[SAMSI logo] 19 T.W. Alexander Drive
P.O. Box 14006
Research Triangle Park, NC 27709-4006
Tel: 919.685.9350
Fax: 919.685.9360
[email protected]
 
OpportunitiesProgramsWorkshopsCalendarAffiliatesReports & PublicationsDirectoryAbout
 

Description of SAMSI

The Statistical and Applied Mathematical Sciences Institute (www.samsi.info), which was established in 2002, is forging a synthesis of the statistical sciences and the applied mathematical sciences with disciplinary science to confront the very hardest and most important data- and model-driven scientific challenges. These challenges are too large and complex to be adequately addressed by individuals, groups or single disciplines. SAMSI’s scope and engagement are national and international, bringing together researchers who would not otherwise interact, and focusing the people, intellectual power and resources necessary for simultaneous advances in the statistical sciences and applied mathematical sciences that lead to ultimate resolution of the scientific challenges.

SAMSI maximizes its impact by being primarily a formulator and stimulator of research. It conducts programs that target the areas most in need of attention and most amenable to high-impact progress, and brings together established and young researchers from academia, industry, national laboratories and government to define the central problems and catalyze the research that addresses those problems.

The human impact of SAMSI is both a goal in itself and the mechanism by which research impact is realized. Visiting young and senior researchers are resident at SAMSI for periods of a month to a year. Postdoctoral fellows participate at the crucial, formative stage of their careers. Graduate and upper level undergraduate students are provided unique insight into the formation of research areas and collaborations. Every SAMSI program conducts public workshops that inform and energize the national statistical and applied mathematical sciences communities to take up the challenges identified by SAMSI. Selective outreach programs to undergraduate students, high-school teachers and faculty from teaching institutions extend SAMSI’s impact still further.

The scale of SAMSI, which matches the scale and urgency of the problems that drive it, exceeds the capabilities of any single institution. Therefore, SAMSI is a partnership between the U.S.A. National Science Foundation and the consortium of Duke University, North Carolina State University (NCSU), the University of North Carolina at Chapel Hill (UNC), and the National Institute of Statistical Sciences (NISS). Support is also provided by William Kenan, Jr. Institute for Engineering, Technology and Science. SAMSI is housed at the NISS building in the Research Triangle Park, North Carolina.

The scientific efforts at SAMSI are organized into regular programs of six months to one year duration, and shorter summer programs, as well as an extensive education and outreach program. Some research programs focus on particular scientific problem areas, while others are defined by statistical and mathematical themes that cut across multiple scientific contexts. Each is led by national and international leaders in the statistical and applied mathematical sciences, coupled with strong involvement of disciplinary scientists. The programming ideas arise from many sources, foremost of which is simply an individual or group proposing – and wishing to help lead – an exciting new program.

SAMSI is led by a Directorate consisting of James Berger (Duke University), Pierre Gremaud (North Carolina State University), Nell Sedransk (National Institute of Statistical Sciences), and Michael Minion (University of North Carolina at Chapel Hill). Scientific input to SAMSI comes from a National Advisory Council, chaired by Carlos Castillo-Chavez (Arizona State University) and Bin Yu (University of California at Berkeley), and a Local Development Committee consisting of leading Research Triangle scholars. The Governing Board, consisting of a leader from each of SAMSI’s partners, together with representatives from the American Statistical Association and Society for Industrial and Applied Mathematics, oversees the operation and development of SAMSI.

Overarching Research Themes

SAMSI programming can partly be understood in terms of an evolving set of overarching mathematical and statistical themes. The themes also indicate commonalities that are exploited in the scheduling of SAMSI programs.

Numerical Modeling of Processes: Physical modeling employing intensive numerical computation is routine in science and engineering; numerical experimentation often supplants physical experimentation. Many SAMSI programs center on development and use of numerical modeling in cross-disciplinary problems involving nonlinear processes, focusing on major gaps in our understanding of the interplay between statistical analysis and mathematical modeling.

Stochastic Computation: There has been a profound increase in the use of stochastic methods to perform computations such as numerical integration (e.g. Markov chain Monte Carlo) and optimization (e.g., annealing and tempering). Basic theoretical questions in probability theory and computational mathematics often arise, such as establishing convergence properties of the computational algorithms. By bringing together statistics, applied mathematics and computational science, SAMSI maintains the current momentum.

Multiscale Modeling: Issues of scale are basic to modern numerical modeling, with adaptive scaling holding great promise for providing computationally feasible, yet realistic, models. When numerical modeling is combined with data – a central feature of many SAMSI programs – multiscale issues come to the fore: data are typically observed at a different scale from that underlying the numerical model. Proper utilization of the data, including characterization of uncertainty, requires models that are consistent across multiple scales. The creation and analysis of such models a major focus of many SAMSI programs.

Challenging Data Sets: Many SAMSI programs involve data sets that are challenging because of their scale – either number of records or number of attributes, their poor quality, or external considerations such as data confidentiality. Other challenges include streaming data and new forms of data such as images, video, audio and text data. The changing nature of data is a grand challenge of science, and SAMSI is at the forefront of addressing it.

Data Assimilation: Issues of assimilating data into models arise when model improvement is needed in the light of available data. This is a critical issue of our times as ever-improving technology has afforded both better observational measurements and much greater computing power in predictive models. SAMSI is well situated to address these issues as they lie at the nexus of physically based modeling and data-driven analysis and the issue has emerged as a focus of many SAMSI programs.

Space-Time Modeling: Space-time modeling is central to a wide variety of disciplinary problems, ranging from meteorology to disease mapping. The statistical issues here center around the fact that spatial modeling and time series modeling are historically very different activities with very different methodologies. Much remains to be done to develop the proper fusion of the two. Advances in applied mathematics relating to partial differential equations and dynamical systems may provide a path to this fusion.

Types of Research Programs

Principal Research Programs are broadly-based programs focusing on interfaces among statistics, applied mathematics, and other sciences. These programs are typically 6–12 months in duration; involve long and short term visitors from across the nation; contain both large and small, intensive research workshops, broad educational workshops and tutorials; involve postdoctoral fellows; include graduate students; seek industrial, governmental agency and national laboratory involvement; and have extensive outreach and educational activities. The programs communicate their results through closing workshops, final program reports, technical reports and dissemination on the Web.

Central to the research in these programs are the working groups, each of which meets at least weekly. These working groups form the heart of any SAMSI program and are the main forum in which collaborations are created and through which the research agendas and individuals are impacted by being at SAMSI. The working groups in each program are officially formed at the opening workshop and the workshop at the end of the program is planned as a summation of the working group activities. Smaller workshops are organized midstream to focus investigation of particularly critical topics that have emerged from the working groups.

Working groups vary in their nature and format. Some are highly focused, at least initially, on knowledge acquisition and sharing. Others focus on specific problem formulation and agenda-setting, while still others devolve quickly into a set of research projects. The exact nature of a working group depends on the interests of the participants, their goals and the extent of shared backgrounds, among many others. Common is always the aim of developing a research agenda and attendant collaborations that derive their energy from combining statistical and (applied) mathematical ideas in a cross-disciplinary context.

A typical principal research program involves 10-15 long-term visiting researchers, numerous short-term visitors, many highly-involved Triangle researchers, 3 postdoctoral fellows, 5 graduate students, and 3-5 workshops with hundreds of participants. Many non-resident researchers participate remotely in the SAMSI working groups through SAMSI’s web-based interactive system.

A unique characteristic of research programs at SAMSI is that they are structured and managed in such a way that, while the primary thrust of SAMSI is to catalyze research (performed by SAMSI participants after they depart), SAMSI pushes the research far enough to ensure that the catalysis "takes". For instance, a SAMSI workshop, while serving the same educational and informational functions as other workshops, also has as an explicit goal to identify research directions to be pursued. Similarly, individual residents at SAMSI do not simply interact by discussing topics of common interest, but are stimulated to begin new collaborations. They ideally will take away from SAMSI not the intention to collaborate, but instead an incipient research project.

Intensive Summer Programs are shorter programs of two weeks to one month, held during the summer and focusing on specific core issues in statistics or applied mathematics. A summer program consists of tutorials, an opening workshop, working groups meeting almost daily, and a closing event.

SAMSI Research Programs (through 2010)

Here is the list of the SAMSI research programs held from the inception of SAMSI through 2010. At the SAMSI website (www.samsi.info) one can find the final report from each of the (finished) programs, as well as program websites that lead to tutorials and workshop talks that were presented during the program, and working group websites where the activities and results of the working groups are summarized.

2002-03
Inverse Problem Methodology in Complex Stochastic Models
Large Scale Computer Models for Environmental Systems
Challenges in Stochastic Computation
2003-04
Network Modeling for the Internet
Multiscale Model Development and Control Design
Data Mining and Machine Learning
2004-05
Latent Variable Models in the Social Sciences
Data Assimilation in Geophysical Systems
Computational Modeling of Infectious Disease
2005-06
National Security and Homeland Defense
Financial Mathematics, Statistics and Econometrics
Astrostatistics
Multiplicity and Reproducibility in Scientific Studies (Summer)
2006-07
Development, Assessment and Utilization of Complex Computer Models
High Dimensional Inference and Random Matrices
Dynamic Treatment Regimes and Multistage Decision-Making (Summer)
Geometry and Statistics of Shape Spaces (Summer)
2007-08
Risk Analysis, Extreme Events and Decision Theory
Random Media
Environmental Sensor Networks
Synthesis and Appraisal of Multiple Sources of Evidence (Summer)
2008-09
Sequential Monte Carlo Methods for Scientific Computing
Algebraic Methods in Systems Biology and Statistics
2009-10
Space-time Analysis for Environmental Mapping, Epidemiology and Climate Change
Stochastic Dynamics

Education and Outreach Program

The Education and Outreach Program at SAMSI is directed toward teachers of primary and secondary students, undergraduate and graduate students and is national in scope.

Industrial Mathematical and Statistical Modeling (IMSM) Workshop for Graduate Students: During each workshop, 35-45 students selected from a national pool work in teams of 5-7 individuals on projects presented by nonacademic scientists. The overall goals of the workshop are to expose mathematics and statistics students to current research problems from government labs and industry as well as to a team approach to problem solving.

SAMSI Workshop for Undergraduates: SAMSI hosts a one-week Undergraduate Workshop each May to introduce students to the manner in which statistics and applied mathematics play fundamental roles in science. The workshop provides students with an intensive introduction to the synergy between applied mathematics and statistics within the context of timely physical applications.

Undergraduate Outreach Days at SAMSI: Twice each year, SAMSI hosts two-day outreach workshops that engage undergraduates from programs around the country. The workshops are tied directly to concurrent SAMSI programs, and therefore expose students to cutting edge research issues. Each workshop consists of overview lectures as well as hands-on activities.

SAMSI/Kenan Fellows Program: The research themes of SAMSI are translated to K-12 schools through a collaboration of SAMSI with the Kenan Fellows Program. We provide enrichment activities for a select group of SAMSI/Kenan Fellows, by sponsoring two SAMSI/Kenan Fellows (during each class) to develop curricula which incorporate mathematical and statistical concepts associated with SAMSI programs.

Opportunities for Visiting Researchers

Principal Researchers are engaged at the initiation stage of a program and provide leadership throughout its course. Ideally, they are resident for much of the program. Sabbatical leaves and teaching buyouts are prominent sources of leveraged funding to allow principal researchers to visit SAMSI.

New Researcher Fellows are outstanding new researchers (typically within 7 years of their Ph.D.) who are provided support to be resident at SAMSI for a semester or year, to participate in the year’s research programs.

Postdoctoral Fellows are extensively involved with all programs. SAMSI funds up to seven full postdoctoral positions per year, and has a number of other postdoctoral associates whose primary appointment is at one of the partner universities. SAMSI provides two layers of mentoring for postdoctoral fellows. The first layer is scientific, with each postdoctoral fellow being assigned to at least one senior Mentor, a researcher on the relevant SAMSI program. The second layer of mentoring is provided by an administrative advisor, who is responsible for general career advice and personal issues.

Graduate Students usually participate in programs together with an advisor who is also a participant. The partner universities support 11 such Graduate Fellows each year for participating in SAMSI programs. Visiting Principal Researchers have the opportunity to also bring their graduate students as SAMSI Graduate Fellows.

Workshop Participants range from key scientific figures who can participate in the program only on a reduced scale, to young researchers who desire to gain entrance to the research area, to individuals primarily interested in technology transfer.

Organizational Collaborations and Diversity

In addition to the relationships with the partner organizations, SAMSI works closely with a number of other organizations. SAMSI and NISS share the Affiliates Program (www.samsi.info/affiliates/index.shtml), whose members include 17 corporate affiliates, 9 government agencies and national laboratories, and 33 university departments. The heavy involvement of government agencies in the affiliates program is quite distinctive, and gives SAMSI connections and visibility in a much-needed area. The affiliates are a major point of entry of SAMSI to the statistics and applied mathematics communities and, on behalf of these communities, have major input into SAMSI programming.

SAMSI also coordinates activities with the other NSF mathematical sciences institutes (see the math sciences institute webpage www.mathinstitutes.org) and with many other national and international research organizations. One of the major activities of the institutes is to encourage diversity in the statistical and applied mathematical sciences. The webpage (www.samsi.info/about/diversity.shtml) indicates SAMSI activities in this regard.




 
 

Entire site © 2001-2010, Statistical and Applied Mathematical Sciences Institute. All Rights Reserved.