Computer Science Research

[Xaos]

I'm currently employed in an Ecology research lab as a summer internship, and I'm around people doing ecological research all day. It really peaked my interest in Computer Science research, so I've been looking at it a lot. One thing I cannot seem to grasp, however, is what exactly Computer Science research is. Is it just creating new algorithms to solve problems? Figuring out the best UI tactics and the best programming strategies? Is it just Math research, but with computers? Or is it computer research with math? I understand I'm probably being very short-sighted and narrow-minded about this, but I really need someone to clear it up. Also, I've looked for papers, but can't find anything that sparks my interest that is of reasonable length...Maybe this is just because I have a very small knowledge about the theory and don't know the real significance of some of the papers?

Sorry for my ranting, but I hope someone has some answers

[Jackolantern]

CS is kind of odd in this respect. The academic world of CS and the professional world are much further apart than most other sciences. In the professional world you will of course be making software, maintaining databases, etc. In the academic world, it does break down by the topic (testing UI designs like you mentioned would be under "Human-Computer Interactions" or HCI), but the software engineering side would likely be trying to discover and improve algorithms and data structures, or trying to tackle other difficult problems (such as faster means of sorting more data points than you can realistically store in main memory).

I will be honest, I hate that crap. I do know a fair number of the core algorithms and data structures they deal with, but the idea of sitting around and thinking about them and testing them for years knowing your reward is a publication in a CS journal just doesn't sound enticing to me. If I was going to do any kind of research, I would rather do research in the physical sciences. I guess if you really want to have your name tied to an algorithm for all times, such as Kruskal or Warnsdorff, go for it. That would be kind of cool, I guess.

The one thing that I do think is kind of interesting in CS research today is the wild west opening up with quantum computing. Just like the coming of digital computers in the 1940's opened up a new world of potential algorithms for solving problems in entirely new ways, quantum computing will likely do the same thing again in our lifetimes. And you don't even need a quantum computer to test them, since a digital computer can perfectly simulate a quantum computer, just at a slower rate. Heck, you can find quantum computer emulators today on almost any platform, including Javascript. Most quantum algorithms in existence today, such as Shor's, were created before the first quantum computer had been tested.

The memory in quantum computers today is just horrendous and unrealistic right now (Google's "top of the line" quantum computers will have 512 qubits), but this is just now digital computers started, too, with the first systems taking up a semi-truck bed just to house 128 bits. And then a few decades later microchips changed everything, and I have no doubt something will happen to revolutionize quantum computers eventually as well. So that could be a very interesting point of research.

[Xaos]

Yeah, the algorithm stuff doesn't sound too interesting -- sounds more like you'd be suited doing Math research than CS. Quantum computers are super interesting, and I agree they'll take off. I'm also kind of interested in interdisciplinary research, stuff like computational economics, computational physics, computational astrophysics, etc.with more of a focus on the CS side than the other side.

[Sharlenwar]

Okay, so what the heck is Quantum Computing? Discuss? I apologize for derailing this thread.

[Xaos]

Oh no, you're okay. Quantum computers basically use micro-molecules to unleash power that traditional computers cannot.


http://computer.howstuffworks.com/quantum-computer.htm

[Jackolantern]

Okay, so what the heck is Quantum Computing? Discuss? I apologize for derailing this thread.

We all know how digital computers work: an individual bit can either be on, 1, or off, 0. Electrical gates is what allows computers to operate so fast in this way.

Now quantum computers exploit a couple of principles of quantum mechanics, particularly superposition (the theory that very, very, very small things, like electrons, exist in many different states at the same time) and quantum entanglement (that you can force two particles with their own independent states to become a single state, thus making their properties dependent on each other). The culmination of these concepts when applied to computer science is the creation of the "qubit" ("quantum bit").

A qubit can be in many more states than a digital bit. Entangled qubits exhibit a very odd behavior. For example, say you have 2 entangled particles that are registered as opposites. If particle A is 0, particle B will always be, and has always been, 1, ever since particle A was 0. This is maintained across any length of space. And superposition allows them to be in multiple states at the same time.

As for the total number of states, 2 qubits can be in any number up to 4 states at a time. 3 qubits can be in any number of to 8 states, and this continues allowing a quantum computer with n qubits to be in any number of states at the same time up to 2^n. A digital computer with n bits can only be in one of 2^n states. This is extremely algorithmically powerful, as demonstrated by the few quantum algorithms that exist today, such as Simon's and Grover's algorithms.

[Sharlenwar]

Okay, so what uses would we have quantum computing for? I know, these might sound dumb, I already know the answers, just entertaining other people's brains.

[Jackolantern]

Okay, so what uses would we have quantum computing for? I know, these might sound dumb, I already know the answers, just entertaining other people's brains.

They can solve certain problems exponentially faster than digital computers. For example, Grover's Algorithm can search an unordered database in O(n^1/2) complexity using O(n) storage spaces. That is incredible, considering that a digital computer would be around O(nlog(n)) complexity with O(n^2) space requirement without hashing (which can be fairly complicated to make a hash strategy that will get better results than O(nlog(n)), and would be impossible to beat O(n^1/2)).

[Xaleph]

Another (famous) example is the traveling salesman problem. This challenge has been living in the CS world for quite some time now and no suitable algorithm has been created to successfully tackle this problem. The problem is quite simple, you have a travelling salesman who wants to figure out the shortest path to take from city to city. Only visiting each city once.

If there are <= N cities where N is for example 10, a naive algorithm could probably solve it.
If N = 40, you`ll have O(log(N^N-1)) which could turn out to be a real pain in the ass. As the algorithm requires exponential growth, the complexity grows. So while you can make really good approximations using heuristics ( either by sampling or plain guessing ) you won`t be able to solve it using normal computers.

Enters quantum computing. As Jack explained, for each consective qubit, your states grow exponentially, whereas 4 bits equal 4^2=16 states, 4qubits equals 4^4 = 256.

8^2= 256 ( 8 normal bits )
8^8 = 1.6 million ( 8 quantum bits )

so imagine 128^128.. This number is so big that I think it has more posibilities then there are particles in the universe. Not sure though.
In any case, there was a calculation that 300qubits had more processing capabilities than all_the_computers_in_the_world_combined. This is the kind of computing you need to solve complex or NP-hard algorithms and problems.

So yeah, CS research does focus on these situations, however, you do need a goal or purpose to research something. For example, a friend of mine did his research paper on augmented reality. He has a lazy eye so he doesn`t know what how or what uor vision is, he cannot estimate depth like we do, he has to do this conciously, whereas we do it subconciously. He also cannot "guess" the size of a coffee cup for example. Whereas, since we have 2 eyes, our brain tells us how big it is.

He challenged himself to tackle how we percieve depth and he wanted to see if there was a way to estimate depth based on monocular vision. For example, calculating the distance from the source ( camera, webcam what have you ) to some objects in the picture. It requires a big database of comparison material but he did pull trough. Based on shape recognition, he managed to come up with some nifty algorithms to calculate distances and sizes of objects.

I think it`s cool to do CS research. I`m thinking of writing my thesis on eye gazing for example. You`d be suprised how hard that can be and what kind of processing power it requires. There is a lot of research fortunelaty so I can start where others left of.

[hallsofvallhalla]

Pshh that is an easy answer. Cut his body into pieces equal to the amount of cities there are and mail each one off. Done! I am going to lunch.