Clustering... The hard way!

[Geoff Bolton]

Okay, hello Coding Horrors!!! A while ago I stumbled upon this delightful piece of code, which appears in a much larger monster that is (thankfully) obsolete. This loop also has a number of shorter variants covering the case of 7, 6, 5, 4 etc cameras. It cracked me up in any case, and I'd like to share it. To this day, I haven't worked out whether the person who wrote it was a fool, a genius, or both! Enjoy =)

//if there are 8 cameras, clustering 15 out of 28 pairs
if (pairCount>=28)
{
// compute the distances among points
errBounce = FLOAT_MAX;
for (i1=0; i1<pairCount-14; i1++)
{
for (i2=i1+1; i2<pairCount-13; i2++)
{
for (i3=i2+1; i3<pairCount-12; i3++)
{
for (i4=i3+1; i4<pairCount-11; i4++)
{
for (i5=i4+1; i5<pairCount-10; i5++)
{
for (i6=i5+1; i6<pairCount-9; i6++)
{
for (i7=i6+1; i7<pairCount-8; i7++)
{
for (i8=i7+1; i8<pairCount-7; i8++)
{
for (i9=i8+1; i9<pairCount-6; i9++)
{
for (i10=i9+1; i10<pairCount-5; i10++)
{
for (i11=i10+1; i11<pairCount-4; i11++)
{
for (i12=i11+1; i12<pairCount-3; i12++)
{
for (i13=i12+1; i13<pairCount-2; i13++)
{
for (i14=i13+1; i14<pairCount-1; i14++)
{
for (i15=i14+1; i15<pairCount; i15++)
{
centroid[0] = (estBounce[i1][0] + estBounce[i2][0] + estBounce[i3][0] + estBounce[i4][0] + estBounce[i5][0] +
estBounce[i6][0] + estBounce[i7][0] + estBounce[i8][0] + estBounce[i9][0] + estBounce[i10][0] +
estBounce[i11][0] + estBounce[i12][0] + estBounce[i13][0] + estBounce[i14][0] + estBounce[i15][0])/15.0;
centroid[1] = (estBounce[i1][1] + estBounce[i2][1] + estBounce[i3][1] + estBounce[i4][1] + estBounce[i5][1] +
estBounce[i6][1] + estBounce[i7][1] + estBounce[i8][1] + estBounce[i9][1] + estBounce[i10][1] +
estBounce[i11][1] + estBounce[i12][1] + estBounce[i13][1] + estBounce[i14][1] + estBounce[i15][1])/1

[Super Lloyd]

Geoff Bolton wrote:

To this day, I haven't worked out whether the person who wrote it was a fool, a genius, or both

Granted I only glanced at the code for, like, 1 minute, but I haven't figured it either! :laugh:

A train station is where the train stops. A bus station is where the bus stops. On my desk, I have a work station.... _________________________________________________________ My programs never have bugs, they just develop random features.

[leppie]

I dont think that code will terminate ;P

xacc.ide
IronScheme - 1.0 RC 1 - out now!
((λ (x) `(,x ',x)) '(λ (x) `(,x ',x))) The Scheme Programming Language – Fourth Edition

[Jeroen De Dauw]

I'd like to see how this guy would create software to manage a global system. How many TiB would the source be? :laugh:

Jeroen De Dauw
Blog ; Wiki

[supercat9]

The for() loops seem a rather verbose way of generating all 37,442,160 combinations of 15 pairs chosen from 28, but I wouldn't say they're necessarily horrible. If it is indeed necessary to enumerate all the combinations, nested 'for' loops seem a verbose but time-efficient way to do it. The w1..w15 calculations seem icky, though. Much better to compute the sum of dis1..dis15, and subtract one item from that computation.

[dybs]

Seems like a classic case for recursion to me...but I'm too tired to both trying to code it now... :)

The shout of progress is not "Eureka!" it's "Strange... that's not what i expected". - peterchen

[Lutoslaw]

Auch! It is terrible. I'm sure this can be reduced to a O(n^2) complexity with a smart algorithm and some precalculations/caching. If not, a simple recursive method can do it. :zzz:

Greetings - Jacek

[James Lonero]

My God. It looks like some of the old stuff lying around at my company.