Comments on: Looking for a challenge, somethign easier than P ?= NP

Question: Looking for a challenge, somethign easier than P ?= NP

b1tr0t — Sun, 19 Apr 2009 22:17:32 -0800

I'd like to find a good online resource that poses interesting computer science problems.

I've been digging through some of the modern classics papers of computer science research. Some are easy enough to understand with a little effort, others are pretty difficult to penetrate. I find that most of them are fairly difficult to find practical applications for.

I'd like to find a list of computer science problems, maybe medium to difficult homework problems for CS grad students. I'm not yet looking for grand challenges or problems big enough for a thesis. I don't want to do your homework, and I don't want to do your contract work either. I also don't want interview questions.

By: crinklebat

crinklebat — Sun, 19 Apr 2009 22:24:33 -0800

Most CS courses at most major universities publish homework problems to their websites. Why not just pick a university, look at the course list, pick a course that sounds interesting and check out the problem sets? In some cases (most familiar to me is Stanford Engineering Everywhere) they even might post lectures and/or solutions and/or exams, etc.

By: Maximian

Maximian — Sun, 19 Apr 2009 22:28:31 -0800

Have you seen Project Euler? The problems range from "moderately difficult" to "extremely difficult", at least by typical CS standards here in the US. Might be a good place to start.

The book "The Structure and Interpretation of Computer Programs" (SICP) contains a wealth of interesting exercises. The book itself is available here. There is also a wiki devoted to these exercises and their solutions.

By: b1tr0t

b1tr0t — Sun, 19 Apr 2009 22:29:42 -0800

SEE is pretty cool, but the CompSci courses they list look like entry level undergraduate stuff. I'm much more interested in upper division graduate problems.

By: b1tr0t

b1tr0t — Sun, 19 Apr 2009 22:31:12 -0800

Project Euler is also cool, and was the inspiration for my question. Most of the PE problems are more along the lines of number theory than computer science.

By: grouse

grouse — Sun, 19 Apr 2009 22:45:04 -0800

The Art of Computer Programming is famous for its exercises, which all have a number indicating their difficulty, ranging from easy to unsolved research problems.

By: crinklebat

crinklebat — Sun, 19 Apr 2009 23:13:15 -0800

b1tr0t, not all the SEE courses are undergrad courses. Just the first three. I've only taken 223A, which was a tough class but not really what you're looking for (it was robotics, so lots of physics), but both 224N and 229 are more typical graduate CS courses which would typically have a good mix of MS/PhD students enrolled with a small representation of quite advanced undergraduates.

Seconding the Art of Computer Programming rec as well, especially if you can borrow it from the library for more than a few weeks...or afford it :) Either way.

By: delmoi

delmoi — Mon, 20 Apr 2009 01:58:46 -0800

What kinds of things interest you? What do you like doing? What kinds of things do you want to be able to do?

Computer science isn't really a linear progression from neophite to real ultimate programmer. Once you get to a certain point, there are lot of branches and paths you can go on. There's 3d programming and simulation, there's AI/Machine Learning, there's high performance computing, there's language design and research, there's also mathematical theory of computation (where P != NP? comes from).

As far as my own free-time coding, I've always just come up with one huge-ass project, which is invariably abandoned at some point after spending a ton of time on it. Usually these projects involve learning something technology or skill I've never used before, so I learn a lot that way. When I was first starting out (like in high school) I would write simple demo programs, though.

Once you get to a certain point, though, you'll basically be able to code anything you can imagine. The challenge is expanding your imagination.

By: Schismatic

Schismatic — Mon, 20 Apr 2009 05:31:14 -0800

IBM Research's Ponder This ranges from pure math questions to ones that are far more like CS to my math/physics brain. Certainly some could be solved from that perspective. If nothing else, it's fun to flip through.

By: alb

alb — Mon, 20 Apr 2009 07:26:44 -0800

The New Turing Omnibus

By: chairface

chairface — Mon, 20 Apr 2009 14:50:03 -0800

I'm not entirely clear what your asking, but routing is still an active area of research with both hardware and software solutions. All solutions are forced to make time/space/cost trade-offs.

By: media_itoku

media_itoku — Mon, 20 Apr 2009 17:35:20 -0800

Have you tried the Python Challenge?

By: b1tr0t

b1tr0t — Mon, 20 Apr 2009 20:48:01 -0800

Once you get to a certain point, though, you'll basically be able to code anything you can imagine. The challenge is expanding your imagination.

I agree - expanding my imagination is what I'm trying to do now.

I'm not entirely clear what your asking, but routing is still an active area of research with both hardware and software solutions. All solutions are forced to make time/space/cost trade-offs.

Routing could be an interesting space to play in, but I don't know enough about the space to set up tractable problems that also won't require setting up a cost-prohibitive lab.