Showing papers by "Lehigh University published in 2017"

StackGAN: Text to Photo-Realistic Image Synthesis with Stacked Generative Adversarial Networks

Abstract: Synthesizing high-quality images from text descriptions is a challenging problem in computer vision and has many practical applications. Samples generated by existing textto- image approaches can roughly reflect the meaning of the given descriptions, but they fail to contain necessary details and vivid object parts. In this paper, we propose Stacked Generative Adversarial Networks (StackGAN) to generate 256.256 photo-realistic images conditioned on text descriptions. We decompose the hard problem into more manageable sub-problems through a sketch-refinement process. The Stage-I GAN sketches the primitive shape and colors of the object based on the given text description, yielding Stage-I low-resolution images. The Stage-II GAN takes Stage-I results and text descriptions as inputs, and generates high-resolution images with photo-realistic details. It is able to rectify defects in Stage-I results and add compelling details with the refinement process. To improve the diversity of the synthesized images and stabilize the training of the conditional-GAN, we introduce a novel Conditioning Augmentation technique that encourages smoothness in the latent conditioning manifold. Extensive experiments and comparisons with state-of-the-arts on benchmark datasets demonstrate that the proposed method achieves significant improvements on generating photo-realistic images conditioned on text descriptions.

First-Line Nivolumab in Stage IV or Recurrent Non–Small-Cell Lung Cancer

Abstract: BackgroundNivolumab has been associated with longer overall survival than docetaxel among patients with previously treated non–small-cell lung cancer (NSCLC). In an open-label phase 3 trial, we compared first-line nivolumab with chemotherapy in patients with programmed death ligand 1 (PD-L1)–positive NSCLC. MethodsWe randomly assigned, in a 1:1 ratio, patients with untreated stage IV or recurrent NSCLC and a PD-L1 tumor-expression level of 1% or more to receive nivolumab (administered intravenously at a dose of 3 mg per kilogram of body weight once every 2 weeks) or platinum-based chemotherapy (administered once every 3 weeks for up to six cycles). Patients receiving chemotherapy could cross over to receive nivolumab at the time of disease progression. The primary end point was progression-free survival, as assessed by means of blinded independent central review, among patients with a PD-L1 expression level of 5% or more. ResultsAmong the 423 patients with a PD-L1 expression level of 5% or more, the media...

Maximizing the right stuff: The trade-off between membrane permeability and selectivity

Abstract: BACKGROUND Synthetic membranes are used for desalination, dialysis, sterile filtration, food processing, dehydration of air and other industrial, medical, and environmental applications due to low energy requirements, compact design, and mechanical simplicity. New applications are emerging from the water-energy nexus, shale gas extraction, and environmental needs such as carbon capture. All membranes exhibit a trade-off between permeability—i.e., how fast molecules pass through a membrane material—and selectivity—i.e., to what extent the desired molecules are separated from the rest. However, biological membranes such as aquaporins and ion channels are both highly permeable and highly selective. Separation based on size difference is common, but there are other ways to either block one component or enhance transport of another through a membrane. Based on increasing molecular understanding of both biological and synthetic membranes, key design criteria for new membranes have emerged: (i) properly sized free-volume elements (or pores), (ii) narrow free-volume element (or pore size) distribution, (iii) a thin active layer, and (iv) highly tuned interactions between permeants of interest and the membrane. Here, we discuss the permeability/selectivity trade-off, highlight similarities and differences between synthetic and biological membranes, describe challenges for existing membranes, and identify fruitful areas of future research. ADVANCES Many organic, inorganic, and hybrid materials have emerged as potential membranes. In addition to polymers, used for most membranes today, materials such as carbon molecular sieves, ceramics, zeolites, various nanomaterials (e.g., graphene, graphene oxide, and metal organic frameworks), and their mixtures with polymers have been explored. Simultaneously, global challenges such as climate change and rapid population growth stimulate the search for efficient water purification and energy-generation technologies, many of which are membrane-based. Additional driving forces include wastewater reuse from shale gas extraction and improvement of chemical and petrochemical separation processes by increasing the use of light hydrocarbons for chemicals manufacturing. OUTLOOK Opportunities for advancing membranes include (i) more mechanically, chemically, and thermally robust materials; (ii) judiciously higher permeability and selectivity for applications where such improvements matter; and (iii) more emphasis on fundamental structure/property/processing relations. There is a pressing need for membranes with improved selectivity, rather than membranes with improved permeability, especially for water purification. Modeling at all length scales is needed to develop a coherent molecular understanding of membrane properties, provide insight for future materials design, and clarify the fundamental basis for trade-off behavior. Basic molecular-level understanding of thermodynamic and diffusion properties of water and ions in charged membranes for desalination and energy applications such as fuel cells is largely incomplete. Fundamental understanding of membrane structure optimization to control transport of minor species (e.g., trace-organic contaminants in desalination membranes, neutral compounds in charged membranes, and heavy hydrocarbons in membranes for natural gas separation) is needed. Laboratory evaluation of membranes is often conducted with highly idealized mixtures, so separation performance in real applications with complex mixtures is poorly understood. Lack of systematic understanding of methodologies to scale promising membranes from the few square centimeters needed for laboratory studies to the thousands of square meters needed for large applications stymies membrane deployment. Nevertheless, opportunities for membranes in both existing and emerging applications, together with an expanding set of membrane materials, hold great promise for membranes to effectively address separations needs.

DeepXplore: Automated Whitebox Testing of Deep Learning Systems

Abstract: Deep learning (DL) systems are increasingly deployed in safety- and security-critical domains including self-driving cars and malware detection, where the correctness and predictability of a system's behavior for corner case inputs are of great importance Existing DL testing depends heavily on manually labeled data and therefore often fails to expose erroneous behaviors for rare inputs We design, implement, and evaluate DeepXplore, the first whitebox framework for systematically testing real-world DL systems First, we introduce neuron coverage for systematically measuring the parts of a DL system exercised by test inputs Next, we leverage multiple DL systems with similar functionality as cross-referencing oracles to avoid manual checking Finally, we demonstrate how finding inputs for DL systems that both trigger many differential behaviors and achieve high neuron coverage can be represented as a joint optimization problem and solved efficiently using gradient-based search techniques DeepXplore efficiently finds thousands of incorrect corner case behaviors (eg, self-driving cars crashing into guard rails and malware masquerading as benign software) in state-of-the-art DL models with thousands of neurons trained on five popular datasets including ImageNet and Udacity self-driving challenge data For all tested DL models, on average, DeepXplore generated one test input demonstrating incorrect behavior within one second while running only on a commodity laptop We further show that the test inputs generated by DeepXplore can also be used to retrain the corresponding DL model to improve the model's accuracy by up to 3%

Phosphorylation of the FUS low‐complexity domain disrupts phase separation, aggregation, and toxicity

Abstract: Neuronal inclusions of aggregated RNA‐binding protein fused in sarcoma (FUS) are hallmarks of ALS and frontotemporal dementia subtypes. Intriguingly, FUS9s nearly uncharged, aggregation‐prone, yeast prion‐like, low sequence‐complexity domain (LC) is known to be targeted for phosphorylation. Here we map in vitro and in‐cell phosphorylation sites across FUS LC. We show that both phosphorylation and phosphomimetic variants reduce its aggregation‐prone/prion‐like character, disrupting FUS phase separation in the presence of RNA or salt and reducing FUS propensity to aggregate. Nuclear magnetic resonance spectroscopy demonstrates the intrinsically disordered structure of FUS LC is preserved after phosphorylation; however, transient domain collapse and self‐interaction are reduced by phosphomimetics. Moreover, we show that phosphomimetic FUS reduces aggregation in human and yeast cell models, and can ameliorate FUS‐associated cytotoxicity. Hence, post‐translational modification may be a mechanism by which cells control physiological assembly and prevent pathological protein aggregation, suggesting a potential treatment pathway amenable to pharmacologic modulation.

Atomic-layered Au clusters on α-MoC as catalysts for the low-temperature water-gas shift reaction

Abstract: The water-gas shift (WGS) reaction (where carbon monoxide plus water yields dihydrogen and carbon dioxide) is an essential process for hydrogen generation and carbon monoxide removal in various energy-related chemical operations. This equilibrium-limited reaction is favored at a low working temperature. Potential application in fuel cells also requires a WGS catalyst to be highly active, stable, and energy-efficient and to match the working temperature of on-site hydrogen generation and consumption units. We synthesized layered gold (Au) clusters on a molybdenum carbide (α-MoC) substrate to create an interfacial catalyst system for the ultralow-temperature WGS reaction. Water was activated over α-MoC at 303 kelvin, whereas carbon monoxide adsorbed on adjacent Au sites was apt to react with surface hydroxyl groups formed from water splitting, leading to a high WGS activity at low temperatures.

AttnGAN: Fine-Grained Text to Image Generation with Attentional Generative Adversarial Networks

Abstract: In this paper, we propose an Attentional Generative Adversarial Network (AttnGAN) that allows attention-driven, multi-stage refinement for fine-grained text-to-image generation. With a novel attentional generative network, the AttnGAN can synthesize fine-grained details at different subregions of the image by paying attentions to the relevant words in the natural language description. In addition, a deep attentional multimodal similarity model is proposed to compute a fine-grained image-text matching loss for training the generator. The proposed AttnGAN significantly outperforms the previous state of the art, boosting the best reported inception score by 14.14% on the CUB dataset and 170.25% on the more challenging COCO dataset. A detailed analysis is also performed by visualizing the attention layers of the AttnGAN. It for the first time shows that the layered attentional GAN is able to automatically select the condition at the word level for generating different parts of the image.

Bulk properties of the medium produced in relativistic heavy-ion collisions from the beam energy scan program

Abstract: © 2017 American Physical Society. We present measurements of bulk properties of the matter produced in Au+Au collisions at sNN=7.7,11.5,19.6,27, and 39 GeV using identified hadrons (π±, K±, p, and p) from the STAR experiment in the Beam Energy Scan (BES) Program at the Relativistic Heavy Ion Collider (RHIC). Midrapidity (|y| < 0.1) results for multiplicity densities dN/dy, average transverse momenta (pT), and particle ratios are presented. The chemical and kinetic freeze-out dynamics at these energies are discussed and presented as a function of collision centrality and energy. These results constitute the systematic measurements of bulk properties of matter formed in heavy-ion collisions over a broad range of energy (or baryon chemical potential) at RHIC.

StackGAN++: Realistic Image Synthesis with Stacked Generative Adversarial Networks

Abstract: Although Generative Adversarial Networks (GANs) have shown remarkable success in various tasks, they still face challenges in generating high quality images. In this paper, we propose Stacked Generative Adversarial Networks (StackGAN) aiming at generating high-resolution photo-realistic images. First, we propose a two-stage generative adversarial network architecture, StackGAN-v1, for text-to-image synthesis. The Stage-I GAN sketches the primitive shape and colors of the object based on given text description, yielding low-resolution images. The Stage-II GAN takes Stage-I results and text descriptions as inputs, and generates high-resolution images with photo-realistic details. Second, an advanced multi-stage generative adversarial network architecture, StackGAN-v2, is proposed for both conditional and unconditional generative tasks. Our StackGAN-v2 consists of multiple generators and discriminators in a tree-like structure; images at multiple scales corresponding to the same scene are generated from different branches of the tree. StackGAN-v2 shows more stable training behavior than StackGAN-v1 by jointly approximating multiple distributions. Extensive experiments demonstrate that the proposed stacked generative adversarial networks significantly outperform other state-of-the-art methods in generating photo-realistic images.

Aqueous Au-Pd colloids catalyze selective CH4 oxidation to CH3OH with O2 under mild conditions

Abstract: The selective oxidation of methane, the primary component of natural gas, remains an important challenge in catalysis. We used colloidal gold-palladium nanoparticles, rather than the same nanoparticles supported on titanium oxide, to oxidize methane to methanol with high selectivity (92%) in aqueous solution at mild temperatures. Then, using isotopically labeled oxygen (O2) as an oxidant in the presence of hydrogen peroxide (H2O2), we demonstrated that the resulting methanol incorporated a substantial fraction (70%) of gas-phase O2. More oxygenated products were formed than the amount of H2O2 consumed, suggesting that the controlled breakdown of H2O2 activates methane, which subsequently incorporates molecular oxygen through a radical process. If a source of methyl radicals can be established, then the selective oxidation of methane to methanol using molecular oxygen is possible.

Identification of single-site gold catalysis in acetylene hydrochlorination

Abstract: There remains considerable debate over the active form of gold under operating conditions of a recently validated gold catalyst for acetylene hydrochlorination. We have performed an in situ x-ray absorption fine structure study of gold/carbon (Au/C) catalysts under acetylene hydrochlorination reaction conditions and show that highly active catalysts comprise single-site cationic Au entities whose activity correlates with the ratio of Au(I):Au(III) present. We demonstrate that these Au/C catalysts are supported analogs of single-site homogeneous Au catalysts and propose a mechanism, supported by computational modeling, based on a redox couple of Au(I)-Au(III) species.

Planning Dynamically Feasible Trajectories for Quadrotors Using Safe Flight Corridors in 3-D Complex Environments

Abstract: There is extensive literature on using convex optimization to derive piece-wise polynomial trajectories for controlling differential flat systems with applications to three-dimensional flight for Micro Aerial Vehicles. In this work, we propose a method to formulate trajectory generation as a quadratic program (QP) using the concept of a Safe Flight Corridor (SFC). The SFC is a collection of convex overlapping polyhedra that models free space and provides a connected path from the robot to the goal position. We derive an efficient convex decomposition method that builds the SFC from a piece-wise linear skeleton obtained using a fast graph search technique. The SFC provides a set of linear inequality constraints in the QP allowing real-time motion planning. Because the range and field of view of the robot's sensors are limited, we develop a framework of Receding Horizon Planning , which plans trajectories within a finite footprint in the local map, continuously updating the trajectory through a re-planning process. The re-planning process takes between 50 to 300 ms for a large and cluttered map. We show the feasibility of our approach, its completeness and performance, with applications to high-speed flight in both simulated and physical experiments using quadrotors.

SARAH: A Novel Method for Machine Learning Problems Using Stochastic Recursive Gradient

Abstract: In this paper, we propose a StochAstic Recursive grAdient algoritHm (SARAH), as well as its practical variant SARAH+, as a novel approach to the finite-sum minimization problems. Different from the vanilla SGD and other modern stochastic methods such as SVRG, S2GD, SAG and SAGA, SARAH admits a simple recursive framework for updating stochastic gradient estimates; when comparing to SAG/SAGA, SARAH does not require a storage of past gradients. The linear convergence rate of SARAH is proven under strong convexity assumption. We also prove a linear convergence rate (in the strongly convex case) for an inner loop of SARAH, the property that SVRG does not possess. Numerical experiments demonstrate the efficiency of our algorithm.

A global multiproxy database for temperature reconstructions of the Common Era

Abstract: Reproducible climate reconstructions of the Common Era (1 CE to present) are key to placing industrial-era warming into the context of natural climatic variability. Here we present a community-sourced database of temperature-sensitive proxy records from the PAGES2k initiative. The database gathers 692 records from 648 locations, including all continental regions and major ocean basins. The records are from trees, ice, sediment, corals, speleothems, documentary evidence, and other archives. They range in length from 50 to 2000 years, with a median of 547 years, while temporal resolution ranges from biweekly to centennial. Nearly half of the proxy time series are significantly correlated with HadCRUT4.2 surface temperature over the period 1850–2014. Global temperature composites show a remarkable degree of coherence between high- and low-resolution archives, with broadly similar patterns across archive types, terrestrial versus marine locations, and screening criteria. The database is suited to investigations of global and regional temperature variability over the Common Era, and is shared in the Linked Paleo Data (LiPD) format, including serializations in Matlab, R and Python.

Elastocapillarity: Surface Tension and the Mechanics of Soft Solids

Abstract: It is widely appreciated that surface tension can dominate the behavior of liquids at small scales. Solids also have surface stresses of a similar magnitude, but they are usually overlooked. However, recent work has shown that these can play a central role in the mechanics of soft solids such as gels. Here, we review this emerging field. We outline the theory of surface stresses, from both mechanical and thermodynamic perspectives, emphasizing the relationship between surface stress and surface energy. We describe a wide range of phenomena at interfaces and contact lines where surface stresses play an important role. We highlight how surface stresses cause dramatic departures from classic theories for wetting (Young–Dupre), adhesion (Johnson–Kendall–Roberts), and composites (Eshelby). A common thread is the importance of the ratio of surface stress to an elastic modulus, which defines a length scale below which surface stresses can dominate.

A giant planet undergoing extreme-ultraviolet irradiation by its hot massive-star host

Abstract: The amount of ultraviolet irradiation and ablation experienced by a planet depends strongly on the temperature of its host star. Of the thousands of extrasolar planets now known, only six have been found that transit hot, A-type stars (with temperatures of 7,300–10,000 kelvin), and no planets are known to transit the even hotter B-type stars. For example, WASP-33 is an A-type star with a temperature of about 7,430 kelvin, which hosts the hottest known transiting planet, WASP-33b (ref. 1); the planet is itself as hot as a red dwarf star of type M (ref. 2). WASP-33b displays a large heat differential between its dayside and nightside, and is highly inflated–traits that have been linked to high insolation. However, even at the temperature of its dayside, its atmosphere probably resembles the molecule-dominated atmospheres of other planets and, given the level of ultraviolet irradiation it experiences, its atmosphere is unlikely to be substantially ablated over the lifetime of its star. Here we report observations of the bright star HD 195689 (also known as KELT-9), which reveal a close-in (orbital period of about 1.48 days) transiting giant planet, KELT-9b. At approximately 10,170 kelvin, the host star is at the dividing line between stars of type A and B, and we measure the dayside temperature of KELT-9b to be about 4,600 kelvin. This is as hot as stars of stellar type K4 (ref. 5). The molecules in K stars are entirely dissociated, and so the primary sources of opacity in the dayside atmosphere of KELT-9b are probably atomic metals. Furthermore, KELT-9b receives 700 times more extreme-ultraviolet radiation (that is, with wavelengths shorter than 91.2 nanometres) than WASP-33b, leading to a predicted range of mass-loss rates that could leave the planet largely stripped of its envelope during the main-sequence lifetime of the host star.

CHARMM-GUI ligand reader and modeler for CHARMM force field generation of small molecules

Abstract: Reading ligand structures into any simulation program is often nontrivial and time consuming, especially when the force field parameters and/or structure files of the corresponding molecules are not available. To address this problem, we have developed Ligand Reader & Modeler in CHARMM-GUI. Users can upload ligand structure information in various forms (using PDB ID, ligand ID, SMILES, MOL/MOL2/SDF file, or PDB/mmCIF file), and the uploaded structure is displayed on a sketchpad for verification and further modification. Based on the displayed structure, Ligand Reader & Modeler generates the ligand force field parameters and necessary structure files by searching for the ligand in the CHARMM force field library or using the CHARMM general force field (CGenFF). In addition, users can define chemical substitution sites and draw substituents in each site on the sketchpad to generate a set of combinatorial structure files and corresponding force field parameters for throughput or alchemical free energy simulations. Finally, the output from Ligand Reader & Modeler can be used in other CHARMM-GUI modules to build a protein-ligand simulation system for all supported simulation programs, such as CHARMM, NAMD, GROMACS, AMBER, GENESIS, LAMMPS, Desmond, OpenMM, and CHARMM/OpenMM. Ligand Reader & Modeler is available as a functional module of CHARMM-GUI at http://www.charmm-gui.org/input/ligandrm. © 2017 Wiley Periodicals, Inc.

Nature of Active Sites and Surface Intermediates during SCR of NO with NH₃ by Supported V₂O₅--WO₃/TiO₂ Catalysts.

Abstract: Time-resolved in situ IR was performed during selective catalytic reduction of NO with NH3 on supported V2O5–WO3/TiO2 catalysts to examine the distribution and reactivity of surface ammonia species on Lewis and Bronsted acid sites. While both species were found to participate in the SCR reaction, their relative population depends on the coverage of the surface vanadia and tungsta sites, temperature, and moisture. Although the more abundant surface NH4+,ads intermediates dominate the overall SCR reaction, especially for hydrothermally aged catalysts, the minority surface NH3,ads intermediates exhibit a higher specific SCR activity (TOF). The current study serves to resolve the long-standing controversy about the active sites for SCR of NO with NH3 by supported V2O5–WO3/TiO2 catalysts.

Topological liquid diode

Abstract: The last two decades have witnessed an explosion of interest in the field of droplet-based microfluidics for their multifarious applications. Despite rapid innovations in strategies to generate small-scale liquid transport on these devices, the speed of motion is usually slow, the transport distance is limited, and the flow direction is not well controlled because of unwanted pinning of contact lines by defects on the surface. We report a new method of microscopic liquid transport based on a unique topological structure. This method breaks the contact line pinning through efficient conversion of excess surface energy to kinetic energy at the advancing edge of the droplet while simultaneously arresting the reverse motion of the droplet via strong pinning. This results in a novel topological fluid diode that allows for a rapid, directional, and long-distance transport of virtually any kind of liquid without the need for an external energy input.

Prophage-mediated defence against viral attack and viral counter-defence.

Abstract: Temperate phages are common, and prophages are abundant residents of sequenced bacterial genomes. Mycobacteriophages are viruses that infect mycobacterial hosts including Mycobacterium tuberculosis and Mycobacterium smegmatis, encompass substantial genetic diversity and are commonly temperate. Characterization of ten Cluster N temperate mycobacteriophages revealed at least five distinct prophage-expressed viral defence systems that interfere with the infection of lytic and temperate phages that are either closely related (homotypic defence) or unrelated (heterotypic defence) to the prophage. Target specificity is unpredictable, ranging from a single target phage to one-third of those tested. The defence systems include a single-subunit restriction system, a heterotypic exclusion system and a predicted (p)ppGpp synthetase, which blocks lytic phage growth, promotes bacterial survival and enables efficient lysogeny. The predicted (p)ppGpp synthetase coded by the Phrann prophage defends against phage Tweety infection, but Tweety codes for a tetrapeptide repeat protein, gp54, which acts as a highly effective counter-defence system. Prophage-mediated viral defence offers an efficient mechanism for bacterial success in host-virus dynamics, and counter-defence promotes phage co-evolution.

Bridge life-cycle performance and cost: analysis, prediction, optimisation and decision-making

Abstract: The development of a generalised framework for assessing bridge life-cycle performance and cost, with emphasis on analysis, prediction, optimisation and decision-making under uncertainty, is briefly addressed. The central issue underlying the importance of the life-cycle approach to bridge engineering is the need for a rational basis for making informed decisions regarding design, construction, inspection, monitoring, maintenance, repair, rehabilitation, replacement and management of bridges under uncertainty which is carried out by using multi-objective optimisation procedures that balance conflicting criteria such as performance and cost. A number of significant developments are summarised, including time-variant reliability, risk, resilience, and sustainability of bridges, bridge transportation networks and interdependent infrastructure systems. Furthermore, the effects of climate change on the probabilistic life-cycle performance assessment of highway bridges are addressed. Moreover, integrati...

Customer participation in services: domain, scope, and boundaries

Abstract: Extant service research considers several aspects of customer participation (CP) but lacks a clear and inclusive typology that delineates CP’s domain, scope, or boundaries. To address this gap, the authors build on a review of extant literature and propose a typology to classify CP into three categories—mandatory, replaceable, and voluntary. They demonstrate how this proposed typology improves the conceptual and empirical clarity of CP research. More specifically, the authors (1) suggest using “customer participation” to replace other terminologies such as coproduction and cocreation to reduce confusion; (2) conceptualize CP, customer engagement, and customer innovation as related but distinct concepts; (3) use the proposed typology to extend existing conceptualizations, integrate prior empirical research, and reconcile conflicting findings. Building on the enhanced conceptual clarity, managerial implications and future research directions are discussed.

SARAH: A Novel Method for Machine Learning Problems Using Stochastic Recursive Gradient

Abstract: In this paper, we propose a StochAstic Recursive grAdient algoritHm (SARAH), as well as its practical variant SARAH+, as a novel approach to the finite-sum minimization problems. Different from the vanilla SGD and other modern stochastic methods such as SVRG, S2GD, SAG and SAGA, SARAH admits a simple recursive framework for updating stochastic gradient estimates; when comparing to SAG/SAGA, SARAH does not require a storage of past gradients. The linear convergence rate of SARAH is proven under strong convexity assumption. We also prove a linear convergence rate (in the strongly convex case) for an inner loop of SARAH, the property that SVRG does not possess. Numerical experiments demonstrate the efficiency of our algorithm.

Wire and arc additive manufactured steel: Tensile and wear properties

Abstract: The present study systematically investigated the mechanical properties of wire-based (wire and arc additive manufacturing, known as WAAM) deposition of steel metals, both stainless steel 304 and mild steel ER70S Graded material properties of stainless steel 304 were observed for wear and hardness in the direction of deposition and in Z height, due to variations in local thermal histories of the metal Wear rates decreased significantly (p = 56 × 10 −12 by one-way ANOVA) along the length of the deposited material, from K = 262 x 10 −5 mm 3 /N m (+/− 232 x 10 −6 mm 3 /N m), to K = 063 mm3 x 10 −5 mm3/N m (+/−308 x 10-6 mm 3 /N m), whereas microhardness values increased significantly (p ∼ 0 by one-way ANOVA) along the same path from μ = 2023 HV and σ = 582 HV to 2109 HV and σ = 591 HV The yield and ultimate strength, however, were not found to be statistically significantly different (p = 055) along the direction of deposition for SS304 During wear testing, a grain refinement was observed directly beneath the wear scar in these materials in a focused ion beam channel observed under scanning electron microscopy Additionally, no significant difference in yield strength was observed in printed mild steel (ER70S) between vertical and horizontal specimens The observed graded mechanical properties in stainless steel 304 allow the opportunity for varying the processing conditions to design parts with locally optimized or functionally graded mechanical properties

Human–water interface in hydrological modelling : current status and future directions

Abstract: Over recent decades, the global population has been rapidly increasing and human activities have altered terrestrial water fluxes to an unprecedented extent. The phenomenal growth of the human footprint has significantly modified hydrological processes in various ways (e.g. irrigation, artificial dams, and water diversion) and at various scales (from a watershed to the globe). During the early 1990s, awareness of the potential for increased water scarcity led to the first detailed global water resource assessments. Shortly thereafter, in order to analyse the human perturbation on terrestrial water resources, the first generation of largescale hydrological models (LHMs) was produced. However, at this early stage few models considered the interaction between terrestrial water fluxes and human activities, including water use and reservoir regulation, and even fewer models distinguished water use from surface water and groundwater resources. Since the early 2000s, a growing number of LHMs have incorporated human impacts on the hydrological cycle, yet the representation of human activities in hydrological models remains challenging. In this paper we provide a synthesis of progress in the development and application of human impact modelling in LHMs. We highlight a number of key challenges and discuss possible improvements in order to better represent the human-water interface in hydrological models.

Low-Rank Tensor Decomposition-Aided Channel Estimation for Millimeter Wave MIMO-OFDM Systems

Abstract: We consider the problem of downlink channel estimation for millimeter wave (mmWave) MIMO-OFDM systems, where both the base station (BS) and the mobile station (MS) employ large antenna arrays for directional precoding/beamforming. Hybrid analog and digital beamforming structures are employed in order to offer a compromise between hardware complexity and system performance. Different from most existing studies that are concerned with narrowband channels, we consider estimation of wideband mmWave channels with frequency selectivity, which is more appropriate for mmWave MIMO-OFDM systems. By exploiting the sparse scattering nature of mmWave channels, we propose a CANDECOMP/PARAFAC (CP) decomposition-based method for channel parameter estimation (including angles of arrival/departure, time delays, and fading coefficients). In our proposed method, the received signal at the MS is expressed as a third-order tensor. We show that the tensor has the form of a low-rank CP, and the channel parameters can be estimated from the associated factor matrices. Our analysis reveals that the uniqueness of the CP decomposition can be guaranteed even when the size of the tensor is small. Hence the proposed method has the potential to achieve substantial training overhead reduction. We also develop Cramer-Rao bound (CRB) results for channel parameters and compare our proposed method with a compressed sensing-based method. Simulation results show that the proposed method attains mean square errors that are very close to their associated CRBs and present a clear advantage over the compressed sensing-based method.

A trust region algorithm with a worst-case iteration complexity of $$\mathcal{O}(\epsilon ^{-3/2})$$O(∈-3/2) for nonconvex optimization

Abstract: We propose a trust region algorithm for solving nonconvex smooth optimization problems. For any $$\overline{\epsilon }\in (0,\infty )$$∈¯?(0,?), the algorithm requires at most $$\mathcal{O}(\epsilon ^{-3/2})$$O(∈-3/2) iterations, function evaluations, and derivative evaluations to drive the norm of the gradient of the objective function below any $$\epsilon \in (0,\overline{\epsilon }]$$∈?(0,∈¯]. This improves upon the $$\mathcal{O}(\epsilon ^{-2})$$O(∈-2) bound known to hold for some other trust region algorithms and matches the $$\mathcal{O}(\epsilon ^{-3/2})$$O(∈-3/2) bound for the recently proposed Adaptive Regularisation framework using Cubics, also known as the arc algorithm. Our algorithm, entitled trace, follows a trust region framework, but employs modified step acceptance criteria and a novel trust region update mechanism that allow the algorithm to achieve such a worst-case global complexity bound. Importantly, we prove that our algorithm also attains global and fast local convergence guarantees under similar assumptions as for other trust region algorithms. We also prove a worst-case upper bound on the number of iterations, function evaluations, and derivative evaluations that the algorithm requires to obtain an approximate second-order stationary point.

Distributed optimization with arbitrary local solvers

Abstract: With the growth of data and necessity for distributed optimization methods, solvers that work well on a single machine must be re-designed to leverage distributed computation. Recent work in this area has been limited by focusing heavily on developing highly specific methods for the distributed environment. These special-purpose methods are often unable to fully leverage the competitive performance of their well-tuned and customized single machine counterparts. Further, they are unable to easily integrate improvements that continue to be made to single machine methods. To this end, we present a framework for distributed optimization that both allows the flexibility of arbitrary solvers to be used on each single machine locally and yet maintains competitive performance against other state-of-the-art special-purpose distributed methods. We give strong primal–dual convergence rate guarantees for our framework that hold for arbitrary local solvers. We demonstrate the impact of local solver selection both theoretically and in an extensive experimental comparison. Finally, we provide thorough implementation details for our framework, highlighting areas for practical performance gains.

(Cross-)Browser Fingerprinting via OS and Hardware Level Features.

Abstract: In this paper, we propose a browser fingerprinting technique that can track users not only within a single browser but also across different browsers on the same machine. Specifically, our approach utilizes many novel OS and hardware level features, such as those from graphics cards, CPU, and installed writing scripts. We extract these features by asking browsers to perform tasks that rely on corresponding OS and hardware functionalities. Our evaluation shows that our approach can successfully identify 99.24% of users as opposed to 90.84% for state of the art on single-browser fingerprinting against the same dataset. Further, our approach can achieve higher uniqueness rate than the only cross-browser approach in the literature with similar stability.

A Carbon Nanotube Reporter of miRNA Hybridization Events In Vivo.

Abstract: MicroRNAs and other small oligonucleotides in biofluids are promising disease biomarkers, yet conventional assays require complex processing steps that are unsuitable for point-of-care testing or for implantable or wearable sensors. Single-walled carbon nanotubes are an ideal material for implantable sensors, owing to their emission in the near-infrared spectral region, photostability and exquisite sensitivity. Here, we report an engineered carbon-nanotube-based sensor capable of real-time optical quantification of hybridization events of microRNA and other oligonucleotides. The mechanism of the sensor arises from competitive effects between displacement of both oligonucleotide charge groups and water from the nanotube surface, which result in a solvatochromism-like response. The sensor, which allows for detection via single-molecule sensor elements and for multiplexing by using multiple nanotube chiralities, can monitor toehold-based strand-displacement events, which reverse the sensor response and regenerate the sensor complex. We also show that the sensor functions in whole urine and serum, and can non-invasively measure DNA and microRNA after implantation in live mice.