Julia Gelfand Department: UCI Libraries
Julia Gelfand Department: UCI Libraries
Data collection, review and analysis is part of basic research in nearly every discipline. The kind of data that is collected and how it is reported varies within the discipline. By definition, data generally "refers to a collection of natural phenomena descriptors including the results of experience, observation or experiment, or a set of premises. This may consist of numbers, words or images, particularly as measurements or observations of a set of variables.? How social scientists or scientists apply data to test hypotheses and make conclusions is dependent on customs within a discipline. Increasingly, there are data archives that support data research in the social and behavioral sciences and increasingly in the sciences. These resources can be characterized as descriptions of certain activities such as voting behavior, demographic and census concentrations, consumer behavior, genetic mapping and are a few of the more common examples of available data sets. Government agencies, academic and scholarly publishing and commercial sources compile such primary and secondary data resources made available for scholarly interpretation. Libraries commonly administer access and assistance in identifying appropriate data for a range of research needs. Data inclusion helps to confirm textual statements and can be enhanced visually in graphic formats, mapped and used to better describe certain activities. Library services and resources that support Social Science Data at the UCI Libraries can be found at http://www.lib.uci.edu/online/subject/subpage.php?subject=socsci_data
Jennifer Shockro Collaborators: Aaron P. Freeman, Siarhei Piatrovich, and Fazlul R. Zubair Faculty Advisor: Haris J. Catrakis Department, School: Mechanical & Aerospace Engineering, Henry Samueli School of Engineering
We present research results on a method we have developed in our group which enables Monte Carlo Computing (MCC) for characterizing multiscale images. The MCC method provides a purely meshless computational approach to the evaluation of the fractal dimension of images, thus eliminating the need for grids or boxes associated with conventional techniques such as box-counting. The basic idea of the MCC method is to utilize point locations which are randomly located within a reference region that contains an object or feature of interest in a multiscale image. The shortest distance from each such randomly-chosen location to the nearest part of the object of interest is computed. These shortest distances provide scales which probe the multiscale geometrical features of the object. One of the most basic and useful parameters of geometrical characterization of image features is the fractal dimension. The present MCC method provides the fractal dimension based on the distribution of shortest-distance scales. By computing the probability density function of the shortest-distance scales and utilizing mathematical theory we have developed in our group, we compute the fractal dimension. Results of the MCC method are presented on sample images of interfaces derived from high-resolution imaging in turbulent flows. The MCC method is a general computing method and thus is expected to have broad applicability in a variety of practical problems such as computer graphics, morphology, image detection, and feature-based searching.
Project URL: http://ideaspages.googlepages.com/mccgroup.html
Ryan C. Sokolowski Collaborators: Josep Salvans-Tort, Adam J. Wachtor, Jennifer Shockro, and Aaron P. Freeman Faculty Advisor: Haris J. Catrakis Department, School: Mechanical & Aerospace Engineering, Henry Samueli School of Engineering
We present research results on fundamentals and applications of Smoothed Particle Computing (SPC) which is a tool useful for solving equations of motion by purely meshless simulations. Also known as Smoothed Particle Hydrodynamics (SPH), this tool enables computations without the need for a grid or mesh. The elimination of the need to generate a grid or mesh is an important reduction in computational cost for applications. The SPC tool solves the equations of motion by representing the field of interest, such as mass, in terms of smoothed particles and computes solutions along characteristics given by the smoothed particle trajectories. An important aspect in terms of achieving purely meshless capabilities of SPC is that the initial configuration of the smoothed particles must not be constrained to be regular. It is essential therefore to be able to generate irregular configurations of smoothed particles and we explore methods for generating irregular distributions of particles. Both for the initial configuration of particles as well as for the dynamical evolution of the particle locations, it is important to monitor computationally the degree of particle clustering vs. particle spreading. For this reason, we explore computational-mathematics measures of particle clustering vs. spreading and we examine computationally the effect of initial smoothed particle irregularity or disorder on the accuracy of SPC results on a model problem of fluid motion consisting of a two-dimensional unsteady array of vortices.
Project URL: http://ideaspages.googlepages.com/spcgroup.html
Yong Xiao Collaborators: Chen Jin and Scott Klasky (ORNL) Faculty Advisor: Zhihong Lin Department, School: Physics and Astronomy
Fusion plasma research is the key research that can secure the energy supply for our future. Due to the extreme complexity of fusion plasma, modern massive parallel computing has been introduced in this field as an indispensable tool. As one of the biggest users of today's national leadership supercomputing resources, our simulation code, Gyrokinetic Toroidal Code(GTC), demonstrates excellent scalability up to tens of thousands processors. The GTC code has made a series of innovations in algorithm, parallelization, and file I/O to reach its optimal performance. These innovations will be illustrated in detail.
J. Alfredo Freites Collaborators: Douglas Tobias and Stephen H. White Faculty Advisors: Douglas Tobias and Stephen H. White Department: Physiology and Biophysics
Interactions between polypeptide chains and lipid bilayers play a crucial role in membrane protein (MP) assembly and stability. Atomistic molecular dynamics (MD) simulations can provide unique insights into the nature and the spatial distribution of these interactions, but due to the complexity of the dynamic protein-membrane interface, a straightforward analysis is uninformative. We present a methodology for the analysis of all-atom MD trajectories, based on Voronoi tessellations (VT), that allows the unambiguous characterization of MP interactions in a lipid bilayer environment on a per residue basis. The method is suitable for the study of large simulation systems with total number of atoms on the order of 100,000. For a given system configuration, the VT algorithm is applied to a subset of sites consisting of the geometric center of each amino acid residue, and simplified representations of the lipid and water molecules in the neighborhood of the protein. Once the VT is complete, the neighborhood of each amino acid residue is identified and parsed as needed. The procedure is repeated along the MD simulation trajectory, yielding a dynamical description of the neighborhood of each amino acid residue. The results are useful for discussing MP structure and assembly in the context of experimental results aimed at identifying MP interfaces in a membrane environment, as well as MP transmembrane topology and structure predictions. We will illustrate the method by analyzing MD trajectories for the LacY symporter embedded in a lipid bilayer in excess water.
Harindar S. Keer[1] Collaborators: Lorant Janosi[3,4], Ioan Kosztin[3] Faculty Advisor: Thorsten Ritz[2] Department, Schools: [1]Dept. of Chemistry, University of California, Irvine, Irvine, CA, USA, [2]Dept. of Physics and Astronomy, University of California, Irvine, Irvine, CA, USA. [3]Dept. of Physics and Astronomy, University of Missouri, Columbia, Columbia, MO, USA, [4]Dept. of Chemical and Biomolecular Engineering, University of Houston, Houston, TX, USA
Light harvesting complexes 2 (LH2) from Rhodospirillum (Rs.) molischianum and Rhodopseudomonas (Rps.) acidophila form ring complexes out of eight or nine identical subunits, respectively. Here, we investigate computationally what factors govern the different ring sizes. Starting from the crystal structure geometries, we embed two 45° for Rs. molischianum and at about 38.5° for subunits of each species into their native lipid-bilayer/water environment and investigate their dynamics by means of both equilibrium and non-equilibrium (steered) molecular dynamics simulations. For each species, by employing the umbrella sampling method, we calculate and compare the free energy profiles, as a function of the angle between the two subunits, for two representative separations of the subunits. We find that two subunits prefer to arrange at distinctly different angles, depending on the species, at about 45° for Rs. molischianum and at about 38.5° for Rps. acidophila, which is likely to be an important factor contributing to the assembly into different ring sizes. The calculation technique employed, represent, arguably, the most accurate calculations possible on a system of this scale. We incorporated a modality to perform Umbrella Sampling in widely used parallel MD simulation software called NAMD. And, that allowed us to increase efficiency of our computing resources, 32 processors Beowulf cluster, by up to a factor of 2.
Craig Brown Faculty Advisor: Walt Scacchi Department: Institute for Software Research, School of Information and Computer Sciences
This presentation will focus on current effort to develop a computer game-based simulator for training technicians in how to diagnose potentially hazardous material/liquid spills within a nano-technology or semiconductor fabrication facility. Previous industrial efforts have found that interactive, training simulators for complex manufacturing systems can cost millions of dollars to develop and deploy. This effort demonstrates how a commercially available "first person action" game can be modified to create a virtual world of semiconductor fabrication that can be used for training technicians in realistic material/liquid spill diagnosis within a variety of hazard scenarios at a very low cost, as well as how such a game-based simulator can be used to support remote diagnosis of spill hazards in remote fabrication facilities that can be accessed over the Internet.
Chris Jensen Collaborators: Faculty Advisor: Walt Scacchi Department: Institute for Software Research, School of Information and Computer Sciences
This presentation will focus on current effort to develop tools and techniques for computer-aided discovery of software development processes employed in remote, open source software projects. Many research groups in different disciplines are now considering how best to exploit emerging free, open source software packages, but with little understanding of how these software are being developed and evolved over time, nor how to contribute to their ongoing development. This research seeks to facilitate such understanding through demonstration of a new suite of computer-based tools that can be used to acquire empirically observable processes of software development found in a selection of open source software projects.
Yuzo Kanomata Collaborators: Alex Szeto, Robert Nideffer Faculty Advisor: Walt Scacchi Department: Institute for Software Research, School of Information and Computer Sciences
This presentation will focus on current effort to setup a Web compatible, Flash media server using open source software. Such a media server support two key kinds of functionality. First, is support for zero configuration Web video conferencing (or video multi-casting), and video broadcasting. Second, is as a back-end server for streaming data event streams for massively multiplayer online games built in Flash. An open source software server offers a low-cost alternative to proprietary Flash servers, which limit scalability by concurrent user license cost. This presentation will highlight some of the issues involved in acquiring and evaluating alternative Flash media servers, as well as experience gained in applying such servers to realized the two desired kinds of massively scalable Flash media services.
Dean Dauger, Ph.D.
Collaborators: Advanced Cluster Systems, LLC,
Dauger Research, Inc.,
UCI Network & Academic Computing Services (NACS),
Wolfram Research, Inc.
Department: NACS
Execution of your algorithms in Mathematica can now be cut from hours to minutes or even seconds. The Supercomputing Engine for Mathematica enables Wolfram Research's Mathematica to be combined with the programming paradigm of today's supercomputers, but within the Mathematica computing environment, utilizing clusters and multicore to easily harness parallel execution with unprecedented flexibility and scalability. Closely following the industry-standard Message- Passing Interface (MPI), this new toolkit creates a standard way for every Mathematica kernel in the cluster to communicate with each other directly while performing computations. In contrast to typical grid implementations of Mathematica that are solely master-slave, this patent-pending solution instead has all kernels communicate with each other directly and collectively the way modern supercomputers do. We introduce how UCI faculty, researchers, and students can access SEM on the ACS/DR-SEM cluster, housed at UCI's Henry Samueli School of Engineering data center, and apply this new supercomputing technology today.
Project URL: http://daugerresearch.com/pooch/mathematica
+--------------------------------------------------------------------+ | 10 Minute Break | +--------------------------------------------------------------------+
Ana Nicoleta Bondar Collaborator: Douglas J. Tobias Faculty Advisor: Stephen H. White Department, School: Physiology and Biophysics, School of Medicine
Intramembrane proteases are a family of membrane proteins that cleave other transmembrane protein helices. The action of the intramembrane proteases is essential for cellular processes such as signalling and gene regulation, and has also been implicated in human diseases such as Alzheimer's. The rhomboid intramembrane protease uses a helix- gating mechanism to dock the cleavage substrate. Understanding how rhomboid protease works will shed light on the general principles of protein/lipid membrane interactions, and is an essential requirement for designing drugs. Computer simulations are critical for deriving an atomic-detail picture of the rhomboid protease in its native membrane. We have examined the dynamics and lipid interactions of the rhomboid protease in two different lipid membrane environments. The simulations provide a detailed understanding of the dynamics of the lipid hydrogen bonding of the rhomboid protease, and help explain the influence of lipids on the catalytic activity observed experimentally. To assess the role of specific structural elements known to be critical for the enzyme activity we examined the dynamics of two mutant phenotypes of rhomboid protease. The mutations cause changes in the orientation and dynamics of the protein and suggest the existence of a relay mechanism that connects remote regions of the protein.
Roy Culver (culverr@uci.edu) Advisor: Professor Feng Liu Department: Mechanical & Aerospace Engineering
Turbomachines are widely used in devices such as jet engines, power generators for large ships, and land based power generators, One reason they are widely used is that they deliver relatively high energy efficiency. Since some estimate that over 7% of the global oil consumed in 2004 was used in jet engines alone, detailed study of these machines, for the purpose of increasing efficiency has the potential to make significant global financial and political impacts. Although there are several factors contributing to efficiency, considering the associated fluid dynamical problem only, we have a set of equations which govern; the Navier-Stokes equations. To solve these, we convert them to a discrete form so that the solution will satisfy the original equation at a finite number of points. Solving these equations directly, we would typically need about 8000 TB of ram! Even if we had that, to do one addition of these numbers using the most powerful computer in the world would take about 10 seconds. The only choice is to further simplify the problem. After simplifying the problem suficiently, a very important part of making this problem solvable (on a useful time scale) is parallel programming. Here we split the discrete set of points into "blocks" and distribute those across several computers, thereby lessening the memory and computational load on each one. The result of all of this work is a detailed picture of the fluid behavior in the turbo- machine which is indispensable in the diagnosis of sources of inefficiency in turbomachine design.
Humberto A. Gallegos Collaborators: Jochen E. Schubert Faculty Advisor: Brett F. Sanders, Ph.D. Department: The Henry Samueli School of Engineering, Civil and Environmental Engineering
The Baldwin Hills Dam failure on December 14, 1963 in Los Angeles County was estimated to have caused more than $15 million dollars in property damage, the loss of the reservoir itself, and most importantly, the loss of five lives. An investigation by the California Department of Water Resources later revealed that the reservoir failed due to an apparent long-term movement of the reservoir's foundation which was located along faults which are planes of foundation weakness. Using software such as ArcGIS and Light Detection and Ranging (LiDAR) data obtained from the Los Angeles Department of Public Works; and, contour plates for both the Baldwin Hills reservoir and the dam's failure breach obtained from the California Department of Water Resources; reconstruction of the failure breach and the dam's terrain in 1963 was possible. Ultimately, the generated terrain for the 1963 dam and breach were used to complete a 2- dimensional, hydrodynamic simulation of the Baldwin Hills Dam failure using BreZo. BreZo is a hydrodynamic modelling tool which uses a Godunov-type finite volume scheme that solves the 2D shallow-water equations and runs on an unstructured grid of triangular cells. It can be used to simulate hydrodynamics in rivers, estuaries, and coastal waters for applications involving wetting and drying of uneven topography for dam-break flows, runup and overland flow resulting from tsunamis, flooding and draining of tidal wetlands, and flood flows down dry river beds. Using flood extent data from the US Army Corps of Engineers, model validation was performed to the modelling results.
Tony Soeller Department: NACS/Research Computing Support
UCI researchers engaged in High Performance Computing have free access to the supercomputing resources of the San Diego Supercomputer Center (SDSC) through the Academic Associates Program (AAP). Researchers can get accounts on supercomputers and a personal allocation of CPU time on each of the SDSC computers by making a simple application through the AAP. Computers available include the DataStar (an IBM terascale machine), TeraGrid (a nationwide system of high-performance computers, data, and research facilities), and BlueGene (an IBM system designed for data-intensive computing). Large data storage systems and a myriad of academic software packages are available at SDSC. Additionally SDSC offers online databases for research work, consulting services to assist researchers with programs run on the computers, as well as workshops such as the SDSC Summer Computing Institute.
Eric V. Schow Collaborator: J. Alfredo Freites Faculty Advisors: Douglas J. Tobias, Stephen H. White Departments: Physics, Institute for Genomics and Bioinformatics
Voltage-dependent potassium (Kv) channels are, among other things, responsible for the propagation of signals through the nervous system. They open and close in response to changes in voltage across the membrane in order to allow the passage of potassium ions out of the cell. In order to better understand the details of Kv channel dynamics on an atomic scale, we have performed an all-atom simulation of the full KvAP tetramer (~250,000 atoms). We have examined intrinsic protein flexibility; the nature of protein- lipid interactions, especially the solvation of several positively charged amino acids known to have an important role in channel gating; and we compare the results of our simulation to several experimental measurements. This research has important implications for understanding the nature of Kv channel gating, and reconciles seemingly contradictory experimental data.
Frank J Wessel Department: Physics and Astronomy
The skin current effect occurs as a result of the transient flow of current on a metallic conductor. Electrons tend to move along the surface of a conductor. Changes in the current's amplitude and direction induces a magnetic field that constrains the electrons to flow near the outer surface of the conductor. The effect increases with frequency and conductor size. At microwave frequencies the core of a conductor would not carry current. Even at AC powerline frequencies heavy copper wire can be replaced with aluminum cables clad with a copper skin without appreciably increasing power losses. This example will compute and compare the skin effect for two circular conductors, modeled using the COMSOL, finite element software.
Harry Mangalam Department: NACS/Research Computing Support
Free and Open Source Software can free you from much of the cost and aggravation of research computing for the rest of your life. Learn the basics and you can accumulate / aggregate / analyze data, compute & model, create and query databases, and write up your results all on Free software. In 10 words, the basics are: Linux, apt-get, GNOME / KDE, bash, Perl / Python, R, gnuplot, SQLite, and OpenOffice. If you desire, you can also run commercial software on Linux such as SAS, Mathematica, Labview, and Windows itself. An excellent starting point is the Software Carpentry Page.
Tony Soeller Department: NACS/Research Computing Support
ArcGIS Server is a web-based Geographic Information System (GIS) which allows users to view spatial data and perform basic GIS queries without having to install GIS software on their local computer. ArcGIS Server extends the audience of a teacher's or researcher's work where spatial data are employed. At UCI, ArcGIS Server is currently used in a teaching application (Dept. of History, North American Indians) and a research application (Dept. of Education, No Child Left Behind). To prepare an application, GIS data are developed using ArcGIS Desktop then the data are uploaded to Server. ArcGIS Server and ArcGIS Desktop software are available through NACS.
Dr. John Porter Collaborator: Prof. Dan Mumm Department: Materials Science and Chemical Engineering, Henry Samueli School of Engineering
Electron microscopy and microanalysis has evolved from a film-based imaging instrument to a computing intensive image processing and image analysis system, based on digital imaging and digital manipulation of diffraction patterns and microanalysis spectra. As microscopists, we are end-users of the vendor provided software, but software evolution can evolve significantly from where it is today. Many of today's capabilities were well understood twenty years ago but it has taken hardware advances to enable today's instrumentation. A digital image 1024 x 1024 pixels requires 1MB of storage. A digital image with 1024 channels of spectral data associated with each pixel requires 1GB of data. An orientation image with a 1024 x 1024 diffraction pattern associated with each pixel becomes even bigger, so smart handling and smart processing of this enormous quantity of data becomes a new task for computer scientists. Serial sectioning, where such images and maps are acquired sequentially as layers of material are removed, simply multiplies the amount of data gathered by the number of layers. Managing and displaying these datasets becomes another task. The human computer interface becomes critical. I have discussed some of this with GIS developers at ESRI. Providing a user friendly means of massaging data can limit the versatility of data handling for the advanced user. MicroGIS appears to me to be a new field - or, more properly, just micro-information system (MIS). But NanoIS may be a more appropriate moniker for funding!
Julia Hoigaard, Ph.D. Department: Social Science Human Subjects Laboratory
I'll present the online research management system, Experimetrix which the School of Social Science has successfully been using for the past three years. The system provides a central cyber location where, 24/7, researchers can post their research needs, students can access the research opportunities and sign up to participate. Credits are assigned online to the students which an administrator or faculty member can access in order to assign extra credit at the end of the quarter. Over 8000 appointments were credited in Spring Quarter using this computerized management system. Thus, a researcher, in a 10-week quarter, can run hundreds of subjects. Social Ecology, Social Science, Education, Management, Medicine are some of the departments using this online research management system. I would encourage other departments who could benefit by using Human subjects to consider participating in this online system. Experimetrix is currently run by a company in Estonia with help available 24/7. They charge approximately $900 per year for the service.