View Full Version : Reverse randomness
calicartel
10-21-2009, 10:52 PM
Hi good ppl i am trying to find the true number based on numbers randomized from it. the goal is to find out of the randomness work and knowing that being able to find the true number from very few randomnized numbers.
The random numbers seam to always be in two possible ranges.
I like help to reverse wich ever is easiest.
True number = 42,223
Case 1. range -+25%
47,795
38,704
38,962
40,653
31,837
42,210
39,878
45,504
46,603
33,413
38,393
Case 1. range -+3%
43,252
42,236
41,604
43,148
43,032
43,272
42,424
40,972
41,522
42,795
41,249
40,970
41,983
41,742
42,459
42,646
If there is any more info i need to provide or if you need more randomized numbers, please ask and i will get it.
I just stumpled opon this site and i do relize you guys mostly do things related to programs and security but the link that brought me here was http://www.reteam.org/papers/e59.pdf and i just hope that maybe here i can find the knowledge needed for this problem of mine. So a huge thanks in advance :)
Can you rephrase what it is you are looking for?. Have you tried fitting a curve to the numbers, say a normal or binomial distribution?. Have you tried any of the many linear congruence tests that exist?. It could be that the Berlekamp Massey algorithm is relevant to your problem, but again I'm not sure what it is you are looking for. What has confused me is that you say you are looking for the true number but then you have told us what the true number is :).
Git
calicartel
10-22-2009, 07:40 AM
No i have not tested those, i wouldnt know where to start to find and setup the tests. Not realy used to this sort of things. I tried with scripts that i made my self to break it down but so far i have failed. I have done it before but they made the randomness "better". before it was only done with a maximum of 2 decimals was was easy to reverese with 2-3 random numbers. I aim to be able to reverse the numbers with 2-4 numbers again, but to be able to do so again.
I am to find out how the randomness is generated. i have given the true number in the above situation only to help find find how the randomness is done. The goal with that figured out to be able to find the true number behind other random generated numbers.
GNIREENIGNE
10-22-2009, 08:47 AM
Is this for gambling online? :p
-G
calicartel
10-22-2009, 09:16 AM
Is this for gambling online? :p
-G
Not for any kind of gambling.
I still don't have any idea of what you mean by "true number". Is it the average or mean or median or something?. As far as where to find the tests, same place as always - Google.
if you want help from here you are going to have to explain yourself much more, I still don't have a clue what you are talking about.
Git
calicartel
10-22-2009, 12:57 PM
I still don't have any idea of what you mean by "true number". Is it the average or mean or median or something?. As far as where to find the tests, same place as always - Google.
if you want help from here you are going to have to explain yourself much more, I still don't have a clue what you are talking about.
Git
Sadly google is to big for its own good, even optimzing the search as much as possible i can not get the result i look for there.
I am sorry if i cant explain much better, english is not my first languages, but i will try again.
I want to be able to find out from wich number the randomized number comes.
Maybe here is the missing clue, the above numbers are not just randomnized from nothing they are randomnize withing a range from an unknown preset number, and its that number i would like to find, its that number i refered to as the "true number". My hope is that the way the number is randomized is predictable and therefor possible to reverse to when given 2-4 of the randomized numbers.
am i making more sense now?
If you can't find Berlekamp Massey using Google, something is seriously wrong.
OK, I have a better idea of your meaning now. There are several tricks you can try. All involve statistics, and all need a much bigger selection of numbers than you have given. I would try Linear Congruence search first (as in that paper you published the URL to) or search for lgcrack.exe. That will take a collection of numbers and test if they are formed by the X1 =(X0 + A) * B form, and can solve for A and B.
Another possibility is that the sequence is formed by an LFSR (Linear Feedback Shift Register). You can test for that with the Berlekamp-Massey algorithm. There are many ways of making a PRNG like this (unfortunately for you)
Whichever way the randomness if formed, it is highly likely that the sequence is added to or multiplied by your "true number". There could be more logical or arithmetic operations too. I don't know if you able to generate streams of these numbers with a known mean, but if you can, I would generate a stream of at least a thousand and then do stats on the results. mean, standard deviation, etc can tell you a lot, as can the shape of the histogram.
Git
calicartel
10-22-2009, 06:11 PM
i think you for that, being unexperinced the main problem when searching around is knowing the names and words, i should be able to generate couple of thosen numbers with known mean, it will take some days only. i will report my success or failor here if i possible need some help finding my way again.
No, I'm offering no more help until you tell us exactly what this problem is. This is a reverse engineering forum for programs, but I have no idea what your numbers are from. Describe the problem please or the thread will be closed.
Git
vBulletin® v3.6.4, Copyright ©2000-2018, Jelsoft Enterprises Ltd.