hacker.org

A finer resolution showed the answer rather plainly... and a dozenfold. But somehow I feel i still do not grasp it, remaining with more questions open than got solved.

Where is the pun?
Inhowfar is the challenge name "a big hint"?
How do the twelve numbers help? Twelve hours of sunshine casting different lines of shadow?

I know having to ask for the punch line voids the fun of a good joke, but I would really like to understand Adum's favorite challenge.

hmmn, i don't know if there's anything too clever about the name. the challenge is meant to be a radar, essentially. the name of the challenge implies you're looking for something, just like a radar does. the 12 numbers give a hint to a clock face. nothing too fancy, but i like this one =)

Very nice having Manipulate in Mathematica to play around with the possibilities - just tried ArrayPlot almost accidentally, and there was the answer!

my solution

1.) analyse the terms by hand.
2.) found some exceptions
3.) tried to analyse the server-side algorithm and entered some Letters
--> Warning: imagecolorat() expects parameter 2 to be long, string given in /home/hacker_apache/html/hacker/html/challenge/misc/twonumbers.php on line 22
4.) check php.net for imagecolorat and wrote a little tool
5.) extracted some pixels



6.) profit!

^^

this is all what i did, quite upside down... as little excuse, i have to say that the hint in the forum where only a reasonable sample of the data is needed brought me out of track... i thought somehow that the information was encoded everywhere and spread over the whole dataset... i did not think about information density in a particular area!

- first i thought it had to do with Fast Fourier Transform as Seti@Home works with it all the time. So, i tried it as hard as possible, hoping that in the coefficients there would be something...
i bought the book "FFT for pedestrians" as well!
- then i started collecting some data and wrote it down on my diary, i found two components which looked pretty linear and thought a FFT on those will produce lot of coefficients
- so, i tried to model Anybody as an oscillatory circuit with L (x in mH) and C (t in pF) and assumed it would resonate at the hydrogen line around 1.41GHz for particular x and t hoping to find ASCII stuff
- i thought at two angles and a pair of orbitals (6 values representing an orbit), or two angles and 4 RGB colors, two angles and ... i studied celestial coordinate systems and sidereal time representations... i thought of plotting the data with OpenGL on a sphere...
- then i started downloading data and tried to visualize some cuts
- then i wrote a random steepest descent hoping there would be some gradient pointing at ASCII stuff
- in the end, i downloaded the entire dataset (please excuse me adum for the server overload...), looked for valid points with ASCII and literals, got the whole dataset loaded in memory in less than 2 seconds...
- with the entire dataset i plotted cuts (with t fixed, varying x) they look pretty linear and tried to model it as base functions, hoping the coefficients would show something
- i then thought of an oscilloscope or at a movie scrolling from left to right, but first i visualized some components as gray scale and finally found the word

As last step i read the forum and understood the clock and the 12 hours, and the shadow! again congratulations to adum for thinking it out!!! btw i think "adumbra" means shadow in latin!

After 2 and half days of getting data I have decided to sample only 1/64 of required.
The images were easily readable, but nevermind ... I have grabbed more than half already so why not to finish it ... .

OK ... I made graphs of separate functions in excell (and their differences as well to make almost sure it's a graphics).

BTW, does everybody used (x*0xff/1000)*0x010101 as RGB?

Interesting that (x*0xffffff/1000) gives more readable results ... .

I just download the whole dataset and simply spot out the pixels that return 12 same numbers. Then with some sort of guess I get the answer. But it's obviously not the best approach since I have to get the whole dataset and the final image I plot is somewhat confusing. So can someone please give me the expected approach?

Seter wrote:I just download the whole dataset and simply spot out the pixels that return 12 same numbers. Then with some sort of guess I get the answer. But it's obviously not the best approach since I have to get the whole dataset and the final image I plot is somewhat confusing. So can someone please give me the expected approach?

I downloaded big portion of the dataset as well, before I start thinking about the meaning of values.
First idea worked well so I started sampling to get the image in smaller resolution first.

I thing the best approach was to think first and start downloading the samples from the opennig.

I bet downloading 1/256 of the image pixels would be sufficient to read the answer.

For me, the whole there-is-an-image-hidden thing was pretty clear right from the beginning, as soon as I had discovered the imagecolorat() warning. Some of the printed values looked like they could be RGB parts (i.e., between 0 and 255), so extracting the image was my first guess. On the other hand, I have no idea what the clock thing is, you guys talk about

I let a Python script run to receive the values for each pixel. After half a day I had half the pixels gathered, so I decided to try plotting it. And even if my order of querying was probably not that smart in hindsight, the result is still readable:



Thanks for the challenge. But what's up with that clock??

This one was fun! To represent 12 values as a pixel I used 2x2 pixel blocks of RGB (exactly 12 values). After rendering first letter I decided to fetch random pixels instead of lines, and it was clear that letters are between 100 and 300(param two), so I got readable answer pretty fast!

No idea why there are 12 iterations ...

Got some sample data, ran into exceptions from the formulas pretty fast ... well.
Querried all the data over night (19MB JSON^^). Rendered all layers into their own image and wondering whether I would/could detect anything in there. Realized I overdid it a bit.