The Rademacher distribution is a probability distribution and can therefore be assigned to the mathematical sub-area of stochastics . With her it is a simple univariate discrete probability distribution on , among other things, to define the symmetric simple random walk on is used.
{
-
1
,
1
}
{\ displaystyle \ {- 1,1 \}}
Z
{\ displaystyle \ mathbb {Z}}
It is named after Hans Rademacher (1892–1969).
definition
The Rademacher distribution is defined on and has the probability function
{
-
1
,
1
}
{\ displaystyle \ {- 1,1 \}}
f
(
n
)
=
{
0
,
5
if
n
=
-
1
0
,
5
if
n
=
1
{\ displaystyle f (n) = {\ begin {cases} 0 {,} 5 & {\ text {if}} n = -1 \\ 0 {,} 5 & {\ text {if}} n = 1 \ end { cases}}}
The distribution function is then
F.
X
(
t
)
=
{
0
if
t
<
-
1
0
,
5
if
-
1
≤
t
<
1
1
if
t
≥
1
{\ displaystyle F_ {X} (t) = {\ begin {cases} 0 & {\ text {if}} t <-1 \\ 0 {,} 5 & {\ text {if}} - 1 \ leq t <1 \\ 1 & {\ text {falls}} t \ geq 1 \ end {cases}}}
properties
Expected value and other measures of location
The expected value of a rademacher distributed random variable is
E.
(
X
)
=
0
{\ displaystyle \ operatorname {E} (X) = 0}
.
The median is
m
~
=
0
{\ displaystyle {\ tilde {m}} = 0}
.
Variance
The variance corresponds to the standard deviation :
Var
(
X
)
=
σ
X
=
1
{\ displaystyle \ operatorname {Var} (X) = \ sigma _ {X} = 1}
.
symmetry
The Rademacher distribution is symmetrical around the 0.
Crookedness
The crooked thing is
v
(
X
)
=
0
{\ displaystyle \ operatorname {v} (X) = 0}
.
Excess and bulge
The excess of the Rademacher distribution is
γ
=
-
2
{\ displaystyle \ gamma = -2}
.
That’s the bulge
β
2
=
1
{\ displaystyle \ beta _ {2} = 1}
.
Higher moments
The -th moments are
k
{\ displaystyle k}
m
k
=
{
0
if k is even
1
if k is odd
{\ displaystyle m_ {k} = {\ begin {cases} 0 & {\ text {if k is even}} \\ 1 & {\ text {if k is odd}} \ end {cases}}}
entropy
The entropy is
H
(
X
)
=
log
2
(
2
)
{\ displaystyle \ mathrm {H} (X) = \ log _ {2} (2)}
measured in bits .
Accumulators
The cumulative generating function is
G
X
(
t
)
=
ln
(
cosh
(
t
)
)
{\ displaystyle g_ {X} (t) = \ ln (\ cosh (t))}
.
That’s the first derivative
G
X
′
(
t
)
=
tanh
(
t
)
{\ displaystyle g '_ {X} (t) = \ tanh (t)}
and therefore the first cumulative . There are closed representations for the higher derivatives .
τ
1
=
0
{\ displaystyle \ tau _ {1} = 0}
Moment generating function
The moment generating function is
M.
X
(
t
)
=
cosh
(
t
)
{\ displaystyle M_ {X} (t) = \ cosh (t)}
.
Characteristic function
The characteristic function is
φ
X
(
t
)
=
cos
(
t
)
{\ displaystyle \ varphi _ {X} (t) = \ cos (t)}
.
Relationship to other distributions
Relationship to the two-point distribution
The Rademacher distribution is a two-point distribution with .
a
=
-
1
,
b
=
1
,
p
=
q
=
0
,
5
{\ displaystyle a = -1, b = 1, p = q = 0 {,} 5}
Relationship to discrete equal distribution
The Rademacher distribution is a discrete uniform distribution on .
x
1
=
-
1
,
x
2
=
1
{\ displaystyle x_ {1} = - 1, x_ {2} = 1}
Relationship to the Bernoulli distribution
Both the Bernoulli distribution with and the Rademacher distribution model a fair coin toss (or a fair, random yes / no decision). The only difference is that heads (success) and tails (failure) are coded differently.
p
=
q
=
0
,
5
{\ displaystyle p = q = 0 {,} 5}
Relationship to the binomial distribution and the random walk
Are independent rademacher distributed random variables, then is
X
1
,
X
2
,
...
{\ displaystyle X_ {1}, X_ {2}, \ dots}
Y
n
: =
∑
i
=
1
n
X
i
{\ displaystyle Y_ {n}: = \ sum _ {i = 1} ^ {n} X_ {i}}
exactly the symmetric random walk on . So is
Z
{\ displaystyle \ mathbb {Z}}
0
,
5
(
n
+
∑
i
=
1
n
X
i
)
∼
B.
i
n
(
n
;
0
,
5
)
{\ displaystyle 0 {,} 5 \ left (n + \ sum _ {i = 1} ^ {n} X_ {i} \ right) \ sim Bin (n; 0 {,} 5)}
so binomially distributed .
Relationship to the Laplace distribution
If rademacher is distributed and is exponentially distributed to the parameter , then laplace is distributed to the position parameter 0 and the scale parameter .
X
{\ displaystyle X}
Y
{\ displaystyle Y}
λ
{\ displaystyle \ lambda}
X
⋅
Y
{\ displaystyle X \ cdot Y}
1
λ
{\ displaystyle {\ frac {1} {\ lambda}}}
Discrete univariate distributions
Continuous univariate distributions
Multivariate distributions
<img src="https://de.wikipedia.org/wiki/Special:CentralAutoLogin/start?type=1x1" alt="" title="" width="1" height="1" style="border: none; position: absolute;">