Re: Converting scales of True Random Numbers causes distribution de-uniformization

sugoi.sama@xxxxxxxxx wrote:
What am I missing here? Is there any way to fix this???

If you want random numbers that don't change in their distribution when
you change the scale, you would have to have numbers that have a
logarithmic distribution.

If you have numbers uniformly distributed on the interval [0,1), which
is what are usually called 'random numbers', and multiply by 5, then of
course the distribution of the 10 possible first digits, instead of
being uniform, favors the digits 1, 2, 3, and 4.

Just replace each random number x by 10^x if all you want is the
distribution of first digits to be constant. There really is no way to
get a truly scale-invariant distribution, because making all exponents
from -infinity to +infinity equally likely is not practical.

John Savard


Relevant Pages