gds@oahu.cs.ucla.edu (Greg Skinner) (03/07/90)
In article <1990Mar6.074924.2026@athena.mit.edu> jik@athena.mit.edu (Jonathan I. Kamens) writes: > Second, why should engineers in disciplines that don't require >in-depth physics knowledge be required to take more than two semesters >of physics? Do computer scientists need more than two semesters? The >engineers that need physics to do their major end up taking more physics >-- the electrical engineers certainly end up taking more physics, even >if that physics is part of a EECS department class, rather than part of >a Physics department class. This is an interesting question to me, something I've been exploring with some of my friends who are MIT alumni. How much math and physics do computer scientists need? As I had commented in a mail message earlier, I remember differential equations much more fondly than Sussman's class (although Abelson taught it when I took it). Perhaps it was because I didn't understand Sussman's class as well as differential equations. However, I have since found lots of uses for differential equations. Around here, for example, if you want to study the modeling and analysis of computer networks, you need a fairly good background in applied math (calculus, differential equations, probablility, linear algebra, some combinatorics) because the classes (even at the undergrad level) use it pretty freely. However, after the core EE courses at MIT, I used about 1% of the math I had learned previously. I never had to use any of the math I listed above for my compilers class, my software engineering class, my architecture class, etc. (Side point: you *do* have to use some math for computability theory and algorithm analysis, but very little of the math I listed above.) I had to use a little calculus in the computer systems class. Generaly speaking, at MIT you tend to do a lot of lab work (programming of some sort). Undergraduates tend to emphasize programming. If you are taking Sussman's class and differential equations concurrently, let's say, because you know you will need to know Sussman's material to handle the rest of the CS classes, whereas there is a chance you will never use differential equations again, you will (if a choice needs to be made), put more effort into Sussman's class. Also, most people have told me that Sussman's class is generally more difficult than differential equations, if not in the difficulty of the material itself but in the sheer amount of time it takes to complete the assigned work. --gregbo