PolyMage: Automatic Optimization for Image Processing Pipelines

This paper presents the design and implementation of PolyMage, a domain-specific language and compiler for image processing pipelines. An image processing pipeline can be viewed as a graph of interconnected stages which process images successively. Each stage typically performs one of point-wise, stencil, reduction or data-dependent operations on image pixels. Individual stages in a pipeline typically exhibit abundant data parallelism that can be exploited with relative ease. However, the stages also require high memory bandwidth preventing effective utilization of parallelism available on modern architectures. For applications that demand high performance, the traditional options are to use optimized libraries like OpenCV or to optimize manually. While using libraries precludes optimization across library routines, manual optimization accounting for both parallelism and locality is very tedious. The focus of our system, PolyMage, is on automatically generating high-performance implementations of image processing pipelines expressed in a high-level declarative language. Our optimization approach primarily relies on the transformation and code generation capabilities of the polyhedral compiler framework. To the best of our knowledge, this is the first model-driven compiler for image processing pipelines that performs complex fusion, tiling, and storage optimization automatically. Experimental results on a modern multicore system show that the performance achieved by our automatic approach is up to 1.81x better than that achieved through manual tuning in Halide, a state-of-the-art language and compiler for image processing pipelines. For a camera raw image processing pipeline, our performance is comparable to that of a hand-tuned implementation.

A systems perspective of host-pathogen interactions: predicting disease outcome in tuberculosis

The complex web of interactions between the host immune system and the pathogen determines the outcome of any infection. A computational model of this interaction network, which encodes complex interplay among host and bacterial components, forms a useful basis for improving the understanding of pathogenesis, in filling knowledge gaps and consequently to identify strategies to counter the disease. We have built an extensive model of the Mycobacterium tuberculosis host-pathogen interactome, consisting of 75 nodes corresponding to host and pathogen molecules, cells, cellular states or processes. Vaccination effects, clearance efficiencies due to drugs and growth rates have also been encoded in the model. The system is modelled as a Boolean network. Virtual deletion experiments, multiple parameter scans and analysis of the system's response to perturbations, indicate that disabling processes such as phagocytosis and phagolysosome fusion or cytokines such as TNF-alpha and IFN-gamma, greatly impaired bacterial clearance, while removing cytokines such as IL-10 alongside bacterial defence proteins such as SapM greatly favour clearance. Simulations indicate a high propensity of the pathogen to persist under different conditions.

A systems perspective of host-pathogen interactions: predicting disease outcome in tuberculosis

The complex web of interactions between the host immune system and the pathogen determines the outcome of any infection. A computational model of this interaction network, which encodes complex interplay among host and bacterial components, forms a useful basis for improving the understanding of pathogenesis, in filling knowledge gaps and consequently to identify strategies to counter the disease. We have built an extensive model of the Mycobacterium tuberculosis host-pathogen interactome, consisting of 75 nodes corresponding to host and pathogen molecules, cells, cellular states or processes. Vaccination effects, clearance efficiencies due to drugs and growth rates have also been encoded in the model. The system is modelled as a Boolean network. Virtual deletion experiments, multiple parameter scans and analysis of the system's response to perturbations, indicate that disabling processes such as phagocytosis and phagolysosome fusion or cytokines such as TNF-alpha and IFN-gamma, greatly impaired bacterial clearance, while removing cytokines such as IL-10 alongside bacterial defence proteins such as SapM greatly favour clearance. Simulations indicate a high propensity of the pathogen to persist under different conditions.

Tuning the electronic effective mass in double-doped SrTiO3

We elucidate the relationship between effective mass and carrier concentration in an oxide semiconductor controlled by a double-doping mechanism. In this model oxide system, Sr1-xLaxTiO3-delta, we can tune the effective mass ranging from 6 to 20m(e) as a function of filling (carrier concentration) and the scattering mechanism, which are dependent on the chosen lanthanum-and oxygen-vacancy concentrations. The effective mass values were calculated from the Boltzmann transport equation using the measured transport properties of thin films of Sr1-xLaxTiO3-delta. We show that the effective mass decreases with carrier concentration in this large-band-gap, low-mobility oxide, and this behavior is contrary to the traditional high-mobility, small-effective-mass semiconductors.

On the large-scale diffuse magnetic field of the sun - II. The contribution of active regions

Dikpati and Choudhuri (1993, 1995) developed a model for the poleward migration of the weak diffuse magnetic field on the Sun's surface. This field was identified with the poloidal component produced by the solar dynamo operating at the base of the convection zone, and its evolution was studied by considering the effects of meridional circulation and turbulent diffusion. The earlier model is extended in this paper by incorporating the flux from, the decay of tilted active regions near the solar surface as an additional source of the poloidal field. This extended model can now explain various low-latitude features in the time-latitude diagram of the weak diffuse fields. These low-latitude features could not be accounted for in the earlier model, which was very successful in modeling the behavior at high latitudes. The time-latitude diagrams show that regions of a particular polarity often have 'tongues' of opposite polarity. Such tongues can be produced in the theoretical model by incorporating fluctuations in the source term arising out of the decaying active regions.

Rotation field in wedge indentation of metals

A study is made of the rotation field in wedge indentation of metals using copper as the model material system. Wedges with apical angles of 60 and 120 are used to indent annealed copper, and the deformation is mapped using image correlation. The indentation of annealed and strain-hardened copper is simulated using finite element analysis. The rotation field, derived from the deformation measurements, provides a clear way of distinguishing between cutting and compressive modes of deformation. Largely unidirectional rotation on one side of the symmetry line with small spatial rotation gradients is characteristic of compression. Bidirectional rotation with neighboring regions of opposing rotations and locally high rotation gradients characterizes cutting. In addition, the rotation demarcates such characteristic regions as the pile-up zone in indentation of a strain-hardened metal. The residual rotation field obtained after unloading is essentially the same as that at full load, indicating that it is a scalar proxy for plastic deformation as a whole.

Magnetic arm-switch-based three-phase series-shunt compensated quality AC power supply

This study presents a novel magnetic arm-switch-based integrated magnetic circuit for a three-phase series-shunt compensated uninterruptible power supply (UPS). The magnetic circuit acts as a common interacting field for a number of energy ports, viz., series inverter, shunt inverter, grid and load. The magnetic arm-switching technique ensures equivalent series or shunt connection between the inverters. In normal grid mode (stabiliser mode), the series inverter is used for series voltage correction and the shunt one for current correction. The inverters and the load are effectively connected in parallel when the grid power is not available. These inverters are then used to share the load power. The operation of the inverters in parallel is ensured by the magnetic arm-switching technique. This study also includes modelling of the magnetic circuit. A graphical technique called bond graph is used to model the system. In this model, the magnetic circuit is represented in terms of gyrator-capacitors. Therefore the model is also termed as gyrator-capacitor model. The model is used to extract the dynamic equations that are used to simulate the system using MATLAB/SIMULINK. This study also discusses a synchronously rotating reference frame-based control technique that is used for the control of the series and shunt inverters in different operating modes. Finally, the gyrator-capacitor model is validated by comparing the simulated and experimental results.

Invisible decays of the lightest Higgs boson in supersymmetric models

We consider supersymmetric models in which the lightest Higgs scalar can decay invisibly consistent with the constraints on the 126 GeV state discovered at the CERN LHC. We consider the invisible decay in the minimal supersymmetric standard model (MSSM), as well its extension containing an additional chiral singlet superfield, the so-called next-to-minimal or nonminimal supersymmetric standard model (NMSSM). We consider the case of MSSM with both universal as well as nonuniversal gaugino masses at the grand unified scale, and find that only an E-6 grand unified model with unnaturally large representation can give rise to sufficiently light neutralinos which can possibly lead to the invisible decay h(0) -> (chi) over tilde (0)(1)(chi) over tilde (0)(1). Following this, we consider the case of NMSSM in detail, where we also find that it is not possible to have the invisible decay of the lightest Higgs scalar with universal gaugino masses at the grand unified scale. We delineate the regions of the NMSSM parameter space where it is possible for the lightest Higgs boson to have a mass of about 126 GeV, and then concentrate on the region where this Higgs can decay into light neutralinos, with the soft gaugino masses M-1 and M-2 as two independent parameters, unconstrained by grand unification. We also consider, simultaneously, the other important invisible Higgs decay channel in the NMSSM, namely the decay into the lightest CP-odd scalars, h(1) -> a(1)a(1), which is studied in detail. With the invisible Higgs branching ratio being constrained by the present LHC results, we find that mu(eff) < 170 GeV and M-1 < 80 GeV are disfavored in NMSSM for fixed values of the other input parameters. The dependence of our results on the parameters of NMSSM is discussed in detail.

Substrate conductivity dependent modulation of cell proliferation and differentiation in vitro

In designing and developing various biomaterials, the influence of substrate properties, like surface topography, stiffness and wettability on the cell functionality has been investigated widely. However, such study to probe into the influence of the substrate conductivity on cell fate processes is rather limited. In order to address this issue, spark plasma sintered HA-CaTiO3 (Hydroxyapatite-Calcium titanate) has been used as a model material system to showcase the effect of varying conductivity on cell functionality. Being electroactive in nature, mouse myoblast cells (C2C12) were selected as a model cell line in this study. It was inferred that myoblast adhesion/growth systematically increases with substrate conductivity due to CaTiO3 addition to HA. Importantly, parallel arrangement of myoblast cells on higher CaTiO3 containing substrates indicate that self-adjustable cell patterning can be achieved on conductive biomaterials. Furthermore, enhanced myoblast assembly and myotube formation were recorded after 5 days of serum starvation. Overall, the present study conclusively establishes the positive impact of the substrate conductivity towards cell proliferation and differentiation as well as confirms the efficacy of HA-CaTiO3 biocomposites as conductive platforms to facilitate the growth, orientation and fusion of myoblasts, even when cultured in the absence of external electric field.

Insight into instabilities in burning droplets

The complex multiscale physics of nano-particle laden functional droplets in a reacting environment is of fundamental and applied significance for a wide variety of applications ranging from thermal sprays to pharmaceutics to modern day combustors using new brands of bio-fuels. Formation of homogenous nucleated bubbles at the superheat limit inside vaporizing droplets (with or without nanoparticles) represents an unstable system. Here we show that self-induced boiling in burning functional pendant droplets can produce severe volumetric shape oscillations. Internal pressure build-up due to ebullition activity ejects bubbles from the droplet domain causing undulations on the droplet surface and oscillations in bulk. Through experiments, we establish that the degree of droplet deformation depends on the frequency and intensity of these bubble expulsion events. In a distinct regime of single isolated bubble residing in the droplet, however, pre-ejection transient time is identified by Darrieus-Landau evaporative instability, where bubble-droplet system behaves as a synchronized driver-driven system with bulk bubble-shape oscillations being imposed on the droplet. The agglomeration of nanophase additives modulates the flow structures within the droplet and also influences the bubble inception and growth leading to different levels of instabilities. (C) 2014 AIP Publishing LLC.

Incipient straining in severe plastic deformation methods

Knowledge of the plasticity associated with the incipient stage of chip formation is useful toward developing an understanding of the deformation field underlying severe plastic deformation processes. The transition from a transient state of straining to a steady state was investigated in plane strain machining of a model material system-copper. Characterization of the evolution to a steady-state deformation field was made by image correlation, hardness mapping, load analysis, and microstructure characterization. Empirical relationships relating the deformation heterogeneity and the process parameters were found and explained by the corresponding effects on shear plane geometry. The results are potentially useful to facilitate a framework for process design of large strain deformation configurations, wherein transient deformation fields prevail. These implications are considered in the present study to quantify the efficiency of processing methods for bulk ultrafine-grained metals by large strain extrusion machining and equal channel angular pressing.

Oscillation dynamics of sessile droplets subjected to substrate vibration

Sessile droplets on a vibrating substrate are investigated focusing on axisymmetric oscillations with pinned contact line. Proper orthogonal decomposition is employed to identify the different modes of droplet shape oscillation and quantitatively assess the droplet oscillation and spectral response. We offer the first experimental evidence for the analogy of an oscillating sessile droplet with a non-linear spring mass damper system. The qualitative and quantitative agreement of amplitude response and phase response curves and limit cycles of the model dynamical system with that observed experimentally suggest that the bulk oscillations in the fundamental mode of a sessile droplet can be very well modeled by a Duffing oscillator with a hard spring, especially near the resonance. The red shift of the resonance peak with an increase in the glycerol concentration is clearly evidenced by both the experimental and predicted amplitude response curves. The influence of various operational parameters such as excitation frequency and amplitude and fluid properties on the droplet oscillation characteristics is adequately captured by the model. (C) 2014 Elsevier Ltd. All rights reserved.