# how to generate random numbers in a range with a fixed average value ?

Discussion in 'Probability' started by B.Y., Sep 4, 2008.

1. ### jer0enGuest

too bad it doesn't have a random function

jer0en, Sep 5, 2008

2. ### illywhackerGuest

If the OP does not care about the distribution (e.g. that it be
uniform on the intersection of the hyperplane and the n-cube),except
that the points must be able to lie anywhere in [0, L], then can he
not just sample uniformly from the n-cube, and then project to the
hyperplane in some way? For example, if the sum of the sample is less
than that required, take the intersection with the hyperplane of the
line from the sample to (1, 1...1), while if it is greater, take the
intersection with the hyperplane of the line from the sample to (0,
0...0).

Or: after generating m samples, let S_{m} be their sum. Let S = nA be
the sum desired. Let T = min {L, S - S_{m}}. Generate the next sample
uniformly from [0, T].

illywhacker;

illywhacker, Sep 5, 2008

3. ### jer0enGuest

it may seem that it does not feel confident to reply to any of the comments
I have made in this post.

but it would stand to reason that, instead, it is trying to replace
mathematics with a phony digital variety, shutting out by ignoring anyone
who cares to disagree with its sophisticated logic, and who does not appear
susceptible to being convinced by it being so deliberately.

like in other fields it has been replacing transistors, resistors and
capacitors up to soda pop and scouring sponges with fakes, not mentioning
media, media content and media content players like the PC, which may
currently be programmed only if you cannot trace the exact behaviour, which
is protected by law.

if you like a resumee, it appears to be aiming to perturb, starve and
overtake humanity by replacing all things real with simulations, digitally
where possible, but all either poisoned or plain inedible, while pretending
to be feasting on it itself.

basicly I suspect that we are dealing with a known here.

jer0en, Sep 5, 2008