topic Re: New Coding Challenges? in BreakPoint

New Coding Challenges?

Thoric — Tue, 24 Feb 2009 12:52:20 GMT

Hi all,

Some years ago I remember reading with interest about the LabVIEW Coding Challenges, but never before felt confident enough to take any up.

The community page seems to indicate that the last one was in 2005. Are the Coding Challenges no longer offered?

Re: New Coding Challenges?

Ben — Tue, 24 Feb 2009 13:50:22 GMT

The coding challenges where intially offered by Ni and when they lost interest (about the time we started beating Ni internal entries ) Bruce Ammons (another LabVIEW Champion, yeah LV Champions! ) stepped up and started sponsoring the challenges. Sponsorship is critical because someone has to write the rules and evaluate the submisions.

There have been a number of attmptes to get a challenge going over on LAVA and the only one that took off and went anywhere was the challenge to re-write the "3-button dialog" since it was a stacked sequence nightmare. I think only three people took that one on. Jim Kring used it as an oppertunity to showcase his JKI tools so he won that one.

So there are two issue that must be confronted if you want to see another challenge.

1) Who will sponsor?

2) A Challenge that people will want to take on. This is not so easy and I should throw out some Kudos to Bruce for coming up with some good ideas. These are challenging since there are some criteria that weigh heavily in the success of teh challenge.

A) The Challenge should be interesting, perhaps solving an age old challenge (like prime numbers) a game (Tic-Tac-Toe), or something useful (the Dictionary Challenge).

B) No non-LV domain expertice required ie I don't have to be a PHD to compete.

C) Can be completed in a resonable amount of time since these are spare time projects.

D) Educational so we learn something as a result.

Re: Educational

Apart form reading Greg McKaskle postings the Dictionary Challenge was one of my major learing experiences. THere is nothing like spending a week-end home alone with LV, a case a beer, and a LV Challenge doing benchmarks comparing speed differences to give you a good understanding of how to maek LV go fast!

Here is an open offer!

If you can get a bunch of the contributors to come up with a good challenge I'll do the work of sponsoring the Challenge*.

Ben

E-mail me at

bar@DSAutomation.com

with a message subject of "LAbVIEW Challenge" if you want to talk off-line.

Ben

*Which means I will not be allowed to play along. No problem! "I must decrease so that others may increase."

Re: New Coding Challenges?

Thoric — Tue, 24 Feb 2009 14:09:16 GMT

That's very generous Ben! Thank for offering to sponsor the Challenge. So we have an appeal for Challenge suggestions.

Might I suggest the first Coding Challenge: A Rubik's Cube solver

Maybe this is a little ambitious, especially if we want to see a 3D Rubik's cube on screen, spinning and circling as it solves itself, but the challenge of being presented with a muddled Rubik's cube (six patterns of 3x3 colours) and being required to perform typical Rubik's actions (spinning planes of blocks on the axis) is quite tantalising. The winner would perhaps be the code that solves the problem in the least number of moves, not necessarily the fastest.

I'd certainly want to give that a go...

Message Edited by Thoric on 02-24-2009 02:09 PM

Re: New Coding Challenges?

Ben — Tue, 24 Feb 2009 15:02:48 GMT

Thoric wrote:
...
Might I suggest the first Coding Challenge: A Rubik's Cube solver
...

I'd certainly want to give that a go...
Message Edited by Thoric on 02-24-2009 02:09 PM

[Ben acting as a sounding board]

Are the graphics part of the challenge or are they part of the environment the Challenge takes place in?

How does time to solve vs number of steps work come eval time?

Are we going to enforce good LV style or will "weivbaled" be acceptable?

Isn't there a single best solution for any givein Rubic Cube configuration?

Can the patterns for moving a square from one location to another be boiled down into a set of algorithms ?

Ben

Re: New Coding Challenges?

Thoric — Tue, 24 Feb 2009 15:25:02 GMT

Nothing wrong with a Devil's Advocate....

Both graphics and algorithms together may make the challenge a bit weighty. Perhaps a subvi could be provided (an action engine) that performs the core of the cube manoeuvres via a method call, such as "Initialise cube", "Rotate in Plane X", "...Y", "...Z" etc, such that the challenge is purely to develop a solver algorithm (with optional GUI).

The challenge ought to be about solving the cube in the shortest number of steps. If challenger was to implement a fancy animated 3D picture GUI with rotating cubes elements, this would take far longer to solve than another that simply shows the cube unfolded on the screen. Therefore one shouldn't include the 'overall solution time' as this includes the GUI updates etc.

Messy diagrams ought not be tolerated 😉

I'm not an expert, and there are already solvers out there for Rubik's Cubes, but from what I can see there are no hard rules for creating the fastest solution. There are certainly sets of 'motions' that help move one square around, but the challenge would be to optimise the use of these algorithms to create the shortest solution.

A number of fixed starting points would be required for the evaluation of each solution, as clearly each 'random' starting point will have a different solution procedure to another.

Re: New Coding Challenges?

Thoric — Tue, 24 Feb 2009 16:59:11 GMT

Additionally there's the Golomb Ruler generator, which follows the same ilk as the Vampire numbers and Prime numbers challenges - but maybe that's all a bit too samey....

Re: New Coding Challenges?

Darren — Tue, 24 Feb 2009 17:30:13 GMT

This thread went stale back in October 2008, but we had fun giving each other puzzles to solve with (or without) LabVIEW.

-D

Re: New Coding Challenges?

tst — Tue, 24 Feb 2009 18:55:04 GMT

Not that this should prevent anyone from doing this, but there is a Rubik's cube VI online and it even has a 3D display and animation. It's probably somewhere in the devzone. I don't remember if it has a solver, but it probably does.

LAVA had a couple of real challenges too (the water effect and the mouse gestures).

Re: New Coding Challenges?

Thoric — Tue, 24 Feb 2009 20:12:55 GMT

Darren - Yes, that's an interesting thread, shame it went stale. What I'm hoping for though is to get NI on board, and restart some official NI Coding Challenges.

tst - Oh yes, I see that someone has already posted a Rubik's Cube solver. Bugger. Although it's not complete, by the sound of it, it certainly takes the originality out of the idea now.

Any of the NI forum greats interesting in suggesting challenges? Altenbach? smerciurio? Dennis Knutson?

Re: New Coding Challenges?

altenbach — Wed, 25 Feb 2009 02:50:17 GMT

Thoric wrote:
Altenbach?

Oh yes, I am all for a new challenge. Unfortunately, I am currently away, halfway around the world, in the middle of nowhere.

I would prefer if the next challenge is relatively simple to increase participation. Ben, if you get some submissions, I'd be happy to help in the selection and to fine-tune the rules to avoid silly solutions.

I have some ideas on my own, but they probably need some work first.

Re: New Coding Challenges?

Thoric — Wed, 25 Feb 2009 09:13:27 GMT

How about a Scrabble word finder. Given a tray of letters, and an arrangement of board words, the challenge would be to find the highest scoring word in the shortest possible time.

Re: New Coding Challenges?

Thoric — Wed, 25 Feb 2009 09:16:03 GMT

...or more challenging still would be a Crossword solver. Parsing information gleaned from the internet, using the clues as search fields, the challenge would be to find the solution to any given crossword.

(This is presuming it must be possible to perform automated Google searches and the like from within LabVIEW... I've not ever tried to do this.)

Re: New Coding Challenges?

Ben — Wed, 25 Feb 2009 14:08:27 GMT

Here is part of a reply I sent via e-mail

Now don't go making it so interesting I will regret not being able to compete.

I have not looked at the Rubik cube solver but could the challenge be to optimize it? Having an existing solution that may not be optimized would be a good learning experience and eliminates the domain expertise req.

I have been tossing ideas around but nothing practical at this point.

I was considering using Shane's Nugget on writing a USB driver and adapt it to the USB controlled Nerf Rocket Launcher

and then develop a challenge based on that maybe knocking down target at locations that are specified at competition time. If my wife goes for it I may even give the rocket launcher to the winner. But I am not sure how I would score something like that without having to purchase a lot of DAQ hardware. If anyoen has ideas please share them. The other issue is how would people test their code without buying a rocket launcher themselves?

A more ambitious plan would use laser tanks to battle it out in an arena. But then again maybe the laser tanks is a more resonable plan. The challenge could use a tank simulator (Like Tanks the video game) to devlop and then switch to live mode for the competition. Bot doe sthat idea require a lot of effort!

An idea I have been challenged with for a while is a way to fix the time warp you experienced by missing out on that thread that went stale last year.

If we could develop an on-going challenge that started out like a normal challenge but then leave it open for other people to latter challenge the best solution at a latter date. This would open the games up every time a new version of LV was released. It would also be good for the younguns to be able to say "I beat Christians best entry!".

At this point I think we should stop trying to come up with a resonable idea but rather toss around crazy ideas that are interesting and THEN back-down to make it resonable.

Ben

Re: New Coding Challenges?

Thoric — Wed, 25 Feb 2009 15:05:13 GMT

Ben wrote:

A more ambitious plan would use laser tanks to battle it out in an arena. But then again maybe the laser tanks is a more resonable plan. The challenge could use a tank simulator (Like Tanks the video game) to devlop and then switch to live mode for the competition. Bot doe sthat idea require a lot of effort!

Now THAT sounds interesting!

I think what you hinted at regarding a 'virtual battleground environment' for us to test our codes within could work - for a tank competition or similar. Someone would need to provide the simulator of course, which would be some initial effort. I presume the simulator would display a 3D representation of the virtual environment?

A competition where we can pitch all our codes against one another in a winner-takes-all type fight to the death would be awesome. Strategy could be far more important than code precision or speed, and hence younguns shouldn't feel at any particular disadvantage.

Other (crazy?) variations on the theme:

Robot football (soccer to my international colleagues )
3D Maze solver (stuck in three-dimensional maze, got to find path out in shortest time)
PacMan/Cat and Mouse chase - longest living mouse is winner concept. (3D?)

I'm sure there's much more we can think of...

Message Edited by Thoric on 02-25-2009 03:05 PM

Re: New Coding Challenges?

Intaris — Wed, 25 Feb 2009 21:26:55 GMT

I posted the old "Travelling salesman" problem on LAVA as a possibility. It's a simple idea, easy to visualise but no known easy mathematical "optimal" solution. Kind of similar to the prime challenge I suppose.

When I was in college, we used to have minesweeper challenges in the lab. We managed to solve even the large mine fields in around 20 seconds after many many weeks of practise. How about implementing a minesweeper solver? Not much in the way of real-world application but something everyoine can relate to and should be fun to see in action.

Shane.

@Ben Rocket launcher probably has a camera which transmits via isochronous mode which isn't supported in VISA.......

Re: New Coding Challenges?

BeCeGa — Fri, 27 Feb 2009 04:38:50 GMT

If we go along with the Robot Soccer Game, and add OOP to the challenge, here i have a small court, (comments and names are in Spanish). Still there are VIs called AI Team 1 and AI Team 2 so we could make direct games between the Teams, The robots have some parameters as battery that could be included to make the Challenge more interesting. To do this we just need to add a VI to check how much the robot moved and subtract the equivalent amount of battery from the robot.

The current AI is quite Simple it just see who is closest to the Ball and if you are not there then it will go back to its start point.

To allow different games (as the AI will always play the same way) the Robot starting positions are moved Randomly.

The only available data to think the moves will be the info stored in you robots, the position of the other robots, and the position of the ball.

With just replacing this VIs we could find who could win.

Also Style in writing the code could be taken in account.

Re: New Coding Challenges?

Neil.Pate — Fri, 27 Feb 2009 12:26:42 GMT

My first attempt at a solver.

It is not pretty, and the style changes a bit part way through, but it works (kindof...)

Grab it here.

If I get a chance, and can figure out the 3D picture API properly I will try and integrate it into the nice 3D cube visualisation that has been posted somewhere else here.

Message Edited by nrp on 02-27-2009 12:26 PM

Re: New Coding Challenges?

Ben — Fri, 27 Feb 2009 13:34:14 GMT

Wow Benjamin!

If this is as cool at it sounds....

I will take a look this week-end.

Neal,

Just don't give away the whole game. Leave some room for improvement.

Ben

Message Edited by Ben on 02-27-2009 07:34 AM

Re: New Coding Challenges?

BruceAmmons — Mon, 02 Mar 2009 01:44:09 GMT

I was sponsoring challenges for a while, but I got pretty busy and demand seemed to drop, so I just never started another one.

I do have one idea for a challenge that I could put together pretty quick. It would be a Sudoku solver. I would provide a set of puzzles (about 100), and the winner would solve all the puzzles in the shortest total time. I already have the puzzle generator and made a solver that can solve the puzzles in about 1 msec/puzzle average (if I remember correctly). I would need to put together the rules, but I can do that fairly quickly.

I can also unofficially announce that NI is expected to be sponsoring a challenge as well. The details aren't resolved yet, but it will involve control of a model of an inverted pendulum. NI is supposed to supply a trial license of their simulation toolkit for the competition. If all goes well, the winner's solution would be used during NI Week on a real inverted pendulum. If NI isn't able to do this challenge, I will run it myself at some point.

I don't recommend a challenge where people's submissions have to compete against each other. That can be much more complex to judge, and time consuming to run all the combinations. The tic-tac-toe challenge was a little tedious in that regard, especially when they started trying to evaluate all the possible remaining outcomes on each iteration.

Bruce

Message Edited by Bruce Ammons on 03-01-2009 08:44 PM

Re: New Coding Challenges?

Ben — Tue, 10 Mar 2009 13:32:57 GMT

How about "Developing a 3d Hierarchy" challenge?

THe final solution could be come a usefull tool for LV developers.

Ben