number generator

[Marc 'BlackJack' Rintsch]

In <1173515984.515050.27400 at n33g2000cwc.googlegroups.com>, Raymond
Hettinger wrote:

> On Mar 9, 7:32 am, Marc 'BlackJack' Rintsch <bj_... at gmx.net> wrote:
>> In <1173453432.893222.308... at j27g2000cwj.googlegroups.com>, cesco wrote:
>> > Given two positive integers, N and M with N < M, I have to generate N
>> > positive integers such that sum(N)=M. No more constraints.
>>
>> Break it into subproblems.  Generate a random number X from a suitable
>> range and you are left with one number, and the problem to generate (N-1)
>> random numbers that add up to (M-X).
> 
> This approach skews the probabilities.  The OP said for example with
> N=5 and M=50 that a possible solution is [3, 11, 7, 22, 7].  You're
> approach biases the probabilities toward solutions that have a large
> entry in the first position.

I know but he said also "No more constraints".  And…
 
> To make the solutions equi-probable, a simple approach is to
> recursively enumerate all possibilities and then choose one of them
> with random.choice().

…it would be faster than creating all possibilities.  :-)

Ciao,
	Marc 'BlackJack' Rintsch