Comment Feed for Channel 9 - C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

wil2300 — Thu, 10 Dec 2009 16:45:33 GMT

Super excellent material. Great series thus far.

posted by wil2300

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

tomkirbygreen — Thu, 10 Dec 2009 17:32:09 GMT

*FX Snoopy Dance! This is a completely awesome series, how ever are you guys going to top it? Can't wait to find out!

posted by tomkirbygreen

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

paks8150 — Thu, 10 Dec 2009 18:51:34 GMT

Just when I thought that this series couldn’t get better. Thanks Erick and Charles for putting so much effort into these lectures. Also, thanks to Dr. Hutton for taking the time to enlighten all of us.

posted by paks8150

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

exoteric — Thu, 10 Dec 2009 19:34:50 GMT

This is extremely excellent! I'm going to burn this to DVDs and wrap in along with the extra book I also bought as a christmas gift.

posted by exoteric

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

dpratt71 — Thu, 10 Dec 2009 20:45:55 GMT

I agree with the other responders, very excellent, thank you Dr. Hutton.

I have to say, in a way I was disappointed with the fact that program fusion was so effective at improving the performance of the program. One of the nicest properties of Haskell is that its functions are so very composable. And because Haskell is pervasively lazy, a lot of unnecessary computation is avoided. It seems to me that "by hand" program fusion runs somewhat counter to function composition.

I wonder if there would be some way of doing this kind of program fusion at compile-time, based on some function-algebra rules engine. Perhaps this relates to Dr. Huttons research.

posted by dpratt71

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

chudq — Thu, 10 Dec 2009 20:52:36 GMT

I really enjoy this lecture. The only topic on this one is about Countdown Problem, a French TV show on resolving a mathematical problem. It'll hurt your brain if you try to resolve the problem.

Here are some links about CP:

On-line Countdown Numbers Problem. Amazingly fast!
Source codes in Haskell by Dr. Graham Hutton.

I pasted the source codes into Notepad and saved as Coundown.hs. Then I used WinHugs to open or load it. To run the problem, just type in main:

Main>main Enter the source numbers : [23, 12, 3, 5, 67, 78] Enter the target number : 101 ...

You will get one and all the solutions. Enjoy it!

By the way, I am not good at Haskell. Can anyone provide some hints on adding timestamps to get one and all solutions? In this way, I'll be able to see the start, end and execution time.

posted by chudq

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

Parmenio — Thu, 10 Dec 2009 22:23:34 GMT

This series really is excellent.

As side bars off this I've really had numerous hours of exploration into functional concepts, finally got my head around Monads, rediscovered some fun in physics and maths (Monads led to Brian Beckman!), and it's fundamentally changed the way I code in c#.

Functional programming really does enable much easier expression of solutions in the first case. Even if further optimization of the solutions ends up re-introducing complexity.

It interesting also some of the impact this has had on my higher level designs ... much of the examples in functional programming centres around solutions at the small level (in code mostly) but the wider ideas of compositionality scale up extremely well to overall solution designs at the level of full enterprise applications.

posted by Parmenio

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

Joerg W Mittag — Thu, 10 Dec 2009 23:31:32 GMT

Has anyone run this program (I mean the inefficient, naive, obvious first one) through the Supero Haskell Supercompiler? I would imagine that a supercompiler should be able to perform this kind of program fusion on its own, without human intervention. Sadly, I just can't seem to figure out how to install the thing.

posted by Joerg W Mittag

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

rgruian — Fri, 11 Dec 2009 07:51:15 GMT

Great post in an otherwise excellent series. I am looking forward to seeing all the remaining chapters.

Thanks!!

posted by rgruian

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

exoteric — Fri, 11 Dec 2009 08:26:22 GMT

I remember seeing this Supero (actually thought it was called Superhero heh) Supercompiler and would also like to know how it would impact code like this. Still, that we see this fusing technique is very valuable.

posted by exoteric

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

Graham Hutton — Mon, 14 Dec 2009 08:48:39 GMT

In reply to chudq, here's a version with timing information: http://www.cs.nott.ac.uk/~gmh/countdown2.hs

posted by Graham Hutton

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

Graham Hutton — Mon, 14 Dec 2009 08:59:34 GMT

Current implementations of fusion in compilers focus on completely eliminating intermediate data structures, whereas in this example the idea is to prune a data structure. So I would be surprised if existing compilers were able to perform this fusion step automatically.

posted by Graham Hutton

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

heebeegeebee — Mon, 14 Dec 2009 10:22:49 GMT

I took Dr. Hutton's Haskell course at Nottingham several years ago and it was great. One of the best taught courses at the university and very interesting!

posted by heebeegeebee

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

jlomax — Mon, 14 Dec 2009 15:57:45 GMT

Absolutely fantastic. Up till now I though of comprehensions as conveniences, now I see they can be very elegant processing mechanisms.

I do wonder, if I tried to write this program using C# and as much Linq as possible, how far could I get?

posted by jlomax

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

Jaunte — Tue, 15 Dec 2009 08:08:19 GMT

Great series.

A few thoughts though:

- Division can further be optimized with first parameter not equal to 1 also (since fractions are not allowed).

- Would have liked to see performance figures for the second optimization only, since it didnt involve refactoring the "pretty" solution.

posted by Jaunte

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

Graham Hutton — Wed, 16 Dec 2009 13:49:06 GMT

Good questions! I tried these changes out, and the extra division optimisation improves the performance of the final solution by around 1-2%. On the other hand, applying the second optimisation without the first only improves the performance of the brute force solution by around 20%, whereas the use of both techniques gives a speedup of around 100 times.

posted by Graham Hutton

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

dev# — Wed, 23 Dec 2009 17:46:14 GMT

From my limited experience, i think adding the prevent div by 1 or mul by 1, might increase in may be 1-2% of cases, but the extra if will kick in the remaining 98-99% of cases, so average case would go down thereotically.

Again i haven't done the perf testing, so can't comment on actual implementation.

posted by dev#

Re: C9 Lectures: Dr. Graham Hutton - Functional Programming Fundamentals Chapter 11 of 13

frou — Fri, 12 Feb 2010 01:24:13 GMT

Extremely clearly presented lecture. You're a pro!

posted by frou