Unique Random Numbers

[bjb007]

I want to generate 14 random numbers
between 1 and 30 without getting any
repeats.

There's no doubt a simple way but I can't
find it. "Unique" or something similar?

Help appreciated.

Edit: I've put this in the repeat loop so that any
repeating number is discarded and the loop counter
set back by one. Works.

Code: Select all

   if tRand is not among the items of field fldRandom then
     put tRand & "," after field fldRandom
   else
     subtract 1 from i
   end if

[mokogobo]

He's some code that works. Throw it in a button and try it out.

Code: Select all

on mouseUp
  -- List of integers to randomize
  local integerList
  repeat with i = 1 to 30
    put i into integerList[i]
  end repeat
  
  -- Combine elements as comma separated list of integers.
  -- Note that combine doesn't necessarily combine elements "in order"
  combine integerList using ","
  answer "AFTER combine:" && integerList
  
  -- Sort the elements if you'd like
  sort items of integerList ascending numeric
  answer "BEFORE:" && integerList
  
  -- Sort items randomly (first pass)
  sort items of integerList numeric by random(1000)
  answer "AFTER:" && integerList
  
  -- Sort items randomly (second pass)
  sort items of integerList numeric by random(1000)
  answer "AFTER:" && integerList
  
  -- Sort items randomly (third pass)
  sort items of integerList numeric by random(1000)
  answer "AFTER:" && integerList
  
  -- Split comma separated list of integers back into an array
  split integerList by comma
end mouseUp

The initial sort of the combined array and multiple applications of the random sort are not necessary, but were included to demonstrate some behavior and differences.

I suggest looking up the "combine" and "sort" commands in the Revolution Documentation. There (and below, in short) it describes some high-level operation details.

Note: The order of the elements is not alphabetical or chronological; it is based on the internal hash order of the array. To alphabetize the list, use the sortcommand:

combine monthlyReceivables using return and comma
sort lines of monthlyReceivables by item 2 of each

Good luck.

[mokogobo]

I've put this in the repeat loop so that any
repeating number is discarded and the loop counter
set back by one. Works.

I'm not certain of the internal workings, but I'd imagine the method you describe (looping until something is found) would be generally less efficient than creating a list of the integers you want (in order), then sorting it randomly, as I did.

They both work, but the trial-and-error approach might not be the wisest of solutions.

[bjb007]

Thanks for your suggestions mokogobo.

Random numbers are fascinating.
It's said that if a series of numbers are
truly random it can't be proved.

As the Rev docs say - the same seed number
will produce the same series of results
so I added a bit of code to change the
seed before each set of numbers is generated.
Don't know if this makes the numbers
more random though!

Any ideas on how to test for randomness?

[Mark]

Hi BJB,

There are statistical tests to test for randomness. Are you sure you really need this?

Getting a unique set of random numbers is simple.

Code: Select all

function uniqueRandom theAmount
  if theAmount < 1 or theAmount > 100 then
    return empty
  else
    repeat with x = 1 to 100
      put x & cr after myList
    end repeat
    repeat theAmount
      put random(number of lines of myList) into myNumber
      put line myNumber of myList & cr after myResult
      delete line myNumber of myList
    end repeat
    return myResult
  end if
end uniqueRandom

This script is untested, but even if it contains any errors, I'm sure you get the idea. You could add parameters to this script, to create a different pool from which the random number are drawn, e.g. 1-1000 instead of 1-100.

Note that computers don't generate true random, but it is close enough for many purposes. If you need true random, you can download the rndLib stack from the developers section of the Economy-x-Talk homepage.

Best regards,

Mark

[bjb007]

Mark

Yes, there is a big difference between
P(seudo)RNGs and the kind used by
online casinos.

Where my RNG prog produces at least
one of every number 0-36 in at most
140 the longest recorded non-appearance
of a number on a real roulette wheel
is 400+.

A big difference.

[Mark]

BJB,

I am not talking of pseudo RNG's and I'm definitely not talking about... I won't even mention it again.

What I am saying is that computer generated random numbers are not random. This is connected to the fact that they need a random seed.

The library on the Economy-x-Talk website, however, does give access to true random numbers.

Best,

Mark

[bjb007]

Mark

Software RNGs are usually referred to
as PRNGs simply because they don't
produce true random numbers.

I'm familiar with random.org which I
believe does produce true random
numbers.

[Mark]

OK, bjb, seems we understand each other, even if using different words :-)

Mark

[Mark Smith]

Here's an alternative approach where the randomness is introduced by a 'sort':

Code: Select all

on mouseUp
   put randList(1, 30, 14)
end mouseUp

function randList pMin, pMax, pAmount
   repeat with n = pMin to pMax
      put n & comma after tList
   end repeat
   delete char -1 of tList
   sort items of tList by random(2^31)
   return item 1 to pAmount of tList
end randList

Many ways to skin a cat....

Best,

Mark Smith