Computational Science, worth studying?

College Majors Engineering Majors
1

<p>I’m an EE student, just about to finish my bachelors and am planning to continue with grad school (masters or phd, dont know yet). Now im not quite sure what direction to continue with, but I stumbled upon Stanfords website for the department of computational science and it seemed very intesresting (because the part of EE that I enjoy the most is the math and I’m reasonably good at programming, but am not too fond of the very theoretical approach to pure math). Anyway I did a bit more research about computational science programs, but found that there are surprisingly few schools that offer programs in this field and Im wondering why? </p>

<p>Is it because:</p>

<ol>
<li>Its a relatively new field?</li>
<li>You are going to aquire the computational science knowledge you need anyways when studying other subjects like physics or chemistry, and therefore its not worth dedicating your whole career to the subject?</li>
<li>Youre better off studying computer science.</li>
<li>There arn’t very many careers out there that require a pure computational science background?</li>
</ol>

<p>On a related note, I’m going to be doing an internship as a quant engineer (quantitative finance), how do you think the admission commite would view this (positive because Im using computers to try and simulate real world situations, or negative “because I’m selling myself to the finance sector”)</p>

2

<p>I’m not sure what you mean by ‘computational science department’ in relation to Stanford. Are you talking about the Mathematical and Computational Science Program? </p>

<p>[Stanford</a> University - Mathematical and Computational Science Program](<a href=“http://www.stanford.edu/group/mathcompsci/]Stanford”>http://www.stanford.edu/group/mathcompsci/)</p>

<p>If so, then there wouldn’t be much point in looking into it, since it’s an undergrad-only department. Stanford students actually often see it as an inferior alternative to studying statistics. (Undergrads interested in majoring in statistics are directed to the MCS major, but since Stanford’s stats department is usually ranked #1, you can see why students are bitter about MCS.) If you’re looking for programs like the one above, it’s no surprise that it’s hard to find them - Stanford has a weird tendency to take well-known majors and turn them into something obscure with Stanford’s own “spin” on the field (MCS instead of stats, “symbolic systems” instead of cognitive science, biomechanical engineering instead of the ‘normal’ BME, etc.)</p>

<p>In answer to your question, it’s definitely not that the skill set isn’t valuable, since it obviously is - but given that it’s new and interdisciplinary, there will be many flexible ways that you can study it, and each university will have a different take on it and will run the study of it differently (administratively, e.g. some will have a department, others a program, others maybe just a lab or center). I also don’t think it could be #2, since it definitely does confer skills that you couldn’t get in physics or chemistry (at least, not efficiently). I’m willing to bet that many of the schools that supposedly don’t offer it simply don’t call it “computational science” or it’s subsumed under CS/stats/math departments. The wiki article also calls it scientific computing, which is a term you hear a lot more (pretty sure there are a lot more programs for that).</p>

<p>edit: The wiki article also says that it’s “most often studied through an applied mathematics or computer science program, or within a standard mathematics, sciences, or engineering program.” So that to me says “it’s an established field but it still has a obvious mother discipline and so is usually counted under that.” (Universities are often reluctant to establish interdisciplinary departments because it means they end up duplicating faculty research interests across departments that are highly related, etc.)</p>

3

<p>Hey, thanks for the interesting insight. I should have been more clear regarding the program, the one I was refering to is this [Institute</a> for Computational and Mathematical Engineering](<a href=“http://icme.stanford.edu/index.php]Institute”>Stanford Institute for Computational & Mathematical Engineering) its MS or PhD only. Glad to hear that its not #2 that was my biggest fear!</p>

<p>This webyite here lists what seems like most universities that offer a computational science degree [SIAM:</a> Graduate Programs in Computational Science](<a href=“http://www.siam.org/students/resources/cse_programs.php]SIAM:”>Programs & Initiatives | SIAM) but quite a few of them are only minors.</p>

4

<p>Yes, computational science is an extrememly valuable program. The reason it is often overlooked is because it is relatively new and because it is something that is done at very high levels of academia, government, and industry, so not many people are aware of it. Also, besides having difficult coursework, not many people understand just what the hell is going on. I know for me, I have done a lot of research on the field and still don’t understand some things. Go ask a random person what “numerical analysis” is and see what they say. Computer science is less mysterious because people can readily visualize its direct applications. Computational science exists to solve problems that normal science disciplines are incapable of solving, making it obvious why people can’t readily visualize what the field is all about.</p>

<p>Don’t make a decision yet. You need to do some more research. Check out MIT’s computation for design and optimization program. They have a report on there from the President’s information technology advisory committee on the importance of computational science. That will give you a lot of insight into what the field is about.</p>

<p><a href=“http://www.nitrd.gov/pitac/reports/20050609_computational/computational.pdf[/url]”>http://www.nitrd.gov/pitac/reports/20050609_computational/computational.pdf&lt;/a&gt;
[MIT</a> Engineering: Initiatives: Center for Computational Engineering](<a href=“http://engineering.mit.edu/research/initiatives/cce.php]MIT”>http://engineering.mit.edu/research/initiatives/cce.php)</p>

5

<p>Thanks for the links, nice to see that US government is advocating computational science :slight_smile: </p>

<p>On a related note, what do you think is a more marketable/useful skill (in industry): computational science or Analog&Digital IC design?</p>

6

<p>I would say YES because the engineering world has more of a handle of “what, where and how” they want to proceed with computational science/engineering. When I was an undergrad (now 20+ years ago), we were told to major in computational mathematics…which (in my case) meant just taking “computational versions” of math courses like ordinary/partial differential equations, linear algebra, optimization and basically a “super minor” in CS. There were no courses in virtualization, computational fluid dynamics, biological applications and what not and definitely no entire academic departments dedicated strictly to the computational world.</p>

<p>Now you have whole undergrad and grad programs in computational science/engineering, with more ENGINEERING applications, not just the math and not just hybrid math/CS degrees. I like the programs even more now (probably too late for me to take them now, but still) because it gives students who are INTERESTED in the field to go into the area AND prepares students to enter other CS/IT areas that are probably not as exciting (but with more jobs) as a solid fallback plan.</p>

7

<p>I like global travelers answer. I too have been wondering about a computational science degree. I am trying to transfer to a computer science dept as a c.c student, but want to double major in math. I know I want to get a masters or even phd, but for me the question is: in what? </p>

<p>The computational science degree is basically a discipline that trains how to use computers to solve complex problems that involve heavy math. You wouldn’t want to try to predict the weather patterns for a week with a pencil and paper- supercomputers are whats used to give you the weekly weather report. </p>

<p>On the west coast, Stanford, UCSB, and UCSD have computational science programs. You learn the math- partial/differential equations, complex analysis; physics- fluid dynamics, particle physics; and computer science- parallel computing, scientific computing. Then you can cater your program with classes across the whole math, physics, and engineering offerings. Especially in graphics and computer simulations. </p>

<p>It’s a great degree if you are interested in sciency stuff. Its a degree where you can work for a vast amount of different companies and disciplines. You would be qualified to analyze water turbulence, balistics damage, or even financial models to name a few areas that require heavy math and simulation skills. </p>

<p>Examples of research areas:
[Program</a> in Computational Science, Mathematics, and Engineering](<a href=“http://csme.ucsd.edu/]Program”>http://csme.ucsd.edu/)
[UCSB</a> | Computational Science & Engineering](<a href=“http://www.cse.ucsb.edu/research/groups.html]UCSB”>http://www.cse.ucsb.edu/research/groups.html)</p>

8

<p>Ok, thanks for the replies. It definitely seems like a subject worth studying then. I guess the next difficulty will be finding what area within computational science seems most interesting…</p>

9

<p>One quibble (though also pointed out by other posters): computational science is not “relatively new” (unless you are comparing it to something like mathematics). In fact, CS arguably came out of computational science. For example, the four founding members of the CS department at Stanford in 1962 were three numerical analysts (Forsythe, Golub, Herriott), and a mathematician-turned AI-researcher (and many other things): McCarthy.</p>

<p>There was an effort to rebrand the field 10-15 years ago, creating interdisciplinary programs (not departments) “staffed” by faculty from many different departments. As you can see from the number of programs, it has not been particularly successful. Computation however is critical in many fields, and the Federal government has set up fellowships to encourage graduate students to consider computational-oriented studies. (Search for Department of Energy Computational Science Graduate Fellowship.) When I was in graduate school, many of the best numerical analysts were engineers (aero-astro simulating flow around aircraft, EE studying how to improve the process of designing chips, civil/mechanical/… modeling fluid flow in a variety of situations or developing crash codes, etc. I went through a CS program, and came out with a CS-tinged background in numerical analysis (and currently work with computational science codes running on ‘supercomputers’).</p>

<p>Computational science is like applied mathematics in the sense that, while there is a common basis, it is applied to something, and what that something is determines what you end up doing as a career. Most technical fields have a strong computational dependence. Even in something like physics the viewpoint is that there are “three legs” to the field: experimentation, theory, and simulation (aka computation).</p>

<p>Going into academia as a computational scientist is not easy going - departments are pretty rigid and have established criteria for evaluating faculty productivity. Interdisciplinary work is not often valued highly (and may be considered a challenge to the existing structure). However, finding research funding is not necessarily harder. In government research labs (DOD, DOE, NOAA, NASA, …) computational science is valued highly, and I think that it is also in some industries (oil companies and automobile manufacturers, aircraft manufacturers, … though I do not know the raw numbers employed in those fields). </p>

<p>This doesn’t really answer the question as to whether the interdisciplinary programs make sense or not, or whether it simply best to focus on the computation from within a traditional department. Many departments will have a computational concentration.</p>