One might say that in the first case, that
of Fig.
, yielding
Eq. (
) starting from
there were no priors on
.
But this is quite not true, because the flat priors
on
and
impinge on the prior on
, due
to the relation
. The easiest way to
see what is going on is by Monte Carlo, that is, in R,
n = 10^7
rM = 100
r1 = runif(n, 0, rM)
r2 = runif(n, 0, rM)
rho = r1/r2
rho.h <- rho[rho<5]
hist(rho.h, nc=200, col='blue', freq=FALSE)
abline(v=1, col='red')
where the selection of the values below
is to
visualize the more interesting region,
shown in the top plot of Fig.
(a more complete script, which also
performs the correct normalization of the histogram,
is shown in Appendix B.4).
The histogram is characterized by
a plateau till
, followed by a slow decreasing.
Curiously, the histogram does not depend on the maximum
value rM.
Figure:
Distribution of
implied
by flat priors on
and
in linear and log scale.
The vertical line in the upper plot
shows the discontinuity of the distribution at
.
 |
Although it might be bizarre, this histogram shows in essence
the prior on
we have been tacitly assumed,
when flat priors on
and
were chosen (as a cross check,
the commented instructions of the script of Appendix B.4, executed one by
one, plot the distribution of
assuming
a flat prior for
and the curious distribution of
the top plot of
Fig.
for
).
In order to have a better insight of what is going on,
the bottom plot of the same figure shows the histogram
of
. The maximum is at
and it decreases symmetrically, exponentially,28as
increases.
This symmetry indicates that the probabilities
to get a value of
below or above 1 are the same.
The same conclusion, within the uncertainties
due to sampling, can be drawn from the histogram in linear scale,
since
is `about
' for
. Similarly,
from the comparison of the two histograms we can
evaluate, by symmetry arguments, that the probability that
is between 0.1 and 10 is equal to 90% (exact value, indeed
as we shall see in a while).
It is interesting to get the distribution shown
in the top plot of
Fig.
making a transformation
of variables, as we have done in Eq. (
) and following
equations:29
where
is the maximum value of
and
.30
At this point, some care is needed with the limits of the integral
over
, due to its `natural' upper limit at
and to that
given by the constraint
, i.e.
.
Therefore, after the trivial integration over
, we are left with
where the upper limit
depends on
in the following way:
and therefore
that we summarize as31
which, indeed, does not depend on the the maximum values
of
and
, as we had already learned playing with
Monte Carlo simulations.32
For completeness, let also make the game of seeing how
flat priors on
and
(up to
and
, respectively)
are reflected into
in the model of Fig.
:
 |
 |
d d |
(104) |
 |
 |
d d |
(105) |
|
 |
d |
(106) |
where the extremes of integration are
and
.
Here is, finally, the pdf of
, in which we have written
explicitly the conditions:
Figure:
Histogram of
implied by
priors on
and
flat up to
s
,
compared to the exact evaluation of the pdf (solid line).
Dashed lines: pdf of
for
s
(higher to lower, that is from steeper to flatter).
 |
An example with
and
s
is
reported in Fig.
,
in which the exact pdf (blue solid line)
is compared with the Monte Carlo result.
The plot also shows the pdf's of
for increasing
maximum values (
s
,
from higher to lower curves).
We see that for
and
also the distribution of
becomes flat. This is an interesting result, showing that,
contrary to the model of Fig.
,
the model of Fig.
can accommodate
in practice flat prior distributions for the three quantities of
interest.33