Edward Seidel, an astrophysicist, will lead the National Science Foundation’s efforts to advance computer science by exploring new ways to connect data, computers, and people starting this September. He says Cyberinfrastructure, or CI — which forges these connections — is necessary for success in industry and academe.
The CI office awards competitive grants to researchers who are doing pathbreaking computer-science work. The office also oversees advances in supercomputing, high-speed networking, data storage, and software development on a national level. Mr. Seidel comes to the foundation from Louisiana State University at Baton Rouge, where he directs the Center for Computation & Technology.
Q. What are your priorities for advancing the foundation’s CI vision?
A. Developing a CI-savvy work force is perhaps the most important long-term investment that needs to be made. We face a critical shortage of computationally skilled researchers and staff to support them. Increasing the number of researchers who understand the importance of CI and how it can transform their fields is just as important as increasing budgets and deploying equipment that becomes obsolete in a few short years.
Q. Are some scientific disciplines better suited to promoting CI than others?
A. Atmospheric research, astrophysics, and fluid dynamics have been early drivers of CI development. At present, absolutely all areas of research, education, and industry are being transformed by advances in CI.
Q. How does the U.S. CI program stack up against programs in other countries?
A. The U.S. has been the leader in development and application of computing to advance science. On the other hand, CI is much broader than supercomputing systems, encompassing software, application development, networks, data, analysis, visualization, algorithms, and so on. In some of these areas, the U.S. can learn much from efforts around the world, especially in Europe and Asia.
Q. What are the greatest opportunities for international collaboration in CI?
A. We need to assemble teams with different kinds of expertise needed to attack complex problems, for example in climate, geosciences, astrophysics. You name it.
Q. What can individual universities do to support CI development?
A. Universities need to hire more faculty who use CI to advance their disciplines. Consider developing local training courses in computational science and the use of CI, and participate in national training events. —Andrea L. Foster
12 Responses to New Ways to Connect Data, Computers, and People
Gene Preuss - February 24, 2012 at 7:02 am
I understood the first three paragraphs….then it was all downhill… :)
johnbarnes - February 24, 2012 at 7:15 am
Actually, Robert, I think it says you stopped the story (that is, what you’re reporting to us) in the middle, and now I want to know how it comes out.
You very nicely demonstrated that they all got “creative” in terms of avoiding doing actual calculus (I’d classify what they were doing as some mixture of analytic geometry and, as you put it, software dumpster diving). At that point probably either a) you had a terrific conversation about the efficiency and effectiveness of the calculus and why it’s the basic tool for this, or b) they found a way to avoid that conversation. Students are good at avoiding difficulty (as they perceive it), and I’ll freely admit that in my own work I like brute-force methods like the graph and the table too. But the job is to see that the path to the answer via the derivative may be harder, but in a real sense it is deeper and more true, and this procedure led you to the door of that conversation. So were you able to have it?
And yes, I agree, drilling it into them as “Because it’s the right way” is unlikely to produce people who can find a) a way, b) several ways, and c) the best way — all three being important.
johnsoad - February 24, 2012 at 11:22 am
Robert, I agree completely. Young children are master learners, and they are always asking questions; if nothing else fits, they will always ask, “But why?”
I have taught a range of biology courses over the years, and the depth of learning I see is always greater when I focus on using questions to drive the course. Constructivist learning theory, and formative assessment both have espoused this approach, and now cognitive neuroscience is showing why it works. We’ve become so convinced of the power of questions and problem-solving that we are building a new textbook model around it. I’d ask a question of the larger community: when did we decide that telling students what to know was more important than getting them to ask questions?
Robert Talbert - February 24, 2012 at 3:11 pm
Thanks for the comment.
To answer your question, the conversation about the algebraic method for maximizing the speed came up in several groups. All of them had a sort of “aha” moment when I pointed it out, and a few went back and tried it out — but it’s interesting that none of them felt strongly enough about this method to go back and change their work on their lab, which they could easily have done. They came up with their approach and they were quite invested in it. So while you and I may think that an algebraic approach using the derivative is “more true”, the students do not agree, and I cannot find any kind of absolute standard that says estimating a rate of change by calculating the limit as (t to 0) of a sequence of average rates of change is any less deep, or any less Calculus, than using the Power Rule and a bunch of algebra. Are not both of these things methods for calculating the derivative?
I’d also point out that every student in the room was doing calculus — even if they weren’t using algebraic rules to take a derivative. At minimum, students had to realize that speed was the norm of the rate of change in position. How many students have we seen in calculus classes who get A’s because they can perform algebraic manipulations, and yet do not comprehend this basic concept?
I feel like learning has been much more successful for my students if they can set up the problem in terms of calculus and then find some effective way — irrespective of whether it’s graphical, algebraic, or numerical — to solve it, than if they can perform a calculation successfully and “efficiently” but had to be told what to do. If students can do the former, I am not going to be picky about whether they are using algebra or a graph or a table. They are all using calculus and that’s what matters.
Socratease2 - February 24, 2012 at 3:29 pm
I agree with the comments below, I don’t know if I should slap your students for being so dense or praise them for trying to find a Rube Goldberg solution to get to the same place (though using far more time). Have the jurors reached a verdict? Yes, your honor, slap them all down the line.
Socratease2 - February 24, 2012 at 3:35 pm
Yes, I think we can all appreciate this socratic approach to mathematics and feel good the students actually applied their math skills creatively, all very fine. But why in God’s name did not one person use the “canonical optimization method” apparently well understood by all and the most efficient way of obtaining answer. Were they told not to do that?
Robert Talbert - February 24, 2012 at 3:58 pm
I am absolutely sure that someone told them how to use algebraic derivative calculations to maximize functions — in their Calculus 1 classes. In fact the students were probably relentlessly drilled on this method for days on end. So, indeed, why didn’t this method spring to mind for them? I suspect it has something to do with the relative effectiveness of telling people what to do rather than letting them answer questions using the calculus concepts that make sense to them. One way may be more “efficient” than others, but who cares if it is, if it doesn’t stick in one’s mind and doesn’t make sense internally?
This experience, if anything, makes me a lot LESS likely to just tell the students what method to use in the future. That seems to get them through the assignment but the half-life on that kind of learning (if that’s what it is) seems pretty brutal.
waratah104 - February 24, 2012 at 6:40 pm
I thought this was brilliant.
Robert Talbert - February 24, 2012 at 9:59 pm
I stand by my students’ work. They were given a question to answer, and they did it correctly, in a mathematically sound way, and can explain exactly why their solutions work the way they do. They gave strong evidence through their work that they understand the fundamental concepts they were studying. They did their work in such a way that they will remember it and be able to apply it to a new problem further down the line. I am proud of them and what they are accomplishing.
elie_s_dad - February 28, 2012 at 2:03 pm
Dear Robert,
Thank you for your interesting and relevant columns. I really agree with you philosophy on learning math.
I am in a professional (i.e. terminal) masters degree in a mathematics. I don’t have any experience other than my own.
Just my 2 cents, but I came to math after doing something unrelated for undergrad. Although I love your ideas, I wouldn’t discount the benefit of also doing work analytically and with pencil and paper.
One example is a class we have on stochastic processes, where we were asked show the Markov property. A simple shortcut would be to observe the vector process obeys a system of SDEs and cite a relevant theorem. However, our Prof forbid us from using this and asked us to prove the Markov property by appealing to the definition of the Markov property. I think the restriction was sensible since the idea of was to understand the Markov property for vector processes (and it was good to use this particular example b/c it appears in many applications).
In my experience, pencil and paper Analytical exercises are very challenging and have a lot of downstream benefits for the student. Using software has a lot of downstream benefits for the student too of course, but students in my generation (at least myself and other American students I know) do not find this to be that challenging or hard to learn on their own (in fact I did similar things professionally before going back to grad school without any education in mathematics).
I applaud your columns and ideas. Just wanted to share my experience.
johnbarnes - March 5, 2012 at 2:49 pm
That is indeed an interesting ending to the story, and you’re right, the ability to see the calculus in the problem is likely to stick with them and be more useful than any one method of solution. Maybe the deeper mystery is why, once they see the problem in the language of calculus, they don’t see that there are many ways to solve it, and look for the one that will be fast and guaranteed accurate. You’re right, if they had Calculus I, they know that method — but it’s not the tool they reach for.
This is certainly a productive case study;there’s much more to be teased out of it in future experiences.
ayetong - April 9, 2012 at 5:23 am
Please refer to the :http://www.mill-industry.net