Department Seminar 2012
Spring and Fall 2012
Title: Reproducible research in molecular biophysics and structural bioinformatics
Speaker: Konrad Hinsen, CNRS Orléans (Centre de Biophysique Moleculaire) and Synchrotron Soleil, Gif-sur-Yvette
Host: Nathalie Reuter/CBU
Abstract: Paper retractions and research scandals related to computational procedures have created an increasing awareness for the need to make computational results in science reproducible: all the input data and all the software used in a research project should be published along with the results, permitting a step-by-step verification of the computations. Reproducibility creates specific challenges for scientists working on biological macromolecules, which are related to the complexity and size of the data sets. I will present these challenges and propose possible solutions.
Title: There is no paradox of logical validity
Speaker: Roy T. Cook, University of Minnesota
Abstract: JC Beall & Julien Murzi, Lionel Shapiro and Bruno Whittle have all recently argued that one can construct a `paradox of logical validity' by supplementing Peano Arithmetic (or any theory strong enough for Godel coding and diagonalization) with a binary validity predicate that holds of (the codes of) two formulas F and G if and only if the argument with F as premise and G as conclusion is logically valid. In this paper I demonstrate that, properly understood, there is no such paradox of logical validity, since the derivations of a contradiction in all three presentations depends on the illegitimate assumption that the rules for the validity predicate are themselves logically valid. A consistency proof for the correct rules for the validity predicate is also provided. The paper concludes by examining whether the argument(s) can be reconstructed to show that other notions of validity (e.g. "analytic validity", "follows as a matter of metaphysical necessity", etc.) are, indeed, paradoxical.
Title: Inside Information - How Visualizing the Human Body is Sparking a Revolution in Science Exhibits for Public Venues
Speaker: Prof. Anders Ynnerman, Linköping University
Abstract: Medical imaging techniques have advanced beyond recognition in the
last few years. Prof Anders Ynnerman will explain how these systems,
normally the preserve of doctors and surgeons, are now being adapted
to provide public visitor venues, such as museums, science centers
and zoos with unique interactive learning experiences. By combining
new medical visualization techniques with interactive multi touch table
technology and intuitive user interfaces, this is opening up new ways
for visitors to interactively explore the normally invisible and learn
about the inside workings of the human body, exotic animals, natural
history subjects or even mummies. Prof Ynnerman will illustrate this
with examples of systems now being incorporated into public venues
About the speaker:
Professor Anders Ynnerman received a Ph.D. in physics from Gothenburg University. During the early 90s he was at Oxford University, UK, and Vanderbilt University, USA. From 1997 to 2002 he directed the Swedish National Supercomputer Centre and from 2002 to 2006 he directed the Swedish National Infrastructure for Computing (SNIC). Since 1999 he has held a chair in scientific visualization at Linköping University and is the director of the Norrköping Visualization Center-C.
Ynnerman is a member of the Swedish Royal Academy of Engineering Sciences and a board member of the Swedish Research Council. In 2007 Ynnerman was awarded the Akzo Nobel Science award and the Golden Mouse award for Swedish IT-person of the year. In 2009 he received the Athena Award for best medical clinical research in Sweden and in 2010 he received the Swedish Knowledge Award for dissemination of scientific knowledge to the public. In 2011 he received the IVA gold medal from the King of Sweden.
Title: Link Activation and Scheduling in Wireless Networks: An Optimization Perspective
Speaker: Prof. Di Yuan, Department of Science and Technology, Linköping University, Sweden
Abstract: In this talk, we revisit the maximum link admission problem with a generic model of wireless network that consists in a set of radio links being coupled by a channel gain matrix. Given the signal-to-interference-and-noise ratio (SINR) requirement, the problem amounts to determining how many of the links can be activated simultaneously at maximum. Link activation is a key element in scheduling and cross-layer resource allocation. We present mathematical formulations of the problem in the context of optimal scheduling, highlighting the impact of formulation on computational efficiency. Next, new results and perspectives are discussed. We present a reformulation of the link activation problem, enabling the global optimum to be reached in orders of magnitude faster, and give an outlook of extensions to interference cancellation and cooperative transmission.
Dr. Kenneth W. Regan, University at Buffalo (SUNY)
Title: Gödel's Lost Letter and P=NP
Abstract: The weblog Gödel's Lost Letter and P=NP was started by Richard Lipton of Georgia Tech in February 2009. I am lucky to have known him and worked with him from the mid-1990's, and after being drawn in to several research topics the first spring and serving as copy-editor for a year, I am now full partner. Among several missions of the blog are promoting the freedom of ideas, excitement over research, personal factors that engender inspiration, and real-time interaction across continents. I will present examples from the blog, including current debate and discussions on quantum computation, solving linear equations, the impact of online courses, and long-standing open problems in mathematics and complexity theory. These include broadcasting some of our joint research and my own private work, focusing on my recent post "Grilling Quantum Circuits". Whether such a post makes a 28-pages-and-growing draft paper accessible for non-experts in the quantum sub-field is part of the story to discuss.
Dines Bjørner (DTU, DK)
Title: Domain Science & Engineering: A New Facet of Informatics
Abstract: The goal of domain engineering is to construct a Domain Description. A domain description informally and formally describes (usually) a man-made domain, e.g., air traffic, banking and health care systems. A domain description may be studied (and improved upon), theorems derived etc., analogous to models in e.g. physics. Domain engineering is as such an applied scientific endeavour. Domain science is the theoretical foundation for creating models: "what can be described", "mereology", etc. The talk will survey this and will briefly indicate that domain models can serve as a basis for requirements engineering, the conventional starting point for software development. The talk will "flash" facets of models of example domains and will hint at philosophical issues such as mereology and domain calculi.
About the speaker: Dines Bjørner has been working at IBM (1960s & 1970s), DTU (1976 onwards), was founding director of UNU-IIST (Macau CN, 1992-1997), and has had numerous guest lecture and research stays all over the world. He has been central in the development of VDM and Raise, and has worked on PL/1 and Ada semantics and tools. The last couple of decades his interest has switched to domain science and engineering.
Stefan Bruckner (Vienna Univ. of Technology)
Title: Visual Analysis and Exploration of Volumetric Data – Challenges and Perspectives
Abstract: Information technology has lead to a rapid increase in the amount of data that arise in areas such as biology, medicine, climate science, and engineering. In many cases, these data – like the underlying phenomena – are volumetric in nature, i.e., they describe the distribution of one or several quantities over a region in space. Volume visualization is the field of research which investigates the transformation of such data sets into images for purposes such as understanding structure or identifying features. However, it is no longer sufficient to only provide means for analyzing individual data sets in isolation. Instead, many thousands of data points, each consisting of a volumetric representation, need to be investigated. Such volumetric data spaces not only require new approaches to data management and transfer, but also necessitate intuitive navigation, interaction, and visualization techniques. In particular, it is crucial to provide efficient facilities to visually explore, query, and retrieve data items, as well as methods to categorize and abstract the space. In this talk, I will outline the challenges involved in building such tools, present approaches on how to overcome some of the obstacles, and provide a perspective on how future research in this field may profoundly impact scientific knowledge discovery.
Thore Husfeldt (Lund University and IT University of Copenhagen)
Title: Science under the algorithmic lens
Abstract: Computer Science is on its way to become the foundation of all of science. Software is the language of science, almost all data is digital, and computer applications permeate all fields. But that's not what this talk is about. Instead, I will give an introduction to the concept of the algorithmic lens. This concept includes the idea of computational thinking: the perspective that reality can be understood as an algorithmic phenomenon, a continuation of the mechanistic worldview from Leibniz via Turing. The presentation is non‐technical and structured along a number of examples. It is admittedly skewed by my background as an algorithms person, but is also meant as a cheeky attempt to present Computer Science as a source of an intellectual mindset, rather than merely a technological tool.
Igor Semaev (The Selmer Center, UoB)
Title: From Jacob Bernoulli to Modern Cryptography.
Abstract: Let n shots be allocated into N boxes according to a probability distribution on boxes. Equivalently, n balls are randomly drawn from an urn with balls of N different colors, where the fraction of balls of each color define a distribution on colors. This convenient model to study probability theory was mentioned by Jacob Bernoulli in Ars Conjectandi published in 1713.
The number of boxes with exactly r shots (in particular, the number of empty boxes), the number of shots before all N boxes are occupied(coupon collector's problem) etc. are among the most attractive problems to study. The model has applications in statistics, statistical mechanics, genetics.
Cryptography aims to provide information security with ciphers, hash functions, digital signatures etc.
In this talk, I will explain why random allocations are there useful. Any special knowledge of the above scientific disciplines is not necessary for understanding the presentation.
About the speaker: Igor Semaev is a professor with the Selmer Center at the Department of Informatics, UoB.