Poster Images and Abstracts

See the winners of the 2010 CI Days Poster Session.

Posters below include the title, authors, keywords, and abstract.

Peri-operative Outcomes Initiative (POI): Using Clinical Systems for Research

AkkeNeel Talsma, PhD, Christine Anderson, PhD, Hyo Geun Geun, MPH, Shubhangi Gumate, BA, Margaret McLaughlin, MS, Irene Felicetti, MLS

Peri-operative Outcomes Initiative, Clinical Systems, Patient Safety

Patient safety research has evolved to study clinical processes that use data from electronic medical records (EMR). This study uses EMR data to identify health systems factors that are related to count discrepancy in the OR. Findings will be applied to the multi-hospital collaborative POI (funded). POI aims to improve peri-operative outcomes.

 


Design Framework for Real-Time Large-Scale, Parallel, Intelligent, CO2 Data Assimilation

V. Yadav, C.J. Antonelli, S.M. Gourdji, K.L. Mueller, A. Chatterjee, A.M. Michalak

Cyberinfrastructure, parallel computing, CO2 data assimilation

This presentation shows the results of applying distributed super-computing techniques for solving largescale linear batch atmospheric inversions to estimate CO2 fluxes efficiently. To overcome the computational limitations of performing these inversions, we propose the application of a parallel largescale data assimilation (DA) software system for global carbon emissions monitoring. Our design framework for a state-of-the-art autonomous software platform performs real-time integration of in-situ and satellite-based atmospheric CO2 measurements within a DA system for producing estimates of global land and oceanic CO2 exchange at weekly to bi-weekly intervals. This software system has an extendable modular design, with capabilities to crawl the web and autonomously ingest CO2 concentration data from multiple electronic sources in different file formats, simultaneous assimilation of observations using different DA algorithms for estimating global CO2 exchange, a first-of-its-kind carbon-climate surveillance system with the capacity to detect and analyze localized variations in CO2 exchange, and the ability to search for disruptive or unusual patterns that are not easily detected due to the massive scale and dimensionality of the data. The visualization component of the software system will provide both raw output and maps of the estimates of CO2 exchange and their associated uncertainties. The development of this system is being supported by a new project funded through the NSF Software Infrastructure for Sustained Innovation program. We hope to incorporate input from potential users throughout the development process, and welcome input either during the CI Days, or anytime thereafter by contacting Vineet Yadav (vineety@umich.edu) and/or Charles Antonelli (cja@umich.edu).

 


Effect of ACL laxity on ACL loading: dynamic simulation using impulsive loads

Jesal Parekh, Dr. Scott McLean, Dr. Mark Palmer

ACL, laxity, inter-segmental acceleration, computational model, finite element method

Background: Noncontact ACL injuries are among common, potentially traumatic and costly sport-related injuries, approximately 200,000 injuries occurring each year. Female athletes exhibit a trend toward higher rates of ACL injuries suggesting the influence of gender-specific characteristics. ACL laxity is compelling because it is one of the crucial differentiating factors between males and females. However, the mechanism through which ACL laxity implicates within an injury is unknown. Therefore, we tested the hypothesis that lax ACL fails to adequately restraint the knee during dynamic activities, resulting in higher inter-segmental accelerations and subsequently, higher strains in the ACL.

Methods: A series of computer simulations were executed on 2-D computational model of the knee, to replicate testing conditions of an in-vitro experiment (Withrow, 2006). Femur, tibia and patella were rendered in sagittal plane from MR images using Mimics® (Materialize, Belgium). The solid model was then imported into Hypermesh (Altair Engineering, USA) and meshed at a density of 0.001 (24990 elements total, 92 elements ACL). Boundary conditions and muscle definitions consistent with Withrow were applied. The bones were defined as linear elastic materials (E=1.5GPa, Poission’s=0.33, Density 286) and based on ligament behavior described by Bischoff et al (2008) ACL response was described by homogenized neo-hookean material law (C10=1.95MPa, D1=6.83GPa-1). The stiffness of ACL is determined by C10 coefficient, higher C10 values implying stiffer ACL. To simulate drop landing, an impulsive load was applied to the distal tibia for 100ms with a peak load of 1200N at 35ms. Abaqus (Simulia, USA) solver was used to obtain dynamic response. Multiple simulations were run each with a different level of ACL stiffness (laxity).

Results: The strains and the inter-segmental accelerations were computed using data extracted from the result files. The peak strain experienced by ACL decreases as laxity decreases. The inter-segmental horizontal acceleration also decreases as laxity decreases.

Discussion: The results indicate an important effect of ACL laxity. Inter-segmental horizontal acceleration is positively correlated with ACL laxity. Moreover, high inter-segmental horizontal accelerations result in the manifestation of elevated strains experienced by the ACL. Combined with the association between ACL laxity and inter-segmental horizontal accelerations, this enables one to elicit the mechanism and associated effects of ACL laxity on ACL strain under dynamic conditions.

 


Cyber-based Discovery of Efficient Metal-Organic Framworks for Carbon Capture

Hyun Seung Koh, Jinhyung Hwang, Donald J. Siegel

MOF, Carbon Capture, Adsorption

Physisorptive approaches to CO2 capture from flue gas have attracted increasing attention given their potential for moderate parasitic regeneration loads. At the low partial pressures typical of post-combustion capture, metal organic frameworks (MOFs) with coordinatively unsaturated sites (CUS) have demonstrated the additional benefit of extremely high CO2 capacities. However, as the number of MOFs is large and rapidly growing, it is not obvious that state of the art materials depict optimal performance. Using Density Functional Theory calculations, we screen 30 metal-substituted CUS-MOFs isostructural to HKUST-1 and MOF-74 for their CO2 affinities. Consistent with experimental data, the calculations identify Mg variants as having amongst the highest CO2 affinities. Substitutions of other alkaline earths and selected transition metals were likewise found to yield large CO2 affinities.

 


Visualization of Engineering Operations Using Outdoor Augmented Reality

Suyang Dong, Vineet R.Kamat

Advanced Visualization

Visualization of engineering processes can be critical for validation of simulation models, and also increasing safety and productivity of field processes. Augmented Reality (AR) visualization blends real-world information with graphical 3D models to create informative composite views that are difficult to replicate on the computer alone. This poster presents a robust and general-purpose mobile computing framework that allows users to readily create complex AR visual simulations and visualize engineering information. The technical challenges of building this framework from the software and hardware perspective are described. The framework has been validated in several case studies, including the visualization of underground infrastructure for applications in excavation planning and control.

 


A Common Web Interface for Managing Empirical Game Theory Studies on a Cluster

Ben-Alexander Cassell, Michael P. Wellman

Empirical Game Theory

The Strategic Reasoning Group at the University of Michigan has long conducted studies in empirical game theory, the study of games derived from empirical observation, through the use of large simulation. Since games grow exponentially in the number of players and strategies, even relatively small games exhaust the computational capabilities of desktop computers. Accordingly, being able to leverage the local cluster, Nyx, has become increasingly important; however, since the perceive complexity of using Nyx is high, many graduate students choose instead to restrict their attention to smaller games rather than learn how to use Nyx. To alleviate this problem we have created an approachable web interface for scheduling arbitrary game simulations onto the Nyx cluster.

 


Using High Performance Computing to Study Electron Transfer in Photovoltaic Materials via Density Functional Theory

Heidi Phillips, Shaohui Zheng, Alex Hyla, Francis Devine, Eitan Geva, Barry Dunietz

Physical Chemistry

Light harvesting relies on photo-induced charge transfer. The silsesquioxane molecule has shown promising photo-induced charge transfer characteristics when functionalized with organic dye molecules, such as red-shifted emission spectra and enhanced two-photon absorption cross-sections. Electronic structure calculations are used in this study to elucidate charge transfer character of these molecular systems. Density Functional Theory (DFT) and Time-Dependent (TD) DFT are implemented in calculating the optimized geometry and excited states of each cube-chromophore system. We compare the TDDFT excitation energies calculated using B3LYP and range-separated hybrid functionals. Calculations were performed using the Q-Chem Program Package versions 3.2 & 4.0. We utilize three types of calculations: Geometry optimizations, frequencies of normal modes, and single point energy TDDFT. Calculations were run on the Gollum cluster, which consists of 222 computing nodes.

 


Cosmic Sky Machine for the Dark Energy Survey

Brandon Erickson

Simulation

This poster presents an outline of our plan to integrate a number of computational tasks that have historically been run manually on various computing resources. We aim to bring all of the computation (simulation) to teragrid, and make the results available to a wider collaboration via a web portal being developed by part of the collaboration based in Brazil. The future observed data will be available on the portal, so we want our simulated data to be available to the collaboration in the same way.

 


Development and Application of the Metscape Bioinformatics Framework for the Analysis and Visualization of Metabolites and Genes

Alla Karnovsky, Terry Weymouth, Tim Hull, Glenn Tarcea, Jinghua Xu, Charles Evans, Maureen Sartor, Barbara Mirel, Brian Athey, Gilbert S. Omenn, Charles Burant

Metabolomics/Bioinformatics

Metabolomics is a rapidly evolving field that holds promise to provide insights into genotype – phenotype relationships in cancers, diabetes, and other complex diseases. One of the major informatics challenges is providing tools that would link metabolite data with other types of high throughput molecular data (e.g. transcriptomics, proteomics), and incorporate prior knowledge of pathways and molecular interactions. We recently developed Metscape, a plug-in for Cytoscape (http://www.cytoscape.org/), that allows users to upload a list of metabolites with experimentally determined concentrations, identify genes and pathways and display them in the context of relevant metabolic networks. Metscape uses an internal relational database that integrates data from KEGG (http://www.genome.jp/kegg/) and EHMN (http://www.ehmn.bioinformatics.ed.ac.uk/). We present the new version of Metscape, which allows users to identify enriched pathways from expression profiling data, build and analyze the networks of genes and metabolites, and visualize the changes in the gene/metabolite data.

We have applied Metscape to characterize the diet-induced changes in beta-cells of pancreatic islets of female Zucker Diabetic Fatty Rat (ZFF), a well defined animal model of type 2 diabetes. Metscape provided evidence for rapid activation of macromolecular biosynthetic pathways, including cholesterol, triglyceride and nucleotide, after 48 hours of consumption of a high sucrose/high fat diet.

 


Exploring New Methods for Protecting and Distributing Confidential Research Data

Felicia LeClere, Bryan Beecher, Steve Burling, Stuart Hutchings

Social science, curation, data cloud, cloud computing

Our project will build and test a data storage and dissemination system for research data containing confidential information. Our system will eliminate the need for data producers to build their own secure computing environments. Cloud computing makes it possible to provide powerful, secure data analysis platforms on demand, relieving researchers of the traditional concerns of data security. Our system will collect information on each user with a web interface, and then produce a custom computing environment for each confidential data contract holder.

 


Dense Packings of Hard Tetrahedra

Amir Haji-Akbari, Michael Engel, Elizabeth R Chen, Aaron S Keys, Xiaoyu Zheng, Rolfe G Petschek, Peter Palffy-Muhoray, Sharon C Glotzer

statistical physics, applied geometry, Monte Carlo

Using Monte Carlo simulations, we observe that a fluid of hard tetrahedra undergoes a first-order phase transition to a dodecagonal quasicrystal, which can be compressed to a packing fraction of Φ = 0.8336. By compressing a crystalline approximant of the quasicrystal, the highest packing fraction we obtain is Φ = 0.8503. If quasicrystal formation is suppressed, the system remains disordered, jams, and compresses to Φ = 0.7858. Jamming and crystallization are both preceded by an entropy-driven transition from a simple fluid of independent tetrahedra to a complex fluid characterized by tetrahedra arranged in densely packed local motifs that form a percolating network at the transition. The quasicrystal that we report represents the first example of a quasicrystal formed from hard particles or from non-spherical building blocks. Our results demonstrate that particle shape and entropy can produce highly complex, ordered structures. Furthermore, we have discovered the densest known packing of regular tetrahedra with a density Φ = 4000 / 4671 = 0.856347. The packing is crystalline with a unit cell of four tetrahedra forming two triangular bipyramids (dimer clusters) and is cluster transitive. The densest packing is a special case of an analytical three-parameter family of dimer packings we have constructed. Numerical compressions starting from random configurations confirm that the reported packing is optimal at least for small systems with up to 16 tetrahedra. For all systems with four or more tetrahedra, we obtain packings with densities higher than the maximum packing density of spheres.

 


Solvent effect on the rate of hexatriene isomerization: A molecular Dynamics study

Surma Talapatra, Frank X. Vazquez, Roseanne Sension, Eitan Geva

molecular dynamics simulation, reactive flux correlation function, transition state theory

The ring opening reaction of 1,3-cyclohexadiene (CHD) is important because it is analogous to the reaction of 7-dehydrocholesterol to form pre-vitamin-D. Previous experimental studies on the multistep reaction reveal that the ring opening of CHD is followed by the isomerization of cis-1,3,5-hexatriene (Zt-HT) from cis to trans conformer (cZt-HT to tZt-HT). Detailed experimental studies on this last step reveal that the rate constants of isomerization in different alcohols and alkanes do not follow the dependence predicted by Kramers theory.

We have investigated the isomerization reaction using molecular dynamics simulations to develop a molecular level explanation for this interesting trend of reaction rates in alcohols and alkanes. Using a number of alcohols (methanol, ethanol, propanol, butanol) and a number of alkanes (cyclohexane, n-hexane, n-heptane, cycloheptane) as solvents, the isomerization reaction rate constant was calculated at various temperatures ranging from 280K to 320K. The rate constant was measured using the reactive flux correlation function and transition state theory. In order to explain the results seen, thermodynamic properties of the reaction were calculated, including changes in free energy, entropy and enthalpy as the cZt-HT goes to the transition state. Primary results, with methnol, propanol, ethanol and cyclohexane, n-hexane, n-heptane, cycloheptane as solvents, reveal that the rate of isomerization in alcohols is indeed slower than in alkanes. The calculated entropy changes indicate that the faster reaction rates in alkanes can be attributed to structural difference brought about by the fact that the alcohol solvents form protic liquids with a relatively rigid hydrogen-bonded network, while alkane solvents form aprotic liquids so that solvent molecules can access the close vicinity of the solute.

 


GPU Accelerated Molecular Dynamics Algorithms for Soft Matter Systems using HOOMD-Blue

Carolyn L. Phillips, Joshua A. Anderson, Sharon C. Glotzer

Soft Matter Physics, High Performance Computing

The rheological, thermodynamic, and self-assembly behavior of liquids, colloids, polymers, foams, gels, granular materials and biological systems are often studied in simulation by using coarse-grained models based on molecular dynamics algorithms. The open source general purpose particle dynamics code HOOMD-Blue has been expanded to include the simulation techniques and pair potentials used to study this class of problems.

 


Design of Patchy Particles using Ternary Self-Assembled Monolayers

Ines Pons-Siepermann, Sharon C. Glotzer

Nanomaterials, self-assembly, patchy particles

Recent simulations have studied the formation of patterns in a binary mixture of immiscible surfactants adsorbed on the surface of a spherical nanocolloid [1]. The resulting patterns (Janus, patches, and stripes) were in good agreement with experimental results [2]. Here we perform dissipative particle dynamics (DPD) simulations to study the patterns obtained by increasing the number of surfactants in the monolayer. The simulations resulted in a variety of new patterns that can be produced through different combinations of simple design elements, like nanocolloid size, degree of surfactant immiscibility, stoichiometry of the mixture, and length difference between surfactants. In all cases, the formation of patterns is driven by the competition of enthalpic losses and entropic gains at phase boundaries.

[1] C. Singh et al. Physical Review Letters 99, 226106 (2007)

[2] A.M. Jackson et al. Nature Materials, 3, 330-336 (2004)

 


Visualizing the Audibility of Sound Fields – An Integrated Computation Method for Room Acoustics Analysis

Sentagi S. Utami

room acoustics, acoustical imaging, computer-simulation, beam-forming, virtual environment

Computer-simulations and field measurements are common methods use to evaluate the quality of room acoustics. The acoustical conditions are interpreted from objective and subjective indices. Performance judgment utilizing human ears is as important as the digital data interpretation obtained from the objective measurement.

This poster presentation demonstrates the ability to visualize sound fields of real and simulated spaces by using integrated and advanced computation methods. Techniques applied within this research utilizes high performance computing for measurement and analysis of large data set as related to signal processing for generating objective indices, auralizing and acoustical imaging of the sound fields.

A system of multi-microphones arrays based on beam forming with acoustical imaging algorithm is utilized for field measurements. Alteration of the physical elements of simulated spaces is using computer-simulations based on geometrical room acoustics. Auralization is utilized to create digital sound that synthesize auditory event as if the human is listening inside the simulated space. Subjective evaluation utilizes a simulation technique that assimilates real-time auditory cognitive experience into computer-simulation data, which combines auralization of computer-simulated spaces and the Virtual Environment (VE) system.

Further application of this approach is the ability to synthesized auditory event while listeners physically are presence inside the virtual scene and evaluate the acoustical conditions at any stage of the design and therefore, practitioners in architectural acoustics do benefit from it.

 


North Campus Research Complex Photovoltaic Systems Potential Analysis

Chanikarn Yimprayoon

Building environmental technology/ Grid-connected photovoltaic systems, solar energy, commercial buildings, building simulation

North Campus Research Complex (NCRC), recently purchased from Pfizer, consists of 1.97 million square feet of laboratory and administration space in thirty main buildings. A total flat roof area of 0.8 million square feet makes the complex amenable for rooftop grid-connected solar cell or photovoltaic (PV) systems. This study quantifies a potential deployment of PV grid connected systems on a commercial building complex as well as demonstrates the available technology and tools that can be used to accelerate the analysis of PV systems performance. PV output calculation is done by using an online tool from National Renewable Energy Laboratory (NREL) called PVWATTS. The results show that if PV horizontal fixed systems are installed over 60% of roof areas and all buildings are modeled as if they are office buildings, the electricity outputs would account for approximately 43% of electricity loads and can significantly reduce electricity peak load demand during the hottest month. After PV systems are installed, the effective of building energy management can be maximized by the use of another online tool called SolarAnywhere which is a satellite based weather monitoring service. Building automation systems can incorporate this information with electricity tariffs and decide what should be the best strategies to manage the PV outputs. Utilities can also use this information to maintain their electricity supply.

 


Ab initio modeling of intrinsic defects in lithium oxides: relevance for Li-air batteries

Maxwell D. Radin

energy sotrage, computational materials science

Li-air batteries are a potentially transformative technology that may offer an order of magnitude increase in energy density over state of the art Li-ion batteries. However, current rechargeable Li-air batteries suffer from low efficiency due to high potentials during charging. Although some transition metal catalysts have been found to increase efficiency, the decomposition process and catalytic mechanisms involved are poorly understood. We hypothesize that the catalysts facilitate decomposition by lowering energy barriers to defect formation/migration in bulk lithium peroxide. The purpose of the project is to use first principles calculations to screen for new catalysts that can increase the efficiency of Li-air batteries. We use density functional theory (DFT) to calculate the thermodynamics and kinetics of defects in bulk lithium peroxide. Preliminary results indicate that lithium and oxygen vacancies are the dominant defects under atmospheric conditions.

 


Mixed Quantum-Classical Approach in the Vibrational Spectroscopy of Methanol-d in Carbon tetrachloride

Kijeong Kwac, Eitan Geva

Theoretical chemistry, simulation

Hydrogen bonds (H-bonds) play an important role in many chemical systems. Methanol is the simplest alcohol which is a highly structured H-bonded liquid at room temperature. The OH stretch mode of methanol is a very sensitive probe of neighboring solvation environment. In this work, we report the mixed quantum-classical molecular dynamics simulation study of the MeOD/CCl4 mixture. We calculated the IR absorption lineshape of the OD stretch mode and compared with the experimental result. Using the result from the mixed quantum-classical simulation, we establish the relation which can map from the electric field along the OD bond to various quantities such as OD bond length and OD stretch frequency.

 


Simulated Observations of the Universe

Brian Nord

Physics, Cosmology, Simulation, Clusters

In order to investigate the accelerating expansion of the universe (caused by dark energy) and its effects on large-scale structure, we use gravity and hydrodynamic simulations on cosmological scales. Figures 1 and 2 below summarize the state of the art of cosmology research. See the poster by Brandon Erickson, “Cosmic Sky Machine for the Dark Energy Survey,” for more background information.

Simulations are useful not only for obtaining a complete view of large-scale structure. Future cosmological constraints will rely on simulated observations: we must test the robustness of the connections drawn between observations and the underlying matter distribution. To this end, we focus on virtual sky surveys constructed from simulated lightcones.

 


Electromagnetic Analysis of Plasma Engulfed Re-Entry Vehicles via a Hybrid FE-BI-VIE Approach

Xi Lin, Onur Bakır, and Eric Michielssen

High-performance computing, Parallel Computing, electromagnetics

Space vehicles often are affected by communication blackout upon re-entering the Earth’s atmosphere. The blackout arises when the vehicle interacts with the atmosphere around it, giving rise to dense plasmas that are impenetrable by electromagnetic waves. The vehicle itself often is covered in a thin and inhomogeneous plasma shell, the density of which decreases rapidly with distance from the vehicle surface. This plasma shell hinders the operation of antennas mounted on the side of the vehicle. As the vehicle moves through space, it also leaves behind a large plasma plume. This plume hinders the operation of antennas mounted on the back of the vehicle. The nature and density of the plasma shell and wake heavily depend on operational and environmental conditions and vary rapidly with the vehicle’s position along its trajectory. To analyze the occurrence of communication blackout and facilitate the design of robust navigation systems, fast simulators capable of accurately characterizing the operation of antennas mounted on plasma-engulfed vehicles are called for.

This work presents a full-wave hybrid technique that addresses existing challenges in analyzing scattering and radiation from plasma-engulfed space vehicles. Different solvers are used in the plasma shell and the wake region. A domain decomposition – finite element (FE) solver is used to model electromagnetic fields in the plasma near the vehicle as this permits the analysis of highly inhomogeneous plasma distributions near the vehicle. A fast Fourier transform-accelerated volume integral equation (VIE) solver is used to model electromagnetic fields in the plasma wake behind the vehicle; this choice of solver is motivated by the (relatively) slow variations of the plasma parameters in the wake, and the computational benefits associated with accurately modeling wave propagation over large distances using integral as opposed to differential equation methods. The FE and VIE solvers are coupled via a boundary integral (BI) and the resulting hybrid equations are solved iteratively. Various computational results that demonstrate the efficiency, accuracy, and modeling versatility of this full-wave hybrid FE-BI-VIE technique will be presented.

 


Stochastic Electromagnetic Analysis via High Dimensional Model Representations

Abdulkadir C. Yücel, Hakan Bagcı, and Eric Michielssen

Computational Electromagnetics

The recent literature abounds with studies aimed at statistically characterizing electromagnetic observables given uncertainty in geometry descriptions and material parameters, as well as excitations. Often, this characterization is achieved via Monte Carlo (MC) methods, which call for the execution of a deterministic electromagnetic simulator for many realizations of the uncertain/random variables sampled with respect to their (assumed/known) probability distribution functions (pdfs). While MC methods are straightforward to implement and readily generate important statistical information (e.g., moments and pdf of observables), their convergence often is prohibitively slow. Stochastic collocation (SC) methods that use generalized polynomial chaos (gPC) expansions to represent observables are not unlike MC methods in that they only require the repeated execution of a deterministic simulator. Even though SC methods are more accurate than MC methods, they become computationally expensive (i) when the observables are rapidly varying and/or discontinuous functions of the random variables and (ii) when the number of random variables is high.

Here, an extension to SC methods that addresses the second of the above shortcomings is presented. The proposed extension leverages high dimensional model representations (HDMRs), viz. hierarchical representations of observables involving sums of so-called multinomial component functions. The lowest-order component functions reveal the “independent” contributions of the random variables while higher-order ones reveal combined contribution of random variable groups. For an observable that only weakly depends on high-order correlations of the random variables, the number of participating component functions can be kept small with little or no effect on the accuracy of the HDMR. The HDMR is constructed iteratively, starting with low-order component functions and only including those high-order ones that “significantly” enhance the accuracy of the representation. The component functions that feature in the HDMR are approximated using adaptive SC-gPC expansions. When compared to classical SC-gPC methods, the proposed hybrid HDMR-SC-gPC approach often permits more accurate expansions with far fewer terms. The efficiency and accuracy of the proposed method will be demonstrated via its application to the statistical characterization of crosstalk on interconnects.

 


Stochastic Characterization of Wave Propagation in Mine Environments

Onur Bakir, Abdulkadir Yücel, Hakan Bagcı, and Eric Michielssen

Electrical Engineering, Computational Electromagnetics, Domain Decomposition, Uncertainty quantification, Wave propagation, Underground Mines

Wireless communication systems are key to ensuring mine safety because of their potential to continue operating during catastrophic events when wired communication systems fail (Miner Act 2006). The design of wireless communication systems for mine environments requires simulation tools capable of analyzing electromagnetic wave propagation through electrically large mine tunnels and galleries. Ideally, these tools should account for the presence of miners, mining equipment, trolleys, and rails, as well as the roughness on the walls. Furthermore, they must yield statistical information of certain observables (e.g., probability distribution function (pdf) of received power) required for wireless channel design, given the uncertainty in mine geometry, the position and shape of obstacles, material and wall roughness properties, and transmitter and receiver locations.

In this work, an efficient stochastic electromagnetic simulation framework that addresses the above challenge is proposed. The proposed framework hybridizes (i) a deterministic electromagnetic simulator for modeling wave propagation in mine environments with (ii) a stochastic collocation (SC) method to extract pertinent statistical data.

The SC method approximates the observable using a generalized polynomial chaos expansion (gPC). The coefficients of the gPC expansion are obtained from inner products, which are computed on a sparse grid constructed using the Smolyak algorithm. At integration (collocation) points, the deterministic electromagnetic simulator is executed to compute the exact value of the observable. Therefore, this method is non-intrusive and straightforward to implement. The approximate observable values obtained from the resulting gPC expansions are then used to replace the actual observable evaluations (i.e. the costly deterministic electromagnetic simulations) when extracting observable pdfs via Monte Carlo methods.

The scheme’s applicability will be demonstrated through the characterization of wireless communication links deployed in electrically long mine tunnels and galleries. The path loss distributions obtained using the proposed method will be compared to those known to apply in mine environments (Rice, Rayleigh, and lognormal).

 


Fully Localized High-Order Div- and Quasi-Curl-Conforming Basis Functions for Multiplicative Calderón Preconditioning of the EFIE

Felipe Valdes, Francesco P. Andriulli, Kristof Cools, and Eric Michielssen

Computational Electromagnetics/dense mesh breakdown, high-order basis functions, preconditioning

Method of moments (MoM)-based electric field integral equation (EFIE) solvers are widely used for analyzing time-harmonic electromagnetic scattering from perfect electrically conducting (PEC) surfaces. That said, MoM-EFIE solvers are no panacea. Indeed, the spectral properties of the EFIE operator render MoM system matrices arising from dense meshes highly ill-conditioned. Recent efforts at preconditioning MoM-EFIE solvers by using Calderón identities in conjunction with zeroth-order Buffa-Christiansen (BC) basis functions and their higher-order extensions have been shown to largely remedy this problem.

Unfortunately, the high-order BC constructs developed to date are not local; that is, unlike their zeroth-order counterparts, their support includes the entire scatterer surface. Here, the concepts presented previously in the high-order BC constructs are used to generate a set of completely localized high-order div- and quasi curl-conforming basis functions. The fact that these functions are localized allows for them to be pre-computed and stored in memory during the iterative solution process, resulting in significant CPU time and memory reductions. Numerical results show that the proposed localization procedure does not adversely affect the functions’ preconditioning properties.

 


On the Regularization of Single Source Combined Integral Equations for Analyzing Scattering from Homogeneous Penetrable Objects

Felipe Valdes, Francesco P. Andriulli, Hakan Bagci, and Eric Michielssen

Computational Electromagnetics/dense mesh breakdown, single source integral equations

The literature abounds with integral equation techniques for analyzing scattering from homogeneous penetrable objects. Dual source techniques, which are by far the most popular, solve a coupled pair of electric, magnetic, or mixed/combined field integral equations for electric and magnetic surface currents. Single source techniques on the other hand, solve one electric, magnetic, or combined field integral equation for an electric or magnetic surface current density.

Equations of the first kind involve hypersingular operators which lead to ill-conditioned matrices when discretized, therefore are susceptible to dense mesh and low frequency breakdown. Moreover, they exhibit resonances; that is, their solution is not unique at a set of discrete frequencies that grows increasingly dense as the electrical size of the scatterer increases. Second kind equations on the other hand, do not suffer from dense mesh nor low frequency breakdown, but they are still susceptible to resonances and hence problematic when applied to the analysis of electrically large scatterers.

A linear combination of first and second kind single source equations has been proposed, yielding a resonant free formulation. Unfortunately, this equation still contains a hypersingular electric field integral operator rendering the entire equation hypersingular and susceptible to dense mesh breakdown.

Here we present a second kind single source equation for analyzing scattering from homogeneous penetrable objects which is resonance free and that is not susceptible to dense mesh or low frequency breakdown.

 


Hertzian Insurgency: Tectonic Recipes for Frequency Inhabitation

Sara Dean

Architecture

Through the development of interventions and narratives, frequency space is activated as a site its inhabitable potential is uncovered and exploited. The hertzian landscape here is seen less as an interface and more as a physical location of interaction and altered condition. This is an important actionable site both as part of an understanding of the world in which we live and the site conditions under which we are working. Through these investigations we can continue to develop a knowledge base for how to better engage the opportunities of the hertzian landscape so that it can be better integrated into our built world.

 


Atomistic Modeling of Materials for High Capacity Lithium-Air Batteries

Jill F. Rodriguez, Donald J. Siegel

Density Functional Theory, Parallel Computing

Rechargeable lithium batteries have become an integral part of portable electronics. Upon recharge, the vast majority of these devices operate via removal of Li+ from an LiCoO2 intercalation electrode before proceeding across the electrolyte and being inserted between the graphene sheets in the graphite anode. Current research efforts on cathode materials focus on identifying alternative cathode materials possessing higher capacities, with the hope of increasing the energy density by a factor of at least two. A less explored field replaces the LiCoO2 with a porous electrode, allowing Li+ to react with O2 found in air (Figure 1). Li-air batteries have the potential to increase the specific energy density by a factor between five and ten (Table 1) because one of the reactants need not be stored in the cell; rather the supply of O2 would come from an air inlet, and react via 2Li + O2 -> Li2O2. Because studies addressing the O2 electrode have been limited and focused on discharging (Li2O2 has been shown to be the dominant product in this process), research is needed in the field of non-aqueous based Li/O2 batteries with the ability to recharge. Such exploration will help us to understand these processes and optimize performance. Prior research has demonstrated that some of the obstacles facing Li-air batteries include: dendrite formation on the lithium anode; slow rates of discharge are currently required to achieve high capacities; sensitivity to impurities in the oxygen stream; and a high potential required for recharging. Our research aims to improve the rechargeability and capacity of Li-air cells. This is due to the much higher potentials required for recharging compared to the (lower) potential present during discharge. Various cathode catalysts have been shown to improve this efficiency by lowering the potential for recharge, but the mechanism by which they work is not understood (Figure 2).

 


A Focal Transcranial Magnetic Stimulator (TMS)

Luis Gomez, Anthony Grbic, Luis Hernandez, Frantishek Cajko, Eric Michielssen

Electromagnetism, Bioelectromagnetism, Transcranial Magnetic Stimulation, Biomedical Engineering

Transcranial magnetic stimulation (TMS) is a technique for studying brain function and treating neurological disorders. TMS holds significant promise as a tool for cognitive neuroscience as well as psychiatric research in the areas of depression and obsessive compulsive disorder. In TMS, one or more coils carrying time varying current pulses located near the scalp generate magnetic fields inside the head that in turn induce electric fields and eddy-currents inside conductive brain tissue. Whenever a nerve fiber is aligned with the induced electric field, a current is produced in the axon, which in turn depolarizes its membrane.

Unfortunately, TMS devices stimulate large lateral regions of tissue near the scalp. Numerous attempts have been make to design TMS coils capable of delivering focused magnetic fields deep into the brain. For example the ‘figure-8′ coil, which is composed of two circular coils side-by-side, was designed to create a more compact excitation than a single coil. The ‘H-coil’ which works by inducing primary electric fields which are mostly tangential to the scalp is able to excite deeper regions of the brain. The purpose of the current study is to introduce Negative Refractive Index Metamaterial (NRI-MM) near-field lenses as a way to improve the targeting and penetration depth of TMS coils.

Recently, it was shown that NRI-MM lenses with can be used to focus magnetic fields in the near-field. MM’s are structures made up sub-wavelength conductive and dielectric sub-structures arranged in a way that they emulate an effective permittivity of choice for a range of frequencies. Our results show that TMS setups leveraging NRI-MM lenses are able to generate more compact excitations for any penetration depth than systems composed of TMS coils alone.