+ 2

[Solved][CharGen]math.random() is not random enough

It seems that all the times i have it run, i am getting a lot of middle numbers. not very many highs or lows. for example, if i want a random number between 0-5 i get a lot of 2s and 3s some 4s and 1s but almost never any 0s or 5s. Update: still not seeing any range variation after using "java.util.concurrent.ThreadLocalRandom;" Solution: i needed to tweek the numbers it was randomly generating to extend the range and reduce the offset.

10th Jun 2017, 8:31 PM
oddeofreq
oddeofreq - avatar
3 Réponses
+ 5
I copied your "Character generator" program, stripped out everything that wasn't a random() call, then calculated the balance of each result over 100,000 iterations (for each category). After compensating for the "stats retries" (for being too low) -- and just counting retries for 11 -- the output looks fairly balanced, so you might be seeing local effects. I didn't test for any sort of grouping, but (like flipping tails 100x in a row) an infinite RNG could hang around a place for "as long as it wants", since I think it only has to even out after infinite time. Forcing apparent smoothness in a shorter time is non-random, but it's an interesting idea... https://code.sololearn.com/cfcWVOC88168/?ref=app (Percentage totals won't be exactly 100% due to rounding)
11th Jun 2017, 12:11 AM
Kirk Schafer
Kirk Schafer - avatar
+ 4
I think localRandomThreads would be the 'most random'. Try testing and comparing Math.random to it. This is the code: import java.util.concurrent.ThreadLocalRandom; int randomNum = ThreadLocalRandom.current().nextInt(0, 6); This should produce a random number from 0-5.
10th Jun 2017, 9:03 PM
Rrestoring faith
Rrestoring faith - avatar
0
Probably you have to modify the seed like C
10th Jun 2017, 8:55 PM
Andrés04_ve
Andrés04_ve - avatar