A divide and conquer strategy for scaling weather simulations with multiple regions of interest

Accurate and timely prediction of weather phenomena, such as hurricanes and flash floods, require high-fidelity compute intensive simulations of multiple finer regions of interest within a coarse simulation domain. Current weather applications execute these nested simulations sequentially using all the available processors, which is sub-optimal due to their sub-linear scalability. In this work, we present a strategy for parallel execution of multiple nested domain simulations based on partitioning the 2-D processor grid into disjoint rectangular regions associated with each domain. We propose a novel combination of performance prediction, processor allocation methods and topology-aware mapping of the regions on torus interconnects. Experiments on IBM Blue Gene systems using WRF show that the proposed strategies result in performance improvement of up to 33% with topology-oblivious mapping and up to additional 7% with topology-aware mapping over the default sequential strategy.

Time variant reliability model updating in instrumented dynamical systems based on Girsanov’s transformation

The problem of updating the reliability of instrumented structures based on measured response under random dynamic loading is considered. A solution strategy within the framework of Monte Carlo simulation based dynamic state estimation method and Girsanov’s transformation for variance reduction is developed. For linear Gaussian state space models, the solution is developed based on continuous version of the Kalman filter, while, for non-linear and (or) non-Gaussian state space models, bootstrap particle filters are adopted. The controls to implement the Girsanov transformation are developed by solving a constrained non-linear optimization problem. Numerical illustrations include studies on a multi degree of freedom linear system and non-linear systems with geometric and (or) hereditary non-linearities and non-stationary random excitations.

Layerwise decomposition of water dynamics in reverse micelles: A simulation study of two-dimensional infrared spectrum

We present computer simulation study of two-dimensional infrared spectroscopy (2D-IR) of water confined in reverse micelles (RMs) of various sizes. The present study is motivated by the need to understand the altered dynamics of confined water by performing layerwise decomposition of water, with an aim to quantify the relative contributions of different layers water molecules to the calculated 2D-IR spectrum. The 0-1 transition spectra clearly show substantial elongation, due to in-homogeneous broadening and incomplete spectral diffusion, along the diagonal in the surface water layer of different sized RMs. Fitting of the frequency fluctuation correlation functions reveal that the motion of the surface water molecules is sub-diffusive and indicate the constrained nature of their dynamics. This is further supported by two peak nature of the angular analogue of van Hove correlation function. With increasing system size, the water molecules become more diffusive in nature and spectral diffusion almost completes in the central layer of the larger size RMs. Comparisons between experiments and simulations establish the correspondence between the spectral decomposition available in experiments with the spatial decomposition available in simulations. Simulations also allow a quantitative exploration of the relative role of water, sodium ions, and sulfonate head groups in vibrational dephasing. Interestingly, the negative cross correlation between force on oxygen and hydrogen of O-H bond in bulk water significantly decreases in the surface layer of each RM. This negative cross correlation gradually increases in the central water pool with increasing RMs size and this is found to be partly responsible for the faster relaxation rate of water in the central pool. (C) 2013 AIP Publishing LLC.

A study of the speed and the accuracy of the Boundary Element Method as applied to the computational simulation of biological organs

In this work, first a Fortran code is developed for three dimensional linear elastostatics using constant boundary elements; the code is based on a MATLAB code developed by the author earlier. Next, the code is parallelized using BLACS, MPI, and ScaLAPACK. Later, the parallelized code is used to demonstrate the usefulness of the Boundary Element Method (BEM) as applied to the realtime computational simulation of biological organs, while focusing on the speed and accuracy offered by BEM. A computer cluster is used in this part of the work. The commercial software package ANSYS is used to obtain the `exact' solution against which the solution from BEM is compared; analytical solutions, wherever available, are also used to establish the accuracy of BEM. A pig liver is the biological organ considered. Next, instead of the computer cluster, a Graphics Processing Unit (GPU) is used as the parallel hardware. Results indicate that BEM is an interesting choice for the simulation of biological organs. Although the use of BEM for the simulation of biological organs is not new, the results presented in the present study are not found elsewhere in the literature. Also, a serial MATLAB code, and both serial and parallel versions of a Fortran code, which can solve three dimensional (3D) linear elastostatic problems using constant boundary elements, are provided as supplementary files that can be freely downloaded.

Validation based sparse Gaussian processes for ordinal regression

This paper proposes a sparse modeling approach to solve ordinal regression problems using Gaussian processes (GP). Designing a sparse GP model is important from training time and inference time viewpoints. We first propose a variant of the Gaussian process ordinal regression (GPOR) approach, leave-one-out GPOR (LOO-GPOR). It performs model selection using the leave-one-out cross-validation (LOO-CV) technique. We then provide an approach to design a sparse model for GPOR. The sparse GPOR model reduces computational time and storage requirements. Further, it provides faster inference. We compare the proposed approaches with the state-of-the-art GPOR approach on some benchmark data sets. Experimental results show that the proposed approaches are competitive.

Signal Detection in Generalized Gaussian Noise by Nonlinear Wavelet Denoising

In this paper, a nonlinear suboptimal detector whose performance in heavy-tailed noise is significantly better than that of the matched filter is proposed. The detector consists of a nonlinear wavelet denoising filter to enhance the signal-to-noise ratio, followed by a replica correlator. Performance of the detector is investigated through an asymptotic theoretical analysis as well as Monte Carlo simulations. The proposed detector offers the following advantages over the optimal (in the Neyman-Pearson sense) detector: it is easier to implement, and it is more robust with respect to error in modeling the probability distribution of noise.

Role of the cloud adjustment time scale in simulation of the interannual variability of Indian summer monsoon

The simulation of precipitation in a general circulation model relying on relaxed mass flux cumulus parameterization scheme is sensitive to cloud adjustment time scale (CATS). In this study, the frequency of the dominant intra-seasonal mode and interannual variability of Indian summer monsoon rainfall (ISMR) simulated by an atmospheric general circulation model is shown to be sensitive to the CATS. It has been shown that a longer CATS of about 5 h simulates the spatial distribution of the ISMR better. El Nio Southern Oscillation-ISMR relationship is also sensitive to CATS. The equatorial Indian Ocean rainfall and ISMR coupling is sensitive to CATS. Our study suggests that a careful choice of CATS is necessary for adequate simulation of spatial pattern as well as interannual variation of Indian summer monsoon precipitation.

Monte Carlo simulation of light transport in tissue for optimizing light delivery in photoacoustic imaging of the sentinel lymph node

Noninvasive or minimally invasive identification of sentinel lymph node (SLN) is essential to reduce the surgical effects of SLN biopsy. Photoacoustic (PA) imaging of SLN in animal models has shown its promise for clinical use in the future. Here, we present a Monte Carlo simulation for light transport in the SLN for various light delivery configurations with a clinical ultrasound probe. Our simulation assumes a realistic tissue layer model and also can handle the transmission/reflectance at SLN-tissue boundary due to the mismatch of refractive index. Various light incidence angles show that for deeply situated SLNs the maximum absorption of light in the SLN is for normal incidence. We also show that if a part of the diffused reflected photons is reflected back into the skin using a reflector, the absorption of light in the SLN can be increased significantly to enhance the PA signal. (C) 2013 Society of Photo-Optical Instrumentation Engineers (SPIE)

System-level SoC near-field (NF) emissions: Simulation to measurement correlation

As System-on-Chip (SoC) designs migrate to 28nm process node and beyond, the electromagnetic (EM) co-interactions of the Chip-Package-Printed Circuit Board (PCB) becomes critical and require accurate and efficient characterization and verification. In this paper a fast, scalable, and parallelized boundary element based integral EM solutions to Maxwell equations is presented. The accuracy of the full-wave formulation, for complete EM characterization, has been validated on both canonical structures and real-world 3-D system (viz. Chip + Package + PCB). Good correlation between numerical simulation and measurement has been achieved. A few examples of the applicability of the formulation to high speed digital and analog serial interfaces on a 45nm SoC are also presented.

Reaction dynamics under confinement: an exact path integral treatment of a two-stage model of stochastic gene expression

Gene expression in living systems is inherently stochastic, and tends to produce varying numbers of proteins over repeated cycles of transcription and translation. In this paper, an expression is derived for the steady-state protein number distribution starting from a two-stage kinetic model of the gene expression process involving p proteins and r mRNAs. The derivation is based on an exact path integral evaluation of the joint distribution, P(p, r, t), of p and r at time t, which can be expressed in terms of the coupled Langevin equations for p and r that represent the two-stage model in continuum form. The steady-state distribution of p alone, P(p), is obtained from P(p, r, t) (a bivariate Gaussian) by integrating out the r degrees of freedom and taking the limit t -> infinity. P(p) is found to be proportional to the product of a Gaussian and a complementary error function. It provides a generally satisfactory fit to simulation data on the same two-stage process when the translational efficiency (a measure of intrinsic noise levels in the system) is relatively low; it is less successful as a model of the data when the translational efficiency (and noise levels) are high.

A study of the speed and the accuracy of the Boundary Element Method as applied to the computational simulation of biological organs

In this work, possibility of simulating biological organs in realtime using the Boundary Element Method (BEM) is investigated, with specific reference to the speed and the accuracy offered by BEM. First, a Graphics Processing Unit (GPU) is used to speed up the BEM computations to achieve the realtime performance. Next, instead of the GPU, a computer cluster is used. A pig liver is the biological organ considered. Results indicate that BEM is an interesting choice for the simulation of biological organs. Although the use of BEM for the simulation of biological organs is not new, the results presented in the present study are not found elsewhere in the literature.

Computationally Efficient Model for Simulation of Boost Converter

This paper presents a computationally efficient model for a dc-dc boost converter, which is valid for continuous and discontinuous conduction modes; the model also incorporates significant non-idealities of the converter. Simulation of the dc-dc boost converter using an average model provides practically all the details, which are available from the simulation using the switching (instantaneous) model, except for the quantum of ripple in currents and voltages. A harmonic model of the converter can be used to evaluate the ripple quantities. This paper proposes a combined (average-cum-harmonic) model of the boost converter. The accuracy of the combined model is validated through extensive simulations and experiments. A quantitative comparison of the computation times of the average, combined and switching models are presented. The combined model is shown to be more computationally efficient than the switching model for simulation of transient and steady-state responses of the converter under various conditions.

Comparative Evaluation of Switching and Average Models of a DC-DC Boost Converter for Real-Time Simulation

This paper presents a comparative evaluation of the average and switching models of a dc-dc boost converter from the point of view of real-time simulation. Both the models are used to simulate the converter in real-time on a Field Programmable Gate Array (FPGA) platform. The converter is considered to function over a wide range of operating conditions, and could do transition between continuous conduction mode (CCM) and discontinuous conduction mode (DCM). While the average model is known to be computationally efficient from the perspective of off-line simulation, the same is shown here to consume more logical resources than the switching model for real-time simulation of the dc-dc converter. Further, evaluation of the boundary condition between CCM and DCM is found to be the main reason for the increased consumption of resources by the average model.

A scaled unscented transformation based directed Gaussian sum filter for nonlinear dynamic system identification

Impoverishment of particles, i.e. the discretely simulated sample paths of the process dynamics, poses a major obstacle in employing the particle filters for large dimensional nonlinear system identification. A known route of alleviating this impoverishment, i.e. of using an exponentially increasing ensemble size vis-a-vis the system dimension, remains computationally infeasible in most cases of practical importance. In this work, we explore the possibility of unscented transformation on Gaussian random variables, as incorporated within a scaled Gaussian sum stochastic filter, as a means of applying the nonlinear stochastic filtering theory to higher dimensional structural system identification problems. As an additional strategy to reconcile the evolving process dynamics with the observation history, the proposed filtering scheme also modifies the process model via the incorporation of gain-weighted innovation terms. The reported numerical work on the identification of structural dynamic models of dimension up to 100 is indicative of the potential of the proposed filter in realizing the stated aim of successfully treating relatively larger dimensional filtering problems. (C) 2013 Elsevier Ltd. All rights reserved.

Capacity of a Gaussian MAC with energy harvesting transmit nodes

We consider a Gaussian multiple access channel (GMAC) where the users are sensor nodes powered by energy harvesters. The energy harvesters may have finite or infinite buffer to store the harvested energy. First, we find the capacity region of a GMAC powered by transmit nodes with an infinite energy buffer. Next, we consider a GMAC with the transmitting nodes equipped with a finite energy buffer. Initially we assume perfect knowledge of the buffer state information at both the encoders and the decoder. We provide an achievable region for this case. We also generalize the achievable region when only partial information about buffer state is available at both the encoders and the decoder.