Suggestions and Strategies

The suggestions that follow below are based on detailed comments from the focus sessions, survey, interviews with mathematical and computational scientists and managers on site visits, individual interviews at meetings, and finally on the experience of the committee members. They are also informed by the recommendations made in the 1996 report. In fact, we believe that the suggestions made in 1996 are still valid. The suggestions we make here are variants on those in the previous report, and the differences are a matter of emphasis rather than radically new proposals.

The principles guiding the educational suggestions here are knowledge of a relevant application, real-world problem-solving experience, facility with computation, communication skills, the ability to work in and provide leadership for an interdisciplinary team, and the desire to develop a sense of a business and its objectives.

5.1 Global Strategies

In July 2009, the Organization for Economic Co-operation and Development issued a report on mechanisms for promoting mathematics in industry [OECD 2009] that contains an impressive but not necessarily comprehensive list of programs. The list includes:

Academic Initiatives

Academic-industrial collaborations

The United States, Canada, and most countries in Europe have developed programs that fall into the most of these categories. As the report points out, many of these programs are organized by academia, and governments provide the main resources via academic channels.

For example, the Canadian center, Mitacs, has a business development team that consists of members who have a strong scientific background, business development abilities, and a commitment to facilitating the development of multi-disciplinary research collaborations with non-academic partners. The members report to the Executive VP, Business Development and work closely with the Mitacs scientific staff. They are distributed in regions of Canada and funding is provided by government and industry partners. The major tasks of the team members include introducing companies to collaborations with academia for advanced research, finding internship opportunities for students by connecting university researchers with industry on research projects, (see [“Mitacs Accelerate” 2012]), creating and maintaining a network of partners in the private and public sectors, formulating new partnerships for Mitacs, and managing relationships with key stakeholders.

A second example of a center with a strong industrial focus is the DFG Research Center, Matheon, in Berlin, Germany. Matheon develops mathematics for key technologies and supports partners in industry, the economy, and science. It also interacts with schools and the general public to increase the visibility of applied mathematics.

The center fosters awareness in industry in several ways. It maintains a website that explains the expertise of Matheon's members with pointers to industrial projects, references, and success stories. See [“Matheon Services” 2012]. It has a transfer office that acts as a knowledge broker and matches requests from industry with the appropriate group. It offers industry internships to graduate students. Students work on R&D problems of industrial partners under the supervision of Matheon scientists for four months. The financing of students and scientists is provided through a partnership of Matheon and the industrial partner. This has turned out to be a win-win situation for industry, students and academia. And it carries out industry-funded projects in the areas of discrete mathematics, optimization, numerical analysis, scientific computing, and stochastic analysis.

The report concludes that “the creation of national and international networks can both stimulate mathematical awareness and creativity concerning industrial problems and avoid duplication of intellectual effort.” We support the OECD initiative and suggest that the United States should strongly support it. One outcome of the initiative should be the development of a best practices document and a discussion of metrics for evaluation of knowledge transfer along the lines of those developed in [Holli 2008].

Perhaps because the report above focuses on mathematics and not specifically on computational science and engineering, it does not address the collaboration between industry and government laboratories on high-performance computing. We discussed this earlier and include recommendations below.

 

5.2 Graduate Education

Graduate education in the mathematical and computational sciences provides graduates with specific technical expertise, the ability to think analytically, formulate problems, and develop mathematical models. All of these are key demands in industry. However, these technical skills are not by themselves sufficient for a graduate to succeed in an industrial career. The graduates in our survey and the industrial scientists and managers in our on-site interviews emphasized a set of additional skills and experiences that are needed within industry. We mentioned these skills above in §3.3 and discuss them in more detail here.

Exposure to a relevant application and real-world problem solving

The majority of the industrial scientists we interviewed work in companies that have internships and use them to find potential employees. Both the respondents to our survey and the industrial scientists strongly suggest industrial or government internships as the best course for gaining experience on real-world problems. Obtaining an internship presupposes coursework in a relevant application discipline. Math departments should maintain a database with links to national summer positions and internships in industry or government laboratories. Faculty and administrators should provide encouragement and support applications. But an even more effective way to connect graduate students with internships is to routinely invite your PhD and Master’s graduates who have taken jobs in industry or government to come back to the department and give a colloquium or workshop. Ask your alumni to speak about their jobs and the opportunities in their organizations.

Expertise in programming

We hired a bright candidate with a PhD in topology. However, because he was too humble about his expertise in programming, he almost lost a job offer.

A story like this one should never happen. Students need to know that programming is often an essential tool in industry, and if they have these skills they should certainly not hide them or be afraid to mention them.

The programming requirement varies by industry, by company, by department and by work group. In a small to medium-size company, everyone may be expected to contribute to the IT effort. The programming languages in which expertise is expected also vary. In some cases, a fourth-generation language, such as MATLAB, R, SAS, or SPSS is sufficient. In other cases, a programming language such as C++ or Java and a high-level scripting language such as Python are required. (Some interviewees even spoke disdainfully of job applicants who know “only” MATLAB.) It is therefore very important for potential job applicants, and their mentors, to find out as early as possible what are the current expectations in their desired industry concerning programming expertise.

High-Performance computing

A faculty is doing a disservice to their students if it does not offer a course in parallel computing.

A 2008 white paper sponsored by DARPA, the DOE, and the Council on Competitiveness [Council on Competitiveness 2008] concluded that “American industry is in the midst of a new 21st-century industrial revolution driven by the application of computer technology to industrial and business problems. HPC plays a key role in designing and improving many industrial products … as well as industrial-business processes.” Our study provided many examples where high-performance computing was helpful or even critical in solving business problems. The lack of qualified personnel was mentioned, in one form or another, by several of the individuals we interviewed.

For students, it is a distinct advantage to develop skills in modeling and computation in a particular application. It is also an advantage to develop skills in high-performance computing. A combination of these skills is certainly in great demand.

Communication and teamwork

It is important that during their education, students work on team projects with more than two members.

Communication skills remain as essential today as they were in 1996. But these skills go beyond writing and presenting well. In business, an individual’s success is most often determined by the success of the team(s) he or she was a member of. To be an effective communicator on a team, you need broad enough technical skills to understand what other members are saying. The ability to listen to and learn from other team members is just as important as the ability to generate your own ideas. You need leadership and presentation skills to get your ideas across, a strategic sense of the team’s goals, and the drive, discipline, and energy to meet project deadlines. One industrial scientist summarized this list as the “get-things-done factor.”

Summary of suggestions for students

Depending on your department’s policies and experience with preparing students for industrial careers, you may need to be pro-active and go beyond the minimum requirements for your degree. Seek out opportunities to speak with alumni of your department who pursued a successful career outside academia. Develop multidisciplinary skills including an application area, computing, and mathematical modeling. As you progress in your chosen discipline, don’t forget to develop a broad understanding of the mathematical sciences. Employers like applicants with a “T-shaped” profile, with depth in one area and breadth of understanding.

Pursue opportunities for summer jobs, cooperative employment, and especially internships in industry or government laboratories. These activities are likely to be helpful in finding a permanent job, as well as whether an industrial career is helping you decide right for you.

When it comes time to select an advisor, look for an advisor or co-advisor who has experience in research collaborations with industry or government scientists. You may want to find your co-advisor outside of your department.

Finally, be on the lookout for talks by industrial or government scientists in departments outside of yours or at meetings of scientific societies.

 

5.3 Suggestions for Faculty and Administrators

The following suggestions are designed to enhance connections among faculty in the mathematical and computational sciences, faculty members in the physical, social, and economic sciences, and nonacademic colleagues working in related disciplines. We note, however, that none of these suggestions are intended to substitute for the requirement to teach the core courses in mathematics or computer science.

A recurring issue in our interviews with industrial scientists and managers was the difficulty of negotiating intellectual property rights with universities. One interviewee said,“ A few universities handle IP issues well, others make it very difficult.” In cases of substantial collaborative research projects, faculty and administrators should be willing to negotiate a way around IP roadblocks. Administrators especially should be warned that an over-zealous approach to intellectual property is a good way to discourage industrial partners and impair the educational mission of the university.

 

5.4 Suggestions for Industrial Scientists and Decision Makers

Collaborations among industrial groups, government laboratories, and academic programs in the mathematical and computational sciences can be mutually reinforcing approaches to achieving organizational goals. This is, in the end, what technology transfer is about. It is not a one-way street. Therefore, we provide a list of suggestions that we believe can help industry, government, and academic organizations use the available mathematical and computational resources for mutual benefit.

For policy makers, our interviews highlighted two areas of government policy that adversely affect the environment for industrial research. Both of these areas extend far beyond the mathematical sciences, but we mention them here for completeness.

First, restrictions on visas for foreign students and nationals continue to be a burden. For example, computer-aided geometric modeling is an area with fewer American students compared to Europe; thus American aerospace and automotive companies that rely on geometric modeling struggle to find enough qualified candidates. While visa policies may depend on many other factors, scientists need to make their voices heard and emphasize that the mobility of qualified scientists can only benefit American competitiveness.

Second, our interviewees at several companies pointed out that intellectual property policies at universities involving government-funded work, since the Bayh-Dole Act of 1980, have tended to discourage cooperation with industry. Although there are examples of successful policies, there are also examples of policies that make it difficult for industry to cooperate. University decision makers should review their policies to make sure that they encourage cooperation.

Renew SIAM · Contact Us · Site Map · Join SIAM · My Account
Facebook Twitter Youtube