Windjam

maratikus Wrote:
-------------------------------------------------------
> ZeroBonus Wrote:
> --------------------------------------------------
> -----
> > Well if you ignore how you derived the 10
> numbers
> > (which add up to 1) then they sound pretty
> random
> > to me...
>
> Think about simple example of two variables. You
> simulate a random point on a square first and then
> you project that point on the main diagonal.
> Clearly, there will be significantly more 'weight'
> in the middle than on the tails. That would be
> appropriate in certain situations but not
> appropriate in some other ones, for example, if
> your want to generate a random allocation that is
> uniformly distributed on the space of feasible
> allocations.

I don't understand what you said but if I need 10 numbers that add up to 1 and I can get those 10 numbers (a different combo each time) which are 'derived' from another sample of 10 numbers, I don't see how that isn't purely random

If you 'hide' column A in ohai's solution, you will see 10 random numbers that add up to 1 every time.