Previous seminars
Previous seminars
2015 seminars
Title: Social Media & Social Network Data Analytics
Date & time: 11am, 4 May 2015
Short description: Social media and networks are a popular place for people to express their opinions about consumer products, to organide or initiate social events, or to spread news. Some questions should be asked in order to understand the social media and social networks: How can we detect and predict the emerging events with the awareness of locations? How can we predict the propagation patterns of online micro-blogs? How can we understand people’s opinions about a current issue, a new product, or an important event such as government elections? This talk is to report our recent research work on the social media and social networks data mining. A few application systems will be reported to answer the above questions.
Presenter: Dr Xue Li
Location: Room R505A, level 5, Axon Building, Building 47, St Lucia Campus, UQ, QLD 4067
Contact: Prof David Abramson, RCC, david.abramson@uq.edu.au
Title: Addressing Performance and Programmability Challenges in Current and Future Supercomputers
Short description: With future high-end systems becoming more parallel with more processors per node, more threads per processor, longer vector lengths, and more complex memory hierarchies, application developers need an environment that can maximise programmability, and ease porting and tuning efforts to close the gap between observed performance and achievable performance, while not losing sight of performance portability across a wide range of systems. Developers need sophisticated tools and adaptive runtime systems that can help solve multi-disciplinary and multi-scale problems, enabling them to achieve high levels of performance. In this talk I will present Cray's programming environment vision and will discuss some of the challenges and open research problems that need to be addressed to build the system software for extreme-scale systems.
Presenter: Dr Luiz DeRose (Cray Inc, USA)
Date & time: 11am, 19 March 2015
Location: B49 R502 Advanced Engineering Building, UQ, St Lucia Campus, UQ, QLD 4067
Contact: Prof David Abramson, RCC, david.abramson@uq.edu.au
Title: Programming Future Large Scale Systems
Date & time: 11am, 18 March 2015
Short Description: The US Department of Energy recently announced that Lawrence Livermore National Laboratory (LLNL) and Oak Ridge National Laboratory (ORNL) would both procure large systems from IBM as a result of the CORAL RFP. These systems, which represent significant steps on the path to exascale, will include NVIDIA Volta GPUs. Thus, a major consideration is how the significant performance potential will be accessed. This talk will provide an overview of the CORAL RFP and a brief discussion of the planned systems. It will then detail how LLNL applications will evolve to use this and other future large-scale systems.
Presenter: Dr Bronis R. de Supinski (Lawrence Livermore National Laboratory)
Location: B49 R502 Advanced Engineering Building, UQ, St Lucia Campus, UQ, QLD 4067
Contact: Prof David Abramson, RCC, david.abramson@uq.edu.au
2014 seminars
All 2014 seminars were held in Room 505, Level 5, Axon Building (47), St Lucia Campus, UQ.
Title: Managing the execution environment for scientific workflows
Date & time: 9-10am, 1 August 2014
Short description: This talk will examine issues related to the execution of scientific workflows in distributed environments. It will describe the characteristics of some applications in astronomy, earthquake science, gravitational-wave physics, and others and their execution needs. It will show the different wide area execution environments, which need to support the workflow execution, with an emphasis on data storage options. The talk will then describe how the applications are mapped to these environments and executed by the Pegasus Workflow Management System and how the environment can be controlled to improve the execution of the workflow applications. Controlling the environment through resource provisioning will focus particularly on the provisioning of wide area networks that enable efficient data transfer to the application and within the workflow.
Presenter: Prof Ewa Deelman, Research Associate Professor & Assistant Director, University of Southern California Information Sciences Institute
Title: Enabling chemical discovery through the lens of a computational microscope
Date & time: 9-10am, 8 August 2014
Short description: With exascale computing power on the horizon, computational studies have the opportunity to make unprecedented contributions to drug discovery efforts. Steady increases in computational power, coupled with improvements in the underlying algorithms and available structural experimental data, are enabling new paradigms for discovery, wherein computationally predicted ensembles from large-scale biophysical simulations are being used in rational drug design efforts. Such investigations are driving discovery efforts in collaboration with leading experimentalists. I will describe our work in this area that has provided key insights into the systematic incorporation of structural information resulting from state-of-the-art biophysical simulations into protocols for inhibitor and drug discovery, with emphasis on the discovery of novel druggable pockets that may not be apparent in crystal structures.
Presenter: Dr Rommie Amaro, Director, National Biomedical Computation Resource, UCSD Department of Chemistry and Biochemistry
Title: Beyond the Data Cliff: Visualisation Strategies for Understanding Large Data
Date & time: 9-10am, 15 August 2014
Description: As scientific instrumentation is advanced, total data volumes increase in size, acquisition rate and complexity. Spectacular examples include the Large Hadron Collider projects, the Allen Human Brain Atlas, the Large Synoptic Survey Telescope projects and closer to home, the Australian Square Kilometre Array Pathfinder telescope surveys. Contemporary informatics approaches are solving the challenges of storing, finding and retrieving large data extracts and collections, and instantiating sophisticated post-processing workflows.
But what of traditional analysis and visualisation, and understanding and comprehending the data at hand? Drawing examples from neuroscience, astrophysics, geoscience and anatomy, I will describe the data that is collected and how it is analysed, and especially visualised, to enable discovery and insight. I will place this work in the context of the impending data cliff that many science and engineering disciplines face, and motivate the need for advanced, immersive visualisation facilities that are closely-coupled to the large data sources and computing systems.
With the scene set, I will introduce Monash University's newest advanced visualisation platform, the CAVE2 (TM), built in collaboration with our partners at the Electronic Visualization Laboratory at the University of Illinois at Chicago. CAVE2 is a large-scale, immersive, head-tracked virtual reality environment backed by a high-performance supercomputer. It boasts 84 million pixels at ultra-high clarity and contrast, and exploits Monash's research-grade network fabric to connect to local and remote data sources and supercomputing centres.
Monash foresees the CAVE2 becoming a critical piece of research infrastructure, fulfilling the role of the world's finest viewfinder in the 21st century microscope, but also serving a unique role as a collaborative display environment for comprehending large data, accelerating and transforming the research process to enable the most important discoveries now and in the future. I will describe and highlight our early successes in the CAVE2 and finally, briefly speculate on the future of immersive visualisation.
Presenter: Dr David Barnes, Senior Research Fellow, CAVE2 Platform Manager, Monash University
Title: Visual Analysis of Personal Health Data
Date & time: 9-10am, 22 August 2014
Short description: Modern analysis methods allow people to collect unprecedented amounts of information about the status of their health. While daily weight and body fat measurements can be collected by almost anyone, fewer people have their blood analyased for biomarkers on a regular basis, or even their DNA and their gut microbiome sequenced. At Calit2, we work with one of the world's most quantified individuals, founding director Larry Smarr, who made available to us much of the medical data he collected about himself so that we could create visual and computational analysis tools. This presentation will report on our quest of distilling information such as patterns and relations out of Dr Smarr's data, in comparison with available data from dozens of other people. We created a visual analysis tool for the 64 million pixel tiled display wall at the Qualcomm Institute, a system driven by 17 high-end graphics PCs and networked with 10 Gbps.
Presenter: Dr Jurgen Schulze, Adjunct Professor, Department of Computer Science, UC San Diego Qualcomm Institute
Title: Exploring Clouds as Enablers of Science
Date & time: 9-10am, 29 August 2014
Short description: Cloud computing has emerged as a dominant paradigm that has been widely adopted by enterprises. Clouds provide on-demand access to computing utilities, an abstraction of unlimited computing resources, and support for on-demand scale up, scale down and scale out. Clouds are also rapidly joining high-performance computing system, clusters and Grids as viable platforms for scientific exploration and discovery. As a result, understanding application formulations and usage modes that are meaningful in such a hybrid infrastructure, and how application workflows can effectively utilise it, is critical. In this talk, I will explore the role of clouds in science and engineering. I will also explore how science and engineering applications can benefit from clouds and how the cloud abstraction can lead to new paradigms and practices. This talk is based on research that is part of the CometCloud autonomic cloud-computing project at the NSF Cloud and Autonomic Computing Center at Rutgers.
Presenter: Professor Manish Parashar, Electrical and Computer Engineering, Rutgers University
Location: Room 505, Level 5, Axon Building (47) UQ, St Lucia Campus
Title: Elevating nanoHUB to the Next Level – Vision and Status 2014
Date & time: 9-10am, 5 September 2014
Short description: More than 330,000 users in 172 countries annually participate in nanoHUB.org, a science and engineering gateway providing the capability to perform online simulation resources through a web browser without the installation of any software. nanoHUB is an online meeting place for simulation, research, collaboration, teaching, learning and publishing. More than 12,000 users run simulation software from their browser in nanoHUB’s science computing cloud. Cumulatively more than 19,000 students in more than 1,000 classes utilised nanoHUB simulations in classrooms and more than 2,200 authors referenced nanoHUB in more than 1,100 scientific publications. The platform has spawned nanoHUB-U and, in turn, Purdue HUB-U, interfaces for online courses that are broadly accessible around the world.
Presenter: Dr Gerhard Klimeck, Professor, Electrical and Computer Engineering and, Director of nanoHUB, the Network for Computational Nanotechnology, Purdue University
Title: Modelling cancer as a transcriptional landscape
Date & time: 9-10am, 12 September 2014
Short description: More than half a century ago, the British developmental biologist and philosopher Conrad Hal Waddington introduced the landscape as a metaphor of how cells differentiate into different types of tissues. This is now recognised as “probably the most famous and most powerful metaphor in developmental biology”. Nonetheless it has remained unclear how such a surface might be computed from actual cell-state data, and if so whether it could be informative or predictive about real-life biology. Here I explore Hopfield neural networks with genome-wide gene expression data as a computational model of Waddington’s landscape. I discuss concepts of state, trajectory and attractors, and present visualisations of the Hopfield surface for subtypes of breast cancer.
Presenter: Dr Mark Ragan, Professor, Genomics of Disease and Development, Institute of Molecular Biology, University of Queensland
Title: The Tell-Phone — Where Mobility Meets Intelligence
Date & time: 9-10am, 26 September 2014
Short description: Mobile Analytics today focuses on a range of interesting problems including the inference of location semantics, user profiling based on mobile phone usage, identification of mobile social networks, predicting the next place of a user/device and recommending places of interest (PoI). The richness and diversity of the data that is being generated today via mobile devices, their connection points (e.g. CDR) and applications (e.g. location-based social networks/social-media) raises several research challenges which are of interest to both the data mining community as well as the mobile computing world. This talk will focus on specific research areas where mobility intersects with intelligence. We have been working on analysing both benchmark and real data from wearable sensors, mobile devices, cell-towers, and location-based social networks. This talk will present our recent research and applications/demonstrations in mobile analytics.
Presenter: Dr Shonali Krishnaswamy, I2R,Singapore
Title: Scaled-up Physics-based Simulations for Earthquake System Science
Date & time: 9-10am, 17 October 2014
Short description: Dynamic simulations of fault ruptures and seismic wave propagation cannot yet achieve the physical scale range needed to explore important domains of earthquake behavior. This talk presents the recent technical effects in scaling up a community finite difference code on advanced heterogeneous supercomputers. We illustrate the acceleration of the critical strain tensor calculations, coupled with co-scheduling capabilities of workflow-managed systems, and how to make a statewide hazard model a goal reachable with existing supercomputers
Presenter: Dr Yifeng Cui, Director HUBzero® Platform for Scientific Collaboration, Purdue University
Title: Multi-Scale Modeling of the Failing Heart: From Mouse to Human
Date & time: 9-10am, 24 October 2014
Short description: Heart failure is a complex syndrome that involves cellular defects in excitation-contraction coupling and contractile mechanisms, neurohormonal and dysregulation and metabolic changes, ventricular structural remodeling, and circulatory alterations. We use multi-scale computational modeling, magnetic resonance imaging and biophysical studies in genetically engineered mice to investigate how cellular and molecular defects can give rise to heart failure at the organ and system scales. Here I will show recent studies on the mechanisms by which gene defects in the contractile regulatory protein myosin regulatory light chain and the cytoskeletal protein vinculin can lead to dilated cardiomyopathy and heart failure. Recent progress in using patient-specific models of dyssynchronous heart failure to understand clinical responses to cardiac resynchronization therapy will also be presented.
Presenter: Professor Andrew McCulloch, University of California, San Diego