Generating a number between 1 and 2 java Math.random()

Question

I'm working on creating a range of numbers between 1 and 2 using the imported math.random() class.

Here's how I got it working so far, however I can't see how it would work:

int tmp = (int)(Math.random()*1)+1;

Anyonould?e know if that actually works to get the range? if not.. then what

EDIT: Looking for either the number 1 or 2.

Solution 1

If you want the values 1 or 2 with equal probability, then int temp = (Math.random() <= 0.5) ? 1 : 2; is all you need. That gives you a 1 or a 2, each with probability 1/2.

Solution 2

int tmp = (int) ( Math.random() * 2 + 1); // will return either 1 or 2

Math.random() returns numbers from [0, 1). Because (int) always rounds down (floors), you actually want the range [1,3), so that [1,2) will be rounded to 1 and [2,3) will be rounded to 2.

Solution 3

Try this.

int tmp = (int)( Math.floor( Math.random() + 1.5 ) )

Math.random() -> [0, 1)

Math.random() + 1.5 -> [1.5, 2.5)

So when you take the floor, it is either 1 or 2 with equal
probability because [1.5, 2.5) = [1.5, 2) U [2, 2.5)
and length([1.5, 2)) = length([2, 2.5)) = 0.5.

Solution 4

Ok, you want to use Math.random, but you could use java.util.Random so no casting or rounding is required:

    java.util.Random random = new java.util.Random();
    int tmp = random.nextInt(2) + 1;

View more solutions

Comments

[peter.petrov]

Not really, you should do < 0.5 and not <= 0.5 for equal probability, right?

[pjs]

@peter.petrov Mathematically it's discrete inversion of the CDF, and CDF's are <= inequalities. Computationally you could argue that the RNG yields 0 but not 1, so there's an asymmetry to the tune of one part in 2**31-1, which practically speaking makes no difference. Since I'm a statistician by trade, I tend to go with a straight-up translation the mathematical definition.

[Adam Gent]

+1 I was looking at these answers and was wondering why no one was using Random. Especially since you can replace it with SecureRandom.

[Admin]

I'm using this code for a class, we're supposed to say in the curriculum and math.random() was the only random taught.

[pjs]

@peter.petrov This also would lend itself easily to splitting the mix to a different proportion. Simply change the proportion value in the test to a new target.

[Raul Guiu]

It is part of java.util, that class is there to solve this kind of problems.

[pjs]

@peter.petrov I think you could argue that mathematically you have two equivalent ranges, while technically I don't. Yours requires more computation by a hair (addition, floor, and casting vs a single comparison). I don't think any statistical test could tell the difference in the results based on any sample size people are actually likely to draw.

[peter.petrov]

OK, I think I see. Thanks.

[Dawood ibn Kareem]

+1, but you don't need the Math.floor.

[ajb]

@pjs I think the asymmetry is more like one part in 252-1, not 231-1. Math.random returns a double which has 52 significant bits of precision.

[pjs]

@ajb Thanks for the info. I think the bottom line is that nobody would be able to tell the difference whether you use a < or a <= on this one.