I have written a computer program which solves Sudoku puzzles

[Asher]

Originally posted by KrazyHorse
Eh. I'm not a computer scientist or even a computer engineer. It works (in all cases I've tested), and that's what counts.

That's the attitude of the engineer. It's the easiest way to tell apart a computer scientist and a software engineer.

[KrazyHorse]

I'm a scientist when it comes to physics. I'm an engineer when it comes to cranking out code to do my bidding (in order to further physics)

[Asher]

I also notice the use of debug-print statements -- do you know how to use a debugger? There's a bit of a learning curve to using it effectively (especially with archaic debuggers like gdb), but they're far more effective than debug-print statements.

[Asher]

Originally posted by KrazyHorse
I'm a scientist when it comes to physics. I'm an engineer when it comes to cranking out code to do my bidding (in order to further physics)

That's how things have to be done in the real world. Hardcore computer scientists are more like theoretical academics than anything else.

I still don't think they've decided on what the optimal algorithm is for solving problems like this.

They're also the people who cringe when you mention using FFT -- "why would you use FFT knowing it is not completely accurate?? use a DFT with high precision!"

Anyway, that's almost a rant, but too pathetic to qualify for it, so it's just ramblings.

[KrazyHorse]

I debug using print statements. I have no clue as to the existence of any sort of utility to do that for me, nor do I know if such a utility exists in IDL. I certainly know that I have no such utility on my parents' home computer.

This code was written in notepad, was ftped to my workstation in Baltimore (1000 km away), and was compiled and run from the IDL command-line interface in an ssh shell.

[KrazyHorse]

Everybody uses FFT.

Exact DFTs can suck my balls.

[Asher]

I take it this was mostly an exercise to make yourself more familiar with IDL,which I presume is a language you use a lot in physics land?

[KrazyHorse]

This was an exercise in reducing boredom when my fiancée and family are away at work, and I have nothing to do (for Christmas holidays). It did not make me any more familiar with IDL (which is used mainly by astrophysicists and cosmologists) since I use IDL every day. I managed to bang this code out in under 4 hours start to finish, which I'm pretty proud of.

Now that it is completed I will probably attempt to do some real work in my spare time, although it's a bit annoying to write code at a location remote from your compiler and the GUI IDL has.

[KrazyHorse]

Have to find the roots of a function which depends on 7 variables (that's what I've reduced a problem down to after a week of paperwork)

[KrazyHorse]

Anyhow, what's so ugly about my code?

I was going to reduce some of those nested loops through the use of clever functions, but when I tried that something ****ed up and I didn't have the heart to debug properly. It's not exactly computationally intense, so I bit the bullet and built those beautiful staircase structures you see...

[Asher]

Since we're showing weird code in ugly languages, I must share this.

Way back when I wa in Jr. High, I wrote an economy system for a science-fiction text-based game. It was written in "MUSHcode", which is an interpreted in-game scripting language used in a variant of MUDs (dating back to the 70s).

It was a huge project, took me 2 years to complete. It's still used today in modified forms in many games.

Here's an example: it's the skeleton code for implementing a command in the Factory objects to allocate power to producing a certain widget.

The full "Factory" source is here, for old times sake: http://www.oompa-loompa.net/mush/softcode/F/FACTORY.TXT

$alloc *=*:th [switch({[u(iscommod,[secure(%0)])][cand([lte(%1,[add([v(free_power)],[v(%0_alloc)])])],[gte(%1,0)])]},11,[u(fun_alloc_power,[secure(%0)],[secure(%1)])],10,[pemit(%#,Not enough power to do that.)],01,[pemit(%#,That is not a valid commodity.)],00,[pemit(%#,That is not a valid commodity.)])]

debugging this wants to make you point your eyes out.

[KrazyHorse]

Beat this for useless: another silly IDL routine to simulate bowling scores based on several variables (e.g. strike percentage etc.)

pro scoring
pins = fltarr(24)
strike = 0.0
spare = 0.0
leftover = 0.7
logleft = alog(leftover)
number = 100000
score = fltarr(number)
for j = 0L, (number - 1) do begin
	ranpins = randomu(s, 24)
	for i = 0, 9 do begin
		if (ranpins[2*i] LE strike) then begin
			pins[2*i] = 0.0
			pins[2*i + 1] = 0.0
		endif
		if (ranpins[2*i] GT strike) then begin
			rescale = (ranpins[2*i] - strike)/(1.0 - strike)
			pins[2*i] = 10.0 < ceil(alog(rescale)/logleft)
			if (ranpins[2*i + 1] LE spare) then begin
				pins[2*i + 1] = 0.0
			endif
			if (ranpins[2*i + 1] GT spare) then begin
				rescale = (ranpins[2*i + 1] - spare)/(1.0 - spare)
				pins[2*i + 1] = pins[2*i] < ceil(alog(rescale)/logleft)
			endif
		endif
	endfor
	for i = 0, 9 do begin
		if (pins[2*i + 1] GT 0.0) then begin
			score[j] = score[j] + 10.0 - pins[2*i + 1]
		endif
		if (pins[2*i] GT 0.0) and (pins[2*i + 1] EQ 0.0) then begin
			score[j] = score[j] + 20.0 - pins[2*i + 2]
		endif
		if (pins[2*i] EQ 0.0) and (pins[2*i + 2] EQ 0.0) then begin
		score[j] = score[j] + 30.0 - pins[2*i + 4]
		endif
		if (pins[2*i] EQ 0.0) and (pins[2*i + 2] GT 0.0) then begin
			score[j] = score[j] + 20.0 - pins[2*i + 3]
		endif
	endfor
endfor
sum = total(score)
sumsquares = total(score*score)
stddev = sqrt((sumsquares - sum*sum/number)/(number - 1.0))
print, sum / number
print, stddev
end

Written during a lunch hour a few months ago

[Urban Ranger]

Originally posted by Asher
Looks like another graph coloring implementation.

Sudoku does't look anything like that at all.

[Bugs ****ing Bunny]

People actually do these things, you know. It's incomprehensible.

[Bugs ****ing Bunny]

Ye gods. This thread is geek Mecca.