Haskell Project

[mameluc]

Hi everyone..
I've been reading the forum for a while, but only now i decided do join..

I'm a Computer Engineering student, and this semester i'm studying Functional Programming.
I need to make a project, but I really hate haskell, and can't get any good ideas.
Until now, i thought of a lexical analyzer for Pascal and something to find routes to places on campus, but they aren't very good.
So, i'm asking for suggestions. Anything is very welcome, except TSP please.

[Yakk]

Pathfinding is amusing. Writing a functional implementation of A* could be fun.

However, I have no idea how competent you are at writing algorithms. I could make assumptions based on the name of your degree, but those assumptions wouldn't be generous.

"A project" is exceedingly vague. What sort of subjects to you understand? What was the guidance asto what kind of projects are expected? Would a fourier-polynomial multiplier be "a project", or should it have wiz-bang GUI and some pseudo-"practical" use?

[WarDaft]

Step 1) Write a genetic algorithm to breed a neural network to an arbitrary problem.

Step 2) Tweak it to improve learning capabilities.

Step 3) Make a conceptualized real time game (obviously it need not run in real time, that just adds busy work with IO code) for the AI to compete at. It could be as simple as ARC with no visible interface.


Depending on the size of the project, do some or all of these steps.

[mameluc]

First of all, thank you for the answers.

Pathfinding is amusing. Writing a functional implementation of A* could be fun.

However, I have no idea how competent you are at writing algorithms. I could make assumptions based on the name of your degree, but those assumptions wouldn't be generous.

"A project" is exceedingly vague. What sort of subjects to you understand? What was the guidance asto what kind of projects are expected? Would a fourier-polynomial multiplier be "a project", or should it have wiz-bang GUI and some pseudo-"practical" use?

Well, I consider myself skilled in writing algorithms. I'm working with minimization problems, related to power distribution losses.
Unfortunately, my teacher was vague too. One of his suggestions was "matriculation processing".
I think it must be some kind of problem 'suitable for functional programming', but his explanations were, again, very vague, about what that means.

If i find no better problem, I'll use A* for pathfinding.

Step 1) Write a genetic algorithm to breed a neural network to an arbitrary problem.

Step 2) Tweak it to improve learning capabilities.

Step 3) Make a conceptualized real time game (obviously it need not run in real time, that just adds busy work with IO code) for the AI to compete at. It could be as simple as ARC with no visible interface.


Depending on the size of the project, do some or all of these steps.

At first, I thought of implementing Neil Fraser's Chain Reaction (cant post links yet).
It would be awesome to use a neural network to improve the AI based on user's input. Also, ARC sounds nice too.

[Yakk]

Heh. Try asking for example problems and grades, if you care about that kind of thing.

I'm more tempted by stunts than anything.
A Haskell interpreter of some imperative language (even a simplified subset) would be amusing. You could even try interpreting bytecode. Or write your own (simple) imperative language.

I've wanted to get my head around co-induction for a while now, and a Haskell project that used that to solve a problem would be interesting to me.

A 3d library that builds binary instructions to send to a GPU (ie, it doesn't control the GPU, it just writes programs for a GPU in some easy to send-to-GPU language), possibly even with hooks intended to make animation easy.

A constructive real number library (where a real number is an arbitrary sequence plus proof that the sequence is Cauchy) is amusing to write. Skip checking the proofs to start with mayhap -- or, you could write a restricted real number library that supports real numbers in a particular form (and not just those that can be proven to be Cauchy). If you need an application, a Calculator that doesn't round (it just displays a limited number of digits).

A massively parallel implementation of some reasonably expensive algorithm. The trick here, of course, could be getting your execution distributed, moreso than the functional part.

[WarDaft]

Well, numerical optimization functions are quite well suited to functional programming.

I've wanted to get my head around co-induction for a while now, and a Haskell project that used that to solve a problem would be interesting to me.

How does coinduction relate to corecursion? Wikipedia only has a useful article on the second. You really have to ditch some preconceptions of what 'the result' is to get all the examples in the corecursion article.

[amindfv]

What school do you go to? Not many colleges use Haskell (yet!)

Tom

[silleknarf]

Last year, there was a competition at the end of one of my courses to write, in haskell, a program that generates an image. Quite arbitray but there were some cool ones, like fractals you could zoom in on indefinitely. I go to University of Edinburgh and Haskell is the first language students learn for CS.