470. Implement Rand10() Using Rand7()

Given a function rand7 which generates a uniform random integer in the range 1 to 7, write a function rand10 which generates a uniform random integer in the range 1 to 10.

Do NOT use system's Math.random().

 

Example 1:

Input: 1
Output: [7]

Example 2:

Input: 2
Output: [8,4]

Example 3:

Input: 3
Output: [8,1,10]

 

Note:

1. rand7 is predefined.
2. Each testcase has one argument: n, the number of times that rand10 is called.

 

Follow up:

1. What is the expected value for the number of calls to rand7() function?
2. Could you minimize the number of calls to rand7()?

 

连续随机浮点数和随机正数还不太一样

如果是浮点数，直接平移到0开始，然后缩放，然后平移到1开始就好了

如果是整数，按照上面那样的思路然后取整得到的不是均匀的整数分布。

一个可行的思路是先得到更大范围的均匀整数分布（先乘以，然后填充里面空余的整数），然后取余

https://leetcode.com/problems/implement-rand10-using-rand7/discuss/150301/Three-line-Java-solution-the-idea-can-be-generalized-to-%22Implement-RandM()-Using-RandN()%22

 

Idea: rand7() -> rand49() -> rand40() -> rand10()

public int rand10() {
    int result = 40;
    while (result >= 40) {result = 7 * (rand7() - 1) + (rand7() - 1);}
    return result % 10 + 1;
}

Time Complexity

The total number of iterations follows geometric distribution. For each iteration in the while loop, the probabilty of exiting the loop is p = 40/49. So the average time complexity T(n) = O(1/p) = O(49/40) = O(1).

 

Correctness

Note that rand49() generates a uniform random integer in [1, 49], so any number in this range has the same probability to be generated. Suppose k is an integer in range [1, 40], P(rand49() = k) = 1/49.

   P(result = k)
= P(rand49() = k in the 1st iteration) +
   P(rand49() > 40 in the 1st iteration) * P(rand49() = k in the 2nd iteration) +
   P(rand49() > 40 in the 1st iteration) * P(rand49() > 40 in the 2nd iteration) * P(rand49() = k in the 3rd iteration) +
   P(rand49() > 40 in the 1st iteration) * P(rand49() > 40 in the 2nd iteration) * P(rand49() > 40 in the 3rd iteration) * P(rand49() = k in the 4th iteration) +
   ...
= (1/49) + (9/49) * (1/49) + (9/49)^2 * (1/49) + (9/49)^3 * (1/49) + ...
= (1/49) * [1 + (9/49) + (9/49)^2 + (9/49)^3 + ... ]
= (1/49) * [1/(1-9/49)]
= (1/49) * (49/40)
= 1/40

 

 

Generalization

Implement randM() using randN() when M > N:
Step 1: Use randN() to generate randX(), where X >= M. In this problem, I use 7 * (rand7() - 1) + (rand7() - 1) to generate rand49() - 1.
Step 2: Use randX() to generate randM(). In this problem, I use rand49() to generate rand40() then generate rand10.

Note: N^b * (randN() - 1) + N^(b - 1) * (randN() - 1) + N^(b - 2) * (randN() - 1) + ... + N^0 * (randN() - 1) generates randX() - 1, where X = N^(b + 1).

 

More Examples

(1) Implement rand11() using rand3():

public int rand11() {
    int result = 22;
    while (result >= 22) {result = 3 * 3 * (rand3() - 1) + 3 * (rand3() - 1) + (rand3() - 1);}
    return result % 11 + 1;
}

Idea: rand3() -> rand27() -> rand22 -> rand11
Time Comlexity: O(27/22)

 

(2) Implement rand9() using rand7():

public int rand9() {
    int result = 45;
    while (result >= 45) {result = 7 * (rand7() - 1) + (rand7() - 1);}
    return result % 9 + 1;
}

Idea: rand7() -> rand49() -> rand45() -> rand9()
Time Comlexity: O(49/45)

 

(3) Implement rand13() using rand6():

public int rand13() {
    int result = 26;
    while (result >= 26) {result = 6 * (rand6() - 1) + (rand6() - 1);}
    return result % 13 + 1;
}

Idea: rand6() -> rand36() -> rand26 -> rand13()
Time Comlexity: O(36/26)