Seminar:
E2 180
Joris Mulder
Assistant Professor
Department of Methodology & Statistics, Tilburg University, Netherlands
Herbert Lee
bf724912872fe7d0118632cf489b60eff8529393
1419449529
0
1425340800
1425344400
Seminar:
E2 180
Igor Prunster
Professor
Department of Statistics, University of Torino, Italy
Abel Rodríguez
09251dd714ca3257a91f48bbcb7f6b99bb471f92
1419449712
0
1425942000
1425945600
Seminar:
E2 180
Artem Sokolov
Postdoctoral Scholar
Biomolecular Engineering, UC Santa Cruz
Juhee Lee
f8df98ce52ee98e9cfe861e6c05dc6e1de112c3d
1419449842
0
1426546800
1426550400
Endogenous Reciprocity
E2-499
Boğaçhan Çelen
Professor
Melbourne Business School
Daniel Friedman
<p>Abstract:<br /> The theory of reciprocity is predicated on the assumption that people are willing to reward kind acts and to punish unkind ones. This assumption raises the question of what kindness is. In this paper, we offer a novel definition of kindness based on a notion of blame. This notion states that for player j to judge whether player i is kind or unkind to him, player j has to put himself in the position of player i, and ask if he would act in a manner that is worse than what he believes player i does. If player j would act in a worse manner than player i, then we say that player j does not blame player i. If, however, player j would be nicer than player i, then we say that player j blames player i. We believe this notion is a natural, intuitive and empirically functional way to explain the motives of people engage in reciprocal behavior. After developing the conceptual framework, we test this concept by using data from two laboratory experiments and find significant support for t!<br /> he theory.<br /> <br /> BIO:<br /> <br /> Boğaçhan Çelen is a Professor of Economics at the University of Melbourne. Prior to joining the University of Melbourne he was a faculty member at Columbia University. He earned his PhD from New York University under the supervision of Professors Douglas Gale and Andrew Schotter. His main research interests include experimental economics and economic theory with special focus on information aggregation in markets and games, and social preferences. His papers appeared in journals such as the American Economic Review, International Economic Review, Management Science, Games and Economic Behavior.</p>
51fff8e8c2d564a3f3bc296497d997154f53f4b4
1421279140
0
1423000800
1423006200
Job Mission as a Substitute for Monetary Incentives: Experimental Evidence
E2-499
Lea Cassar
University of Zurich
Daniel Friedman
<p>Abstract:<br /> Are monetary and non-monetary incentives used as substitutes in motivating effort? I address this question in a laboratory experiment in which the choice of the job characteristics (i.e., the mission) is part of the compensation package that principals can use to influence agents' effort. Principals offer contracts that specify a piece rate and a charity---which can be either the preferred charity of the agent, or the one of the principal. The agents then exert a level of effort that generates a profit to the principal and a donation to the specified charity. My results show that the agents exert more effort than the level that maximizes their own pecuniary payoff in order to benefit the charity, especially their preferred one. The principals take advantage of this intrinsic motivation by offering lower piece rates and by using the choice of the charity as a substitute to motivate effort. However, I also find that because of fairness considerations, the majority of principal!<br /> s are reluctant to lower the piece rate below a fair threshold, making the substitution between monetary and non-monetary incentives imperfect. These findings have implications for the design of incentives in mission-oriented organizations and contribute to our understanding of job satisfaction and wage differentials across organizations and sectors.<br /> <br /> BIO:<br /> <br /> Lea Cassar is a Ph.D. candidate in Economics at the University of Zurich under the supervision of Prof. Roberto Weber. Previously, Lea has received an MPhil degree in Economics from the University of Oxford and a Bachelor’s degree from LUISS University in Rome. Her research is at the intersection between contract theory and behavioral economics. In particular, Lea uses theoretical and experimental methods to study how intrinsic motivation affects the design of incentives and contracts in organizations. Her theoretical work has recently won the Best Paper Award in “Public Organizations” from UniCredit & Universities Foundation, while her experimental work has been awarded the Science of Philanthropy Initiative PhD grant. Throughout her studies, Lea has been actively involved in social work both in Europe and in Africa, gaining, therefore, first-hand experience working with non-profit organizations and charities.<br /> </p>
cf27f6f02b07e5478da458086d2ae0330c93de38
1421279274
0
1423173600
1423179000
Contextual Semantics: From Quantum Mechanics to Logic, Databases and Constraints
Engineering 2 Building, Room 475
Samson Abramsky
Professor, FRS, University of Oxford
Computer Science Department Colloquium
<div style="font-size: 13px;">Abstract:<br />Quantum Mechanics presents a disturbingly different picture of physical reality to the classical world-view. These non-classical features also offer new resources and possibilities for information processing. At the heart of quantum non-classicality are the phenomena of non-locality, contextuality and entanglement. We shall describe recent work in which tools from Computer Science are used to shed new light on these phenomena. This has led to a number of developments, including a novel approach to classifying multipartite entangled states, and a unifying principle for Bell inequalities based on logical consistency conditions. At the same time, there are also striking and unexpected connections with a number of topics in classical computer science, including relational databases, constraint satisfaction, and natural language semantics.<br /> <br />The talk will give a self-contained introduction to these ideas.</div>
<div> </div>
<div>Short Bio:</div>
<div><span> </span></div>
<div><span>Samson Abramsky is </span><a href="http://en.wikipedia.org/wiki/Christopher_Strachey">Christopher Strachey</a><span> </span><span>Professor of Computing at the University of Oxford. </span><span> Previously he held chairs at the</span><span> </span><a href="http://www.ic.ac.uk/">Imperial College of Science, Technology and Medicine</a><span>, and at the</span><span> </span><a href="http://www.ed.ac.uk/">University of Edinburgh</a><span>.</span></div>
<p><span>He is a Fellow of the</span><span> </span><a href="http://www.royalsoc.ac.uk/">Royal Society</a><span> </span><span>(2004), a Fellow of the</span><span> </span><a href="http://www.ma.hw.ac.uk/RSE/rse.htm">Royal Society of Edinburgh</a><span> </span><span>(2000), a Member of</span><span> </span><a href="http://www.acadeuro.org/">Academia Europaea</a><span> </span><span>(1993), and a Fellow of the</span><span> </span><a href="http://awards.acm.org/fellow/all.cfm">ACM</a><span> </span><span>(2014). He is also the winner of the 2007</span><span> </span><a href="http://www2.informatik.hu-berlin.de/lics/#awards">LiCS Test-of-Time award</a><span> </span><span>(a 20-year retrospective) and the recipient of the British Computer Society Lovelace Medal in 2013.</span></p>
<p><span>He has played a leading role in the development of</span><span> </span><span>game semantics</span><span>, and its applications to the</span><span> </span><span>semantics of programming languages</span><span>. Other notable contributions include his work on</span><span> </span><span>domain theory in logical form</span><span>, the</span><span> </span><span>lazy lambda calculus</span><span>,</span><span> </span><span>strictness analysis</span><span>,</span><span> </span><span>concurrency theory</span><span>,</span><span> </span><span>interaction categories</span><span>, and</span><span> </span><span>geometry of interaction</span><span>. More recently, he has been working on</span><span> </span><span>high-level methods</span><span> </span><span>for</span><span> </span><span>quantum computation and information.</span></p>
6f9936cf6474726d882bd95c18151fc825ea71dd
1421364766
0
1422471600
1422475200
Seminar:
E2 180
Daniel Venturi
Assistant Professor (Research)
Division of Applied Mathematics, Brown University
Pascale Garaud
<p><strong>Abstract</strong></p>
<p><strong> </strong></p>
<p>The predictability of mathematical and computational models of physical reality relies on model selection, calibration, validation, and on how well these models forecast specific quantities of interest with quantified uncertainty. Even with recent theoretical and computational advancements, no broadly applicable techniques exist that could deal with the challenging problems of uncertainty propagation and management in high-dimensions, model inadequacy and lack of regularity often exhibited by the quantities of interest. One example of such systems is buoyancy-driven flows subject to high-dimensional random initial or boundary conditions which can undergo several flow transitions leading to discontinuities in the ensemble of flow states, and random frequencies that are difficult to resolve. Another example is models of systems biology driven by colored random noise, such as tumor growth models, which can lead to stochastic resonances and other types of complex dynamics. High-fidelity stochastic simulations of such systems are computationally intractable.</p>
<p> </p>
<p>To overcome these problems, I propose a new paradigm which I will outline in this talk, for variable-fidelity stochastic modeling, simulation and information fusion in high-dimensional systems. The method simultaneously takes into account variable fidelity in models as well as variable-fidelity in probability space (e.g., sparse probabilistic collocation versus multi-element polynomial chaos methods). Information fusion is obtained in a Bayesian setting by using recursive co-kriging. In the new paradigm the interaction between simulations and experiments also becomes more meaningful and more dynamic. Throughout the presentation I will provide numerical examples and applications of the proposed methods to stochastic flow problems, models of systems biology and design/optimization under uncertainty.</p>
<p> </p>
<p> </p>
<p><strong>Bio</strong></p>
<p><strong> </strong></p>
<p>Prof. Venturi received his B.S. and Sc.M. degrees in Mechanical Engineering at the University of Bologna. In 2002 he joined the Department of Energy, Nuclear and Environmental Engineering at the University of Bologna where he received his Ph.D. degree in fluid dynamics and thermal sciences in 2006. He was a post-doctoral fellow in the same department from 2006 to 2010. Since 2010 he has been appointed as assistant professor of applied mathematics (research) at Brown University. His research interests embrace a wide range of topics, including stochastic fluid dynamics, reduced-order modeling, and high-performance scientific computing. In particular, his recent research activities have been focused on the Mori-Zwanzig formulation for coarse-graining high-dimensional systems, multi-fidelity stochastic modeling, design/optimization under uncertainty, stochastic domain decomposition methods, and approximation theory for functional differential equations (e.g., equations involving the Hopf characteristic functional). Prof. Venturi is the leading principal investigator (PI) of an AFOSR (Air Force Office of Scientific Research) project on the Mori-Zwanzig formulation of PDF equations (2014-2017). He is also the co-PI of an ARO (Army Research Office) project on stochastic domain decomposition methods (2014-2017) and a co-investigator of a DARPA (Defense Advanced Research Projects Agency) project related to design/optimization under uncertainty (2014-2015).</p>
7040b90f0435f79828eab68ab0f23e61f17b2e14
1422386768
1422386768
1422900000
1422903600
Seminar:
E2 180
Ruian Ke
Postdoctoral Scholar
Los Alamos National Laboratory
Pascale Garaud
08c4002ca7ff644f940a5f5d45c7cf0fd9cb36f4
1421734051
0
1423504800
1423508400
Seminar:
E2 180
Weiwei Hu
Assistant Professor
University of Southern California
Pascale Garaud
79eff5080df71c0e9f04cee7b06f8a6169d4daff
1421734206
0
1423850400
1423854000
Seminar:
E2 180
Eli Schlizerman
Assistant Professor
Department of Applied Mathematics, University of Washington
Pascale Garaud
7cc8b6e0c5dd2c19e79423a9569a5dfbee3264ee
1421734332
0
1424455200
1424458800
Seminar:
E2 180
Yongcan Cao
National Research Council Research Associate
Air Force Research Laboratory
Pascale Garaud
291cfc3623c4c1efceb5f9526b64ed3c71ea2968
1421734568
0
1424714400
1424718000
Cooperation in the Finitely Repeated Prisoner's Dilemma
Engineering 2 - Room 499
Sevgi Yuksel
PhD Candidate, New York University
Daniel Friedman
<p>Abstract:<br />More than half a century after the first experiment on the finitely repeated prisoner's dilemma, evidence on whether cooperation decreases with experience, as predicted by backward induction, remains inconclusive. This paper provides a meta-analysis of prior experimental research and reports the results of a new experiment to elucidate how cooperation varies with the environment in this canonical game. We describe forces that affect initial play (formation of cooperation) and unravelling (breakdown of cooperation). First, contrary to the backward induction prediction, the parameters of the repeated game have a significant effect on initial cooperation. We identify how these parameters impact the value of cooperation -- as captured by the size of the basin of attraction of Always Defect -- to account for an important part of this effect. This finding is in contradiction to the traditional hypothesis that longer horizon affect plays because it increases the number of steps of backward induction. Second, despite differences in round one behavior, for all parameter combinations, evolution of behavior is consistent with subjects understanding and following the logic of backward induction: Subjects converge to using threshold strategies which conditionally cooperate until a threshold round; conditional on using cooperative threshold strategies, the mode defection round moves earlier with experience.<br /> <br />BIO:<br /><br />Sevgi Yuksel is finishing her Ph.D. at New York University this year. Her advisors are Debraj Ray and Guillaume Frechette. Sevgi's research interests cover two fields: Microeconomic Theory/Political Economy and Experimental/Behavioral Economics. On the theoretical side, she has been studying information acquisition and aggregation among agents with heterogeneous preferences. On the experimental side, her work focuses on understanding when and how cooperation arises in repeated interactions.</p>
64a5850bbef5939d6fbfbf579af7f83455ef19ff
1421795725
0
1423254600
1423260000
Advancement: Real-time Experience Driven Procedural Content Generation
Engineering 2, Room 506
Cameron Alston
PhD Student
Computer Science
Jim Whitehead
<p><strong>Abstract:</strong> Given the increasing development cost and complexity of the games that are being produced, and the diversity in skill, background, and personal preferences of the ever increasing gamer audience, it is intuitive to think that there is a desire for games that are able to entertain and connect with as many players as possible and on as many levels as possible. Recent academic research in the area of player modeling and procedural content generation has opened up a new area for automatic and personalized generation of game content, allowing the design of the game to be flexible and for the players' experience to be unique to each different player. It is the aim of my work to extend the findings of previous research in the area of experience-driven procedural content generation (ED-PCG) and to improve upon the previously explored methods by introducing generated content in real-time during the execution of a game level, and to validate the ideas that game design and player experience can be personalized using ED-PCG methods.</p>
33eecf51b0f6d410bd9ce61878ad5f1d68eb0692
1422040213
0
1422291600
1422298800
A new paradigm for variable-fidelity stochastic modeling and information fusion in physical and biological systems
E2-180
Daniele Venturi
Dr.
Brown University
Pascale Garaud
<p><strong>Abstract</strong></p>
<p><strong> </strong></p>
<p>The predictability of mathematical and computational models of physical reality relies on model selection, calibration, validation, and on how well these models forecast specific quantities of interest with quantified uncertainty. Even with recent theoretical and computational advancements, no broadly applicable techniques exist that could deal with the challenging problems of uncertainty propagation and management in high-dimensions, model inadequacy and lack of regularity often exhibited by the quantities of interest. One example of such systems is buoyancy-driven flows subject to high-dimensional random initial or boundary conditions which can undergo several flow transitions leading to discontinuities in the ensemble of flow states, and random frequencies that are difficult to resolve. Another example is models of systems biology driven by colored random noise, such as tumor growth models, which can lead to stochastic resonances and other types of complex dynamics. High-fidelity stochastic simulations of such systems are computationally intractable.</p>
<p> </p>
<p>To overcome these problems, I propose a new paradigm which I will outline in this talk, for variable-fidelity stochastic modeling, simulation and information fusion in high-dimensional systems. The method simultaneously takes into account variable fidelity in models as well as variable-fidelity in probability space (e.g., sparse probabilistic collocation versus multi-element polynomial chaos methods). Information fusion is obtained in a Bayesian setting by using recursive co-kriging. In the new paradigm the interaction between simulations and experiments also becomes more meaningful and more dynamic. Throughout the presentation I will provide numerical examples and applications of the proposed methods to stochastic flow problems, models of systems biology and design/optimization under uncertainty.</p>
<p> </p>
<p> </p>
<p><strong>Bio</strong></p>
<p><strong> </strong></p>
<p>Prof. Venturi received his B.S. and Sc.M. degrees in Mechanical Engineering at the University of Bologna. In 2002 he joined the Department of Energy, Nuclear and Environmental Engineering at the University of Bologna where he received his Ph.D. degree in fluid dynamics and thermal sciences in 2006. He was a post-doctoral fellow in the same department from 2006 to 2010. Since 2010 he has been appointed as assistant professor of applied mathematics (research) at Brown University. His research interests embrace a wide range of topics, including stochastic fluid dynamics, reduced-order modeling, and high-performance scientific computing. In particular, his recent research activities have been focused on the Mori-Zwanzig formulation for coarse-graining high-dimensional systems, multi-fidelity stochastic modeling, design/optimization under uncertainty, stochastic domain decomposition methods, and approximation theory for functional differential equations (e.g., equations involving the Hopf characteristic functional). Prof. Venturi is the leading principal investigator (PI) of an AFOSR (Air Force Office of Scientific Research) project on the Mori-Zwanzig formulation of PDF equations (2014-2017). He is also the co-PI of an ARO (Army Research Office) project on stochastic domain decomposition methods (2014-2017) and a co-investigator of a DARPA (Defense Advanced Research Projects Agency) project related to design/optimization under uncertainty (2014-2015).</p>
cd0338f709e5527bfee5fbbfd2e7b69d93444ad1
1422385975
1422385975
1422900000
1422903600
A new paradigm for variable-fidelity stochastic modeling and information fusion in physical and biological systems
E2 180
Daniel Venturi
Assistant Professor (Research)
Division of Applied Mathematics, Brown University
Pascale Garaud
<p><strong>Abstract</strong></p>
<p><strong> </strong></p>
<p>The predictability of mathematical and computational models of physical reality relies on model selection, calibration, validation, and on how well these models forecast specific quantities of interest with quantified uncertainty. Even with recent theoretical and computational advancements, no broadly applicable techniques exist that could deal with the challenging problems of uncertainty propagation and management in high-dimensions, model inadequacy and lack of regularity often exhibited by the quantities of interest. One example of such systems is buoyancy-driven flows subject to high-dimensional random initial or boundary conditions which can undergo several flow transitions leading to discontinuities in the ensemble of flow states, and random frequencies that are difficult to resolve. Another example is models of systems biology driven by colored random noise, such as tumor growth models, which can lead to stochastic resonances and other types of complex dynamics. High-fidelity stochastic simulations of such systems are computationally intractable.</p>
<p> </p>
<p>To overcome these problems, I propose a new paradigm which I will outline in this talk, for variable-fidelity stochastic modeling, simulation and information fusion in high-dimensional systems. The method simultaneously takes into account variable fidelity in models as well as variable-fidelity in probability space (e.g., sparse probabilistic collocation versus multi-element polynomial chaos methods). Information fusion is obtained in a Bayesian setting by using recursive co-kriging. In the new paradigm the interaction between simulations and experiments also becomes more meaningful and more dynamic. Throughout the presentation I will provide numerical examples and applications of the proposed methods to stochastic flow problems, models of systems biology and design/optimization under uncertainty.</p>
<p> </p>
<p> </p>
<p><strong>Bio</strong></p>
<p><strong> </strong></p>
<p>Prof. Venturi received his B.S. and Sc.M. degrees in Mechanical Engineering at the University of Bologna. In 2002 he joined the Department of Energy, Nuclear and Environmental Engineering at the University of Bologna where he received his Ph.D. degree in fluid dynamics and thermal sciences in 2006. He was a post-doctoral fellow in the same department from 2006 to 2010. Since 2010 he has been appointed as assistant professor of applied mathematics (research) at Brown University. His research interests embrace a wide range of topics, including stochastic fluid dynamics, reduced-order modeling, and high-performance scientific computing. In particular, his recent research activities have been focused on the Mori-Zwanzig formulation for coarse-graining high-dimensional systems, multi-fidelity stochastic modeling, design/optimization under uncertainty, stochastic domain decomposition methods, and approximation theory for functional differential equations (e.g., equations involving the Hopf characteristic functional). Prof. Venturi is the leading principal investigator (PI) of an AFOSR (Air Force Office of Scientific Research) project on the Mori-Zwanzig formulation of PDF equations (2014-2017). He is also the co-PI of an ARO (Army Research Office) project on stochastic domain decomposition methods (2014-2017) and a co-investigator of a DARPA (Defense Advanced Research Projects Agency) project related to design/optimization under uncertainty (2014-2015).</p>
c3ff895c5de0037a8eba9bd3b1b73e12e3722dd0
1422554958
0
1422900000
1422903600
Defense: Attitude Control of NanoSatellites Using Shifting Masses
Engineering 2, Room 599
Simone Chesi
PhD Student
Applied Mathematics & Statistics
Qi Gong
<div>
<div><strong>Abstract:</strong> Since the development of the CubeSat standard, the number of nanosatellites developed by universities, federal agencies and commercial companies has greatly increased as well as their mission complexity. This thesis focuses on the development of attitude control systems for nanosatellites. The central idea is to utilize the external disturbance forces as control input. Such technique is particular important for small and nanosatellites in low Earth orbit where the major disturbance torque, represented by the aerodynamic torque, limits the ability of spacecraft to maneuver and to achieve stabilization. </div>
<div> </div>
<div>The main contribution of this thesis is the development of a novel attitude control technique that uses active variation of the aerodynamic torque through center of mass shifting. By moving three shifting masses, the distances between the spacecraft's center of mass and the center of pressure of the spacecraft external surfaces can be modified, which results in changes in both magnitude and direction of the total aerodynamic torque acting on the satellite. In this fashion, the shifting masses can convert undesired aerodynamic disturbance into a useful control torque to stabilize spacecraft. In contrast to other actuators, the proposed attitude control system based on shifting masses is more efficient in environment with high disturbance, e.g., low Earth orbit. To achieve three-axis stabilization, a nonlinear adaptive feedback control is developed. The stability of the closed-loop system is analyzed using Lyapunov stability theory and demonstrated through simulations. </div>
<div> </div>
<div>The secondary contribution of this thesis is the application of the center of mass relocation to develop a novel automatic mass balancing system for spacecraft simulators. Spacecraft three-Axis simulators provide frictionless and, ideally, torque-free hardware simulation platforms that are crucial for validating spacecraft attitude, determination, and control strategies. To reduce the gravitational torque, the distance between the simulator center of mass and the center of rotation needs to be minimized. This work proposes an automatic mass balancing system for spacecraft simulators, which utilizes only the three shifting masses during the balancing process, without need of further actuators. The proposed method is based on an adaptive nonlinear feedback control that aims to move, in real-time, the center of mass towards the spacecraft simulator's center of rotation. The stability of the feedback system and the convergence of the estimated unknown parameter (the distance between the center of mass and the center of rotation) are analyzed. The proposed method is experimentally validated using the CubeSat Three-Axis Simulator at the Spacecraft Robotics Laboratories of the Naval Postgraduate School.</div>
</div>
c0d17375e1b95582f24e99e0513fe63f26faceef
1422560852
0
1423247400
1423254600
Seminar: Ultraminiature computational sensing and imaging with phase anti-symmetric diffractive optics
BE 330
Patrick Gill
Principal Research Scientist at Rambus Labs
Joel Kubby
<p>Description:</p>
<p><br />Our aim is to add eyes to any electronic device, no matter how small. To that end, we have developed a tiny and easily manufactured diffraction grating that takes the place of a lens in a traditional camera, allowing us to make miniscule light sensors. Through a novel exploitation of a symmetry property, we are able to ensure that our sensors work with broadband light. I will discuss the theory of operation behind our lensless smart sensors, then present some imaging and motion results from our latest hardware experiments.</p>
<p> </p>
<p>Patrick R. Gill is a Principal Research Scientist at Rambus Labs, in the Computational Sensing and Imaging group. He received his B.Sc. (Hons.) degree in Physics and Mathematics from the University of Toronto, Canada, in 2001 and his Ph.D. in Biophysics studying sensory neuroscience from the University of California at Berkeley in 2007. Prior to working at Rambus, he pursued postdoctoral work at Cornell University in Alyosha Molnar's lab designing and characterizing diffractive light sensors built using native CMOS structures. He has won Canadian national championships in both physics and mathematics, he received the Best Early Career Paper Award from the 2013 Computational Optical Sensing and Imaging Conference, and his technology won a Best of Mobile World Congress award for 2014.</p>
67cd9a8c92d7f631901902adf9c7258821f6654b
1422575729
0
1426798800
1426802400