Welcome and Promotions

Letter from J. Allen Cox

December 28, 2004

Over the years Honeywell scientists and engineers have maintained various forms of relationships with universities in general and the University of Minnesota in particular. In many cases these relationships were spawned at the individual level between a professor and a Honeywell employee familiar with the professor’s work, and frequently they took the form either as consultative agreements or as collaboration on specific government contracts. However, in 1982 the establishment of the IMA offered a completely new approach to promote university-industry collaboration by formally creating an organization for that very purpose and driving the process by requiring industry’s “buy-in” from the start. The buy-in by industry came in two forms: hard cash for membership and student support and specific problems in applied mathematics. The IMA responded by delivering outstanding performance at a great value to the industrial partners; the best evidence of this is found both in the continued strong support for the IMA after 24 years and in the number of similar centers that have been created around the country based on essentially the same model.

My own personal involvement with the IMA started in 1981 as the point-of-contact in the Honeywell Laboratories. Initially, my participation consisted of presentations at the IMA on problems in my field of interest (optical systems and photonics), attending seminars, and recommending Honeywell on the governing board. Frankly, there was a bit of a struggle in the first few years trying to define an optimum mechanism to establish specific collaborative projects within the IMA. The industrial postdoctoral program turned out to be just the answer and, starting with the initial year in 1990, led to four consecutive postdoctoral appointments through 1996 under my mentorship and directed by Avner Friedman.

All four positions supported technology development within the Photonics section at the Honeywell Labs generally concerned with optoelectronic components and systems for very high speed data communications and specifically in the fields of diffractive optics and optical waveguides. The first two postdocs, David Dobson and Gang Bao, developed several design and modeling codes for diffractive elements. These codes played a direct role in winning, and executing, a ~$5M DARPA program to develop a MEMS-based, tunable spectral filter for infrared imaging. The software was also used in a number of other projects to design antireflective structures, polarizing elements, special laser mirrors, and laser beam homogenizing elements. The work that the last two postdocs, Nathan Katz and Lei Wang, did on modeling mode propagation in optical fibers and coupling of VCSELs* to optical fibers was used in one DARPA program, but it was much more valuable for Honeywell in developing a number of VCSEL products for high speed data communications. In total, the effort of these four collaborations resulted in at least twelve patents awarded to Honeywell. The VCSEL business was sold in early 2004 to Finisar for $75M, and although we can’t claim the IMA work was critical to the sale, it did result in a number of patents and product designs which certainly contributed to the overall value.

Other Honeywellers have also had significant involvement with the IMA (and MCIM, for that matter). Blaise Morton, Gunter Stein, and Tariq Samad are especially noteworthy for coordinating a number of collaborations in the area of control systems. Tariq is currently the Honeywell representative on the governing board and is a strong proponent for continuing Honeywell’s participation in the IMA, and I think the future of this relationship looks both secure and promising.

*VCSEL = vertical-cavity surface-emitting laser

J. Allen Cox, Ph.D.
Senior Research Fellow
Honeywell Laboratories

Letter from Tariq Samad

Every major technological advance in modern times can be attributed in significant part to investment in mathematics and its applications. From computers to communications, genetics to geophysics, our understanding of nature—and of our own artifacts—is actionable to the extent it is formalized. We seek to model phenomena through mathematics, to optimally design and operate our engineered systems using mathematical techniques, to foster worldwide intellectual collaborations through the universal language of mathematics, and even to capture the limits of our understanding and abilities with mathematical precision.

In this case, what’s been true in the past will doubtless hold in the future. Organizations—whether at the level of states or corporations or academic centers—that invest in mathematics will be the ones that will be at the forefront of technological progress, to their advantage over their peers and competitors and with attendant economic and societal benefits for their stakeholders.

Our state has a true jewel in this regard. The Institute for Mathematics and Its Applications (IMA) at the University of Minnesota is internationally recognized as a truly one-of-a-kind center that connects expertise in all subfields of mathematics with problems in numerous spheres of industry, government, and society. The IMA’s distinguished cadre of visiting scientists, its annual thematic programs and numerous workshops, the IMA Public Lectures, and other mechanisms have all been designed to increase the impact of the abstract discipline of mathematics on the real world and real-world problems. The IMA privileges the organizations (including the State of Minnesota) that support and engage with it, not only through the Institute’s worldwide recognition for excellence but, and more importantly, by affording them competitive advantages as a result of the scientific and technological developments accruing from IMA participation.

Tariq Samad

Tariq Samad is a Corporate Fellow at the Honeywell ACS Advanced Technology Laboratory, and represents Honeywell on the Industrial Advisory Board of the IMA.

Mathematics News from Normandale Community College

We thank Julie Guelich, Dean of Normandale, for the following update on their very active mathematics programs.

Normandale Community College and the University of Minnesota Department of Mathematics have had a close relationship for many years. More than 50% of the mathematics faculty members at Normandale received their graduate mathematics training at the U of MN. The joint Mathematics/Computer Science Department at Normandale has 32 faculty members, 4 of whom are adjunct members. Faculty members who received their graduate training at the U of MN are well-prepared to teach the mathematics courses offered at Normandale. All faculty members at Normandale have at least a master's degree, and many have a doctorate. New faculty members at Normandale have a three-year period of probation, during which time they create a teaching portfolio. Faculty members are not required to carry out mathematical research as a condition of employment, since the emphasis at Normandale is on the scholarship of teaching. Many faculty routinely give presentations at professional conferences relating to the curriculum and teaching of lower division coursework. Increasingly of late, assessment of student learning is a major focus of post-secondary education. Colleges are required to report on assessment efforts to the North Central Section of the Higher Learning Commission, the accrediting body for colleges and universities in the United States. Knowledge of and experience with the process of assessment for mathematics coursework is important for new faculty members who wish to teach at Normandale or any other two- or four-year college.

Many of the students in the Math/CSci Department at Normandale transfer to the Institute of Technology at the U of MN. The Normandale curriculum was designed with the University in mind, and coursework transfers easily. During the past year, a new degree, “Associate of Arts with Emphasis in Mathematics,” was created. It includes the lower division coursework for a baccalaureate degree in mathematics. Normandale also offers two other degrees – the Associate of Science in Engineering Foundations and the Associate of Science in Computer Science – that articulate with programs at the Institute of Technology. In August 2004, Normandale received a four-year National Science Foundation CSEMS grant of $400,000 for scholarships for mathematics, engineering, and computer science students. The Normandale mathematics faculty members hope that these scholarships will support Normandale students who are strong in math and science, allowing them to attend college full-time without working at outside jobs.

Our Alumnus, Joe Schumi - A Career in Actuarial Science

I received a Bachelor of mathematics from the Institute of Technology in 1966. At the time, it seemed this best prepared me for graduate school in mathematics. It was a good time to be in math or the sciences as the race to the moon provided generous funding.

However five years later, after Nixon was elected president, money was not so plentiful. While previously a Minnesota Ph.D. graduate could obtain a tenure-track position at a school with a nationally ranked football program, though perhaps not a nationally ranked math department, by 1971, when I received my Ph.D., some were fortunate to obtain a position at a division III school.

But not everyone was so fortunate. While teaching part time, I had become aware of the actuarial profession through on campus Actuarial Career Day presentations by the Twin Cities Actuarial Club. As a result, I took and passed the first two actuarial exams and obtained a few interviews. Unfortunately, the sentiment expressed by potential employers was that someone like me would leave the actuarial profession as soon as the academic market improved.

In the early 70’s, the St. Paul Companies was expanding its actuarial department. This was driven to a large extent by changes in the insurance industry restricting the ability of companies to share resources and requiring companies to do more on their own. St. Paul, unlike the life insurance companies I had spoken with previously, was a property-casualty insurance company, so taking a position at St. Paul represented a career choice between the two major ‘brands’ of actuaries in the United States. (Note from the editors: Joe eventually rose to the rank of Vice President.)

[Most people are probably aware that attaining recognition as an actuary requires passing a series of nine or ten exams. While the early topics are mathematical, the latter exams include accounting, taxation, regulation as well as specific insurance topics. While advanced mathematics is not required, good study habits are, as the exams are mainly self study. And while the companies support the program with exam passing bonuses and some on the job study time, the actuarial student is expected to study 300 to 400 hours on their own time every six months until the exams are completed. As the pass ratio is generally close to 40%, good students will often complete the exams in under 10 years.]

New recruits at the St. Paul were assigned to pricing units that helped develop rates for insurance policies or corporate units that worked mainly on reserves for insurance claims that had occurred but had not yet been paid. In my first assignment, I worked with the underwriters of our products for farmers and ranchers. The work included analyzing the recent results to determine if rates needed to be changed to meet the companies' profit objectives and trying to project future results. Early on, my job was to computerize rate-making formulae, first using Fortran in a mainframe environment, and after about ten years, moving to a personal computer environment.

As one progressed through the exams, the areas of responsibility broadened and newer actuarial staff were assigned as subordinates.

By the 1980’s there were many people with advanced mathematical training entering the financial professions and there were changes taking place in nature of the products a company was expected to deliver. Customers were looking for insurance products that were more closely tailored to their needs — policies that covered only very large losses or unusual accumulations of small losses. Insurance companies were looking for ways to more efficiently purchase the insurance they buy to protect themselves from catastrophic losses. Competition required that these estimates be increasingly precise. These products presented more mathematically challenging problems, which were often not solvable in tidy closed forms, but with the advent of powerful desk top computers, were computationally tractable. This was also the time that the so called ‘rocket scientists’ were entering the financial markets with tools such as the Black-Scholes formula for pricing options.

While these new models were very sophisticated, they rely on estimates of parameters and the basic assumptions that underlie insurance. The proper use of these tools then requires the actuary to improve the quality of the parameter estimates and continuously test the validity of assumptions in an ever changing environment. One of the basic principles underlying insurance is that combining the random financial results of number of similar but statistically independent entities reduces the aggregate uncertainty. Sometimes these assumptions can be taken for granted. [Until a 9.0 earthquake struck off the coast of Indonesia, who would have thought that buildings in Thailand and Somalia didn’t represent very independent risks.]

Competition continued to drive the insurance industry, the narrowing of margins driving more and more companies to the brink and/or into the arms of suitors which often led to the elimination of positions deemed redundant. Eventually, my former employer merged with one company, then another. Fortunately, I had a sufficient number of years of age and years of service to retire.

In any case, it was a fascinating time to be a mathematician in a non-traditional setting.

Joe Schumi

Talk by Michael Postol

Dr. Michael Postol, a 1990 Ph.D. graduate of our department who is employed by the National Security Agency, visited in July 2004 and gave an interesting talk on his ongoing work on “Intrusion Detection”. His presentation was attended by colleagues from several departments. Mike wrote his Ph.D. thesis in the area of algebraic topology. An abstract of the talk follows.

Title: Computer Intrusion Detection Using Features from Graph Theory and Algebraic Topology

This talk describes the problem of profiling the users of a network for the purpose of intrusion detection. We look at three graphs associated with each session on a Windows NT machine. A number of features are extracted from these graphs and fed to a classifier which uses “random forest” techniques for distinguishing between the graphs associated with one user and those associated with another. The features can be taken from the properties of the graphs themselves as well as from a variety of simplicial complexes which can be built using the information contained in the graphs.

The talk presents some initial results as well as a number of ideas we plan to pursue in the near future.