result, the Web runtime system must simultaneously address two challenges: responsiveness and energy-efficiency. Conventional Web runtime systems fall short due to their reactive nature: they react to a user event only after it is triggered. The reactive strategy leads to local optimizations that schedule event executions one at a time, missing global optimization opportunities. This paper proposes Proactive Event Scheduling (PES). The key idea of PES is to proactively anticipate future events and thereby globally coordinate scheduling decisions across events. Specifically, PES predicts events that are likely to happen in the near future using a combination of statistical inference and application code analysis. PES then speculatively executes future events ahead of time in a way that satisfies the QoS constraints of all the events while minimizing the global energy consumption. Fundamentally, PES unlocks more optimization opportunities by enlarging the scheduling window, which enables coordination across both outstanding events and predicted events. Hardware measurements show that PES reduces the QoS violation and energy consumption by 61.2% and 26.5%, respectively, over the Android default Interactive CPU governor. It also reduces the QoS violation and energy consumption by 63.1% and 17.9%, respectively, compared to EBS, a state-of-the-art reactive scheduler.

cs.DC↗
arXiv2019

ASV: Accelerated Stereo Vision System

Estimating depth from stereo vision cameras, i.e., "depth from stereo", is critical to emerging intelligent applications deployed in energy- and performance-constrained devices, such as augmented reality headsets and mobile autonomous robots. While existing stereo vision systems make trade-offs between accuracy, performance and energy-efficiency, we describe ASV, an accelerated stereo vision system that simultaneously improves both performance and energy-efficiency while achieving high accuracy. The key to ASV is to exploit unique characteristics inherent to stereo vision, and apply stereo-specific optimizations, both algorithmically and computationally. We make two contributions. Firstly, we propose a new stereo algorithm, invariant-based stereo matching (ISM), that achieves significant speedup while retaining high accuracy. The algorithm combines classic "hand-crafted" stereo algorithms with recent developments in Deep Neural Networks (DNNs), by leveraging the correspondence invariant unique to stereo vision systems. Secondly, we observe that the bottleneck of the ISM algorithm is the DNN inference, and in particular the deconvolution operations that introduce massive compute-inefficiencies. We propose a set of software optimizations that mitigate these inefficiencies. We show that with less than 0.5% hardware area overhead, these algorithmic and computational optimizations can be effectively integrated within a conventional DNN accelerator. Overall, ASV achieves 5x speedup and 85% energy saving with 0.02% accuracy loss compared to today DNN-based stereo vision systems.

cs.CV↗
arXiv2019

High mass and halo resolution from fast low resolution simulations

Generating mocks for future sky surveys requires large volumes and high resolutions, which is computationally expensive even for fast simulations. In this work we try to develop numerical schemes to calibrate various halo and matter statistics in fast low resolution simulations compared to high resolution N-body and hydrodynamic simulations. For the halos, we improve the initial condition accuracy and develop a halo finder "relaxed-FOF", where we allow different linking length for different halo mass and velocity dispersions. We show that our relaxed-FoF halo finder improves the common statistics, such as halo bias, halo mass function, halo auto power spectrum in real space and in redshift space, cross correlation coefficient with the reference halo catalog, and halo-matter cross power spectrum. We also incorporate the potential gradient descent (PGD) method into fast simulations to improve the matter distribution at nonlinear scale. By building a lightcone output, we show that the PGD method significantly improves the weak lensing convergence tomographic power spectrum. With these improvements FastPM is comparable to the high resolution full N-body simulation of the same mass resolution, with two orders of magnitude fewer time steps. These techniques can be used to improve the halo and matter statistics of FastPM simulations for mock catalogs of future surveys such as DESI and LSST.

astro-ph.CO↗
arXiv2019

Learning to Predict the Cosmological Structure Formation

Matter evolved under influence of gravity from minuscule density fluctuations. Non-perturbative structure formed hierarchically over all scales, and developed non-Gaussian features in the Universe, known as the Cosmic Web. To fully understand the structure formation of the Universe is one of the holy grails of modern astrophysics. Astrophysicists survey large volumes of the Universe and employ a large ensemble of computer simulations to compare with the observed data in order to extract the full information of our own Universe. However, to evolve trillions of galaxies over billions of years even with the simplest physics is a daunting task. We build a deep neural network, the Deep Density Displacement Model (hereafter D$^3$M), to predict the non-linear structure formation of the Universe from simple linear perturbation theory. Our extensive analysis, demonstrates that D$^3$M outperforms the second order perturbation theory (hereafter 2LPT), the commonly used fast approximate simulation method, in point-wise comparison, 2-point correlation, and 3-point correlation. We also show that D$^3$M is able to accurately extrapolate far beyond its training data, and predict structure formation for significantly different cosmological parameters. Our study proves, for the first time, that deep learning is a practical and accurate alternative to approximate simulations of the gravitational structure formation of the Universe.

astro-ph.CO↗
arXiv2019

SciPy 1.0--Fundamental Algorithms for Scientific Computing in Python

SciPy is an open source scientific computing library for the Python programming language. SciPy 1.0 was released in late 2017, about 16 years after the original version 0.1 release. SciPy has become a de facto standard for leveraging scientific algorithms in the Python programming language, with more than 600 unique code contributors, thousands of dependent packages, over 100,000 dependent repositories, and millions of downloads per year. This includes usage of SciPy in almost half of all machine learning projects on GitHub, and usage by high profile projects including LIGO gravitational wave analysis and creation of the first-ever image of a black hole (M87). The library includes functionality spanning clustering, Fourier transforms, integration, interpolation, file I/O, linear algebra, image processing, orthogonal distance regression, minimization algorithms, signal processing, sparse matrix handling, computational geometry, and statistics. In this work, we provide an overview of the capabilities and development practices of the SciPy library and highlight some recent technical developments.

cs.MS↗
arXiv2019

UNIT project: Universe $N$-body simulations for the Investigation of Theoretical models from galaxy surveys

We present the UNIT $N$-body cosmological simulations project, designed to provide precise predictions for nonlinear statistics of the galaxy distribution. We focus on characterizing statistics relevant to emission line and luminous red galaxies in the current and upcoming generation of galaxy surveys. We use a suite of precise particle mesh simulations (FastPM) as well as with full $N$-body calculations with a mass resolution of $\sim 1.2\times10^9\,h^{-1}$M$_{\odot}$ to investigate the recently suggested technique of Angulo & Pontzen 2016 to suppress the variance of cosmological simulations We study redshift space distortions, cosmic voids, higher order statistics from $z=2$ down to $z=0$. We find that both two- and three-point statistics are unbiased. Over the scales of interest for baryon acoustic oscillations and redshift-space distortions, we find that the variance is greatly reduced in the two-point statistics and in the cross correlation between halos and cosmic voids, but is not reduced significantly for the three-point statistics. We demonstrate that the accuracy of the two-point correlation function for a galaxy survey with effective volume of 20 ($h^{-1}$Gpc)$^3$ is improved by about a factor of 40, indicating that two pairs of simulations with a volume of 1 ($h^{-1}$Gpc)$^3$ lead to the equivalent variance of $\sim$150 such simulations. The $N$-body simulations presented here thus provide an effective survey volume of about seven times the effective survey volume of DESI or Euclid. The data from this project, including dark matter fields, halo catalogues, and their clustering statistics, are publicly available at http://www.unitsims.org.

astro-ph.CO↗
arXiv2019

Predictions for the Abundance of High-redshift Galaxies in a Fuzzy Dark Matter Universe

During the last decades, rapid progress has been made in measurements of the rest-frame ultraviolet (UV) luminosity function (LF) for high-redshift galaxies ($z \geq 6$). The faint-end of the galaxy LF at these redshifts provides powerful constraints on different dark matter models that suppress small-scale structure formation. In this work we perform full hydrodynamical cosmological simulations of galaxy formation using an alternative DM model composed of extremely light bosonic particles ($m \sim 10^{-22}$ eV), also known as fuzzy dark matter (FDM), and examine the predictions for the galaxy stellar mass function and luminosity function at $z \geq 6$ for a range of FDM masses. We find that for FDM models with bosonic mass $m = 5\times10^{-22}$ eV, the number density of galaxies with stellar mass $\rm M_* \sim 10^7 M_{\odot}$ is suppressed by $\sim 40\%$ at z = 9, $\sim 20\%$ at z = 5, and the UV LFs within magnitude range of -16 < $M_{\rm UV}$ < -14 is suppressed by $\sim 60\%$ at $z = 9$, $\sim 20\%$ at $z = 5$ comparing to the CDM counterpart simulation. Comparing our predictions with current measurements of the faint-end LFs ($-18 \leqslant M_{\rm UV} \leqslant -14$), we find that FDM models with $m_{22} < 5\times10^{-22}$ are ruled out at $3σ$ confidence level. We expect that future LF measurements by James Webb Space Telescope (JWST), which will extend down to $M_{\rm UV} \sim -13$ for $z \lesssim 10$, with a survey volume that is comparable to the Hubble Ultra Deep Field (HUDF) would have the capability to constrain FDM models to $m\; \gtrsim 10^{-21}$ eV.

astro-ph.CO↗
arXiv2019

A Thermodynamic Picture of Financial Market and Model Risk

By treating the financial market as a thermodynamic system, we establish a one-to-one correspondence between thermodynamic variables and economic quantities. Measured by the expected loss under the worst-case scenario, financial risk caused by model uncertainty is regarded as a result of the interaction between financial market and external information sources. This forms a thermodynamic picture in which a closed system interacts with an external reservoir, reaching its equilibrium at the worst-case scenario. The severity of the worst-case scenario depends on the rate of heat dissipation, caused by information sources reducing the entropy of the system. This thermodynamic picture leads to simple and natural derivation of the characterization rules of the worst-case risk, and gives its Lagrangian and Hamiltonian forms. With its help financial practitioners may evaluate risks utilizing both equilibrium and non-equilibrium thermodynamics.

q-fin.RM↗
arXiv2019

Model Risk Measurement under Wasserstein Distance

The paper proposes a new approach to model risk measurement based on the Wasserstein distance between two probability measures. It formulates the theoretical motivation resulting from the interpretation of fictitious adversary of robust risk management. The proposed approach accounts for equivalent and non-equivalent probability measures and incorporates the economic reality of the fictitious adversary. It provides practically feasible results that overcome the restriction of considering only models implying probability measures equivalent to the reference model. The Wasserstein approach suits for various types of model risk problems, ranging from the single-asset hedging risk problem to the multi-asset allocation problem. The robust capital market line, accounting for the correlation risk, is not achievable with other non-parametric approaches.

q-fin.MF↗
arXiv2019

Non-Parametric Robust Model Risk Measurement with Path-Dependent Loss Functions

Understanding and measuring model risk is important to financial practitioners. However, there lacks a non-parametric approach to model risk quantification in a dynamic setting and with path-dependent losses. We propose a complete theory generalizing the relative-entropic approach by Glasserman and Xu to the dynamic case under any $f$-divergence. It provides an unified treatment for measuring both the worst-case risk and the $f$-divergence budget that originate from the model uncertainty of an underlying state process.

q-fin.MF↗
arXiv2019

Kinematic distributions of the $η_c$ photoproduction in $ep$ collisions within the nonrelativistic QCD framework

We study the $η_c$ photoproduction in $ep$ collisions in this paper. The short-distance coefficients for $c\bar{c}(^1S_0^{[1]})$, $c\bar{c}(^1S_0^{[8]})$, $c\bar{c}(^3S_1^{[8]})$, and $c\bar{c}(^1P_1^{[8]})$ photoproductions are evaluated at leading order in $α_s$ expansion, where the color-singlet contribution is achieved for the first time. We have carefully analyzed different kinematic distributions of the cross sections and found that the color-singlet contribution is considerably suppressed comparing with the color-octet parts. This feature renders the $η_c$ photoproduction process an ideal laboratory to test the color-octet mechanism in nonrelativistic QCD. By taking different sets of long-distance matrix elements, we have observed some apparently distinguishable predictions, which can be utilized to scrutinize the validity of these matrix elements.

hep-ph↗
arXiv2019

Precise Attack Synthesis for Smart Contracts

Smart contracts are programs running on top of blockchain platforms. They interact with each other through well-defined interfaces to perform financial transactions in a distributed system with no trusted third parties. But these interfaces also provide a favorable setting for attackers, who can exploit security vulnerabilities in smart contracts to achieve financial gain. This paper presents SmartScopy, a system for automatic synthesis of adversarial contracts that identify and exploit vulnerabilities in a victim smart contract. Our tool explores the space of \emph{attack programs} based on the Application Binary Interface (ABI) specification of a victim smart contract in the Ethereum ecosystem. To make the synthesis tractable, we introduce \emph{summary-based symbolic evaluation}, which significantly reduces the number of instructions that our synthesizer needs to evaluate symbolically, without compromising the precision of the vulnerability query. Building on the summary-based symbolic evaluation, SmartScopy further introduces a novel approach for partitioning the synthesis search space for parallel exploration, as well as a lightweight deduction technique that can prune infeasible candidates earlier. We encoded common vulnerabilities of smart contracts in our query language, and evaluated SmartScopy on the entire data set from etherscan with $>$25K smart contracts. Our experiments demonstrate the benefits of summary-based symbolic evaluation and show that SmartScopy outperforms two state-of-the-art smart contracts analyzers, Oyente and Contractfuzz, in terms of running time, precision, and soundness. Furthermore, running on recent popular smart contracts, SmartScopy uncovers 20 vulnerable smart contracts that contain the recent BatchOverflow vulnerability and cannot be precisely detected by existing tools.

cs.CR↗
arXiv2019

Phenomenological NLO analysis of eta(c) production at the LHC in the collider and fixed-target modes

In view of the good agreement between the LHCb prompt-eta(c) data at sqrt(s)=7 and 8 TeV and the NLO colour-singlet model predictions --i.e. the leading v^2 NRQCD contribution--, we provide predictions in the LHCb acceptance for the forthcoming 13 TeV analysis bearing on data taken during the LHC Run2. We also provide predictions for sqrt(s)=115 GeV for proton-hydrogen collisions in the fixed-target mode which could be studied during the LHC Run3. Our predictions are complemented by a full theoretical uncertainty analysis. In addition to cross section predictions, we elaborate on the uncertainties on the p bar-p branching ratio --necessary for data-theory comparison-- and discuss other usable branching fractions for future studies.

hep-ph↗
arXiv2019

Bounded Projective Functions and Hyperbolic Metrics with Isolated Singularities

We establish a correspondence on a Riemann surface between hyperbolic metrics with isolated singularities and bounded projective functions whose Schwarzian derivatives have at most double poles and whose monodromies lie in ${\rm PSU}(1,\,1)$. As an application, we construct explicitly a new class of hyperbolic metrics with countably many singularities on the unit disc.

math.DG↗
arXiv2019

Disconnected Covariance of 2-point Functions in Large-Scale Structure

Optimal analyses using the 2-point functions of large-scale structure probes require accurate covariance matrices. A covariance matrix of the 2-point function comprises the disconnected part and the connected part. While the connected covariance only becomes important on small scales, the disconnected covariance is dominant on large scales, where the survey window has a significant impact. In this work, we develop an analytical method to compute the disconnected covariance, accounting for the window effect. Derived under the flat-sky approximation, our formalism is applicable to wide surveys by swapping in the curved-sky window functions. Our method works for both the power spectrum and the correlation function, and applies to the covariances of various probes including the multipoles and the wedges of 3D clustering, the angular and the projected statistics of clustering and shear, as well as the cross covariances between different probes. We verify the analytic covariance against the sample covariance from the galaxy mock simulations in two test cases: (1) the power spectrum multipole covariance, and (2) the joint covariance of the projected correlation function and the correlation function multipoles. Our method achieve good agreement with the mocks, while at a negligible computational cost. Unlike mocks, our analytic covariance is free of sampling noise, which often leads to numerical problems and the need to inflate the errors. In addition, our method can use the best-fit power spectrum as input, in contrast to the standard procedure of using a fiducial model that may deviate significantly from the truth. We also show that a naive diagonal power spectrum covariance underestimates the signal-to-noise ratio compared to our analytic covariance. The code that accompanies this paper is available at https://github.com/eelregit/covdisc.

astro-ph.CO↗