Issue with Random class, or bad code?

General
#1

This weekend, I created what I thought would be a quick project. The desktop app runs on Windows and needs to generate a list of lists of random numbers within a range. (say, 1000 combinations of 10 numbers within 0 and 50). I expected that within a short range, there would be repeats. What I found however is more akin to patterns, which leads me to wonder whetehr there is something broken with the Random class.

OK: here is the code portion that generates the numbers. I generate the numbers, check for repeating numbers (a requirement is that each sequence cannot have 2 or more members with the same value) and then place the numbers in the array called M(), then copy the array to a listbox for viewing and analysis. Apol,ogies for the use of nondescriptive variables. Here is a legend:

T1: temporary counter for the main loop (values between 6 and 10 depending on user selection)
T2: temporary counter for the sequence check loop (values between 0 and T1-1)
T3 random number temp variable, before it is placed in M()

for T1 = 1 to NM ' Boucle pour la gnration d'une squence T3 =0 d = Xojo.Core.Date.Now R.Seed = d.nanosecond T3 = R.InRange(LI, LS) ' nombre alatoire entre la limite infrieure et la limite suprieure EqualMember = false if T1 = 0 then 'Le premier membre est toujours accept M(T1 - 1) = T3 else for T2= 0 to T1-1 'chercher un membre gal la nouvelle valeur if M(T2) = T3 then EqualMember = true end if next if EqualMember = false then M(T1 - 1) = T3 'T3 est unique, on l'accepte else T1 = T1-1 'T3 n'est pas unique, on recommence end if end if next

Here is a small sample of results, using 7 columns ( the program allows to select 6 to 10 columns) The patterns are clearly visible with larger samples also (more columns and more sequences). In this example, the limits are 1 and 50.

18 46 24 3 30 9 36
36 15 42 20 48 26 5
5 32 11 38 17 44 22
22 1 28 7 34 13 40
40 18 46 24 3 30 9
9 36 15 42 20 48 26
26 5 32 11 38 17 44
44 22 1 28 7 34 13
13 40 18 46 24 3 30
30 9 36 15 42 20 48
48 26 5 32 11 38 17
17 44 22 1 28 7 34
34 13 40 18 46 24 3
3 30 9 36 15 42 20
20 48 26 5 32 11 38
38 17 44 22 1 28 7
7 34 13 40 18 46 24
24 3 30 9 36 15 42
42 20 48 26 5 32 11
11 38 17 44 22 1 28
28 7 34 13 40 18 46
46 24 3 30 9 36 15
15 42 20 48 26 5 32
32 11 38 17 44 22 1
1 28 7 34 13 40 18
18 46 24 3 30 9 36
36 15 42 20 48 26 5
5 32 11 38 17 44 22
22 1 28 7 34 13 40
40 18 46 24 3 30 9
9 36 15 42 20 48 26
26 5 32 11 38 17 44

So here is my interrogation: Is my code at fault, or is there an apparent repeating pattern with the Random class, basically making it unusable where reliable pseudorandom sequences are required? Any insight is welcome.

#2

try for starters to use RANDOMIZESEED
not SEED … plus normally you only need (should) seed the random one… unless you wish to recreate a previous sequence

and without spending a lot of time attempting to figure out what you code does… remember this. You say you are using a limited range (1 to 50) and you are taking 7 values… but then you are being conditional on what values you keep, and creating various new random numbers until the pattern matches a criteria. it could be that the logic is determining what you are seeing.

If you where to just take a pure stream of random numbers from 1 to 50, I doubt very much if you would see repeating patterns outside of those due to statistical probability

#3

Thank you Dave. Regarding the code to keep only non repeating values, this is one of the first things I suspected of being bad. I don’t think that it is the cause: here is what I get without the filter (10 columns, a sample within a series of 200 sequences:

34 34 34 34 34 34 34 34 34 34
13 13 13 13 13 13 13 13 13 13
40 40 40 40 40 40 40 40 40 40
18 18 18 18 18 18 18 18 18 18
46 46 46 46 46 46 46 46 46 46
24 24 24 24 24 24 24 24 24 24
3 3 3 3 3 3 3 3 3 3
30 30 30 30 30 30 30 30 30 30
30 30 30 30 30 30 30 30 9 9

Now, I moved the seed outside the loop as you suggested and used R.RandomizeSeed once, instead of using R.Seed in the loop. The filter for non-repeating values is still off. The sample looks a lot better! I will analyse further, but your solution appear to be spot on!. Thanks again!

27 13 45 41 12 45 14 6 17 6
13 30 29 38 40 45 36 9 48 38
45 16 28 4 17 36 3 49 18 29
2 42 20 32 27 18 29 25 38 44
11 17 29 10 39 27 18 14 49 9
7 44 39 5 33 36 4 39 42 41
9 37 39 31 1 32 10 27 7 31
19 13 24 40 19 29 18 45 13 26
26 34 41 35 25 30 14 1 4 43
35 27 35 23 12 42 43 25 34 12
18 49 19 31 6 45 14 42 11 6
19 44 21 6 46 32 8 37 2 2
5 47 33 10 44 18 29 22 15 40
25 36 5 4 23 5 10 5 19 12
22 27 45 40 12 17 14 45 3 40
41 42 46 44 27 36 31 49 28 1
16 6 31 17 22 26 48 3 12 16
35 46 18 45 41 31 43 10 23 19
21 11 2 44 41 36 13 49 31 30
9 10 25 31 12 29 1 26 25 27
39 6 40 36 5 1 41 23 16 25
48 17 38 49 2 45 36 22 35 35
16 46 6 15 39 20 32 11 11 37
12 39 28 33 24 11 15 15 7 29

#4

Random is not random. It’s pseudo-random. One of the properties of a Random sequence, in almost all programming languages, is that for a particular seed value, it is guaranteed to produce the same results. This is very important to programmers. It is also why you should never use it for cryptography.

#5