Re: Placing Balls in Urns and Expected values

In article <1166371192.995942.16570@xxxxxxxxxxxxxxxxxxxxxxxxxxx>,
"R.G. Vickson" <C6L1V@xxxxxxx> wrote:
Rob Johnson wrote:
In article <1166352415.987805.284770@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
"kunzmilan" <kunzmilan@xxxxxxxx> wrote:
There are n^n possibilities how to place n indexed balls into n indexed
urns, if balls are indexed, too.
From this number, there are n possibilities how to place all balls into
only one urn.
We have n!/(n-2)!1!1! how to chose two urn to place odd number of balls
in them, and eventually n!/(n-2)!2! , if the number of balls in both
urns is the same. These counts do not count with indexing of balls and
leave (n-2) urns empty.
If only urns are indexed, the sum of all such coefficients for all
partitions reduces to
(2m -1)!/m!(m-1).
The sum of both coefficients divided by this coefficient gives the
probability of (n-2) urns to be empty.

To compute probability, we must index the balls. Consider the case
where n = 2; 2 balls and 2 urns. If we do not index the balls, there
are 3 arrangements:

|o| | | | | | | | | |o|
|o| | | |o| |o| | | |o|
- - - - - -

but they are not equally likely. Coloring the balls (b for blue, r
for red) does not change the probabilities, however, it does show
the proper probabilities; each ball can be in each urn with equal
probability:

|b| | | | | | | | | | | | | |b|
|r| | | |b| |r| |r| |b| | | |r|
- - - - - - - -

That is, each ball will be in the left urn 50% and the right urn
50%, as shown above.

So labelling the balls determines the proper probabilities.

Very nice little demonstration. Of course, that only holds for
"classical" balls; for indistinguishable quantum-mechanical balls, the
first (unlabelled) version holds for bosonic balls, and only the middle
configuration of the unlabelled version is possible at all for
fermionic balls.

Since the original problem was stated as

Suppose that n balls are randomly placed in n urns in such a
way that each ball is equally [likely] to go into each urn.
What are the expected number of empty urns.

The part about "each ball is equally [likely] to go into each urn"
indicates a labelled ball approach. Your description of "bosonic
balls" counts partitions, where only the number of balls in each
container matters. This requires a different model. Do each of
the partitions of the "bosonic balls" have equal probability?

Rob Johnson <rob@xxxxxxxxxxxxxx>
take out the trash before replying
to view any ASCII art, display article in a monospaced font
.

Relevant Pages

  • Re: Placing Balls in Urns and Expected values
    ... urns, if balls are indexed, too. ... the sum of all such coefficients for all ... To compute probability, ...
    (sci.math)
  • Re: chess problem
    ... I'm not sure the urns formula will do it; I think it's a little more ... Now imagine all the cars in a row, ... make it impossible to park the SUV we need to place the s - c empty ... The urns are the c+1 slots between the cars, the balls are the s-c ...
    (sci.math)
  • Re: m balls k urns
    ... >urns shoud be distinguishable. ... There are m indistinguishable balls and k distinguishable ... distinguishable urns, with repetitions allowed. ... throws of 3 dice with 6 sides each there are. ...
    (sci.logic)
  • Urns and balls
    ... A number of elementary problems in Probability have the equivalent model consisting in to draw balls randomly among urns. ...
    (sci.stat.math)
  • Estimating correlations between groups using binomial sampling
    ... Consider a large number of urns containing balls of various colours. ... Looking at the raw data, I suspect that the numbers balls of different ... I might estimate the correlation ...
    (sci.stat.math)
Loading