View source for Gamma

{{Probability distribution |
  name       =Gamma|
  type       =density|
  pdf_image  =[[Image:Gamma distribution pdf.png|325px|Probability density plots of gamma distributions]]|
  cdf_image  =[[Image:Gamma distribution cdf.png|325px|Cumulative distribution plots of gamma distributions]]|
  parameters =<math>k > 0\,</math> [[shape parameter|shape]] <br /><math>\theta > 0\,</math> [[scale parameter|scale]] |
  support    =<math>x \in [0, \infty)\!</math>|
  pdf        =<math>x^{k-1} \frac{\exp{\left(-x/\theta\right)}}{\Gamma(k)\,\theta^k}\,\!</math>|
  cdf        =<math>\frac{\gamma(k, x/\theta)}{\Gamma(k)} \!</math>|
  mean       =<math>k\theta \!</math>|
  median     =no simple closed form|
  mode       =<math>(k-1) \theta\text{ for }k \geq 1\,\!</math> |
  variance   =<math>k \theta^2\,\!</math>|
  skewness   =<math>\frac{2}{\sqrt{k}}\,\!</math>|
  kurtosis   =<math>\frac{6}{k}\,\!</math>|
  entropy    =<math>k + \ln\theta + \ln\Gamma(k) \!</math><br /><math>+ (1-k)\psi(k) \!</math>|
  mgf        =<math>(1 - \theta\,t)^{-k}\text{ for }t < 1/\theta\,\!</math>|
  char       =<math>(1 - \theta\,i\,t)^{-k}\,\!</math>|
}}
In [[probability theory]] and [[statistics]], the '''gamma distribution'''  is a two-parameter family of continuous [[probability distribution]]s. It has a [[scale parameter]] ''θ'' and a [[shape parameter]] ''k''.  If ''k'' is an [[integer]] then the distribution represents the sum of ''k'' independent [[exponential distribution|exponentially distributed]] random variables, each of which has a mean of ''θ''  (which is equivalent to a rate parameter of ''θ''<sup>&nbsp;&minus;1</sup>) .

The gamma distribution is frequently a probability model for waiting times; for instance, in life testing, the waiting time until death is a random variable that is frequently modeled with a gamma distribution.<ref>See Hogg and Craig Remark 3.3.1. for an explicit motivation.test</ref>  Gamma distributions were fitted to rainfall amounts from different storms, and differences in amounts from seeded and unseeded storms were reflected in differences in estimated ''k'' and <math>\theta</math> parameters <ref>{{citation|last=Rice|first=John|title=Mathematical Statistics and Data Analysis|edition=Second|publisher=Duxbury Press|year=1995|page=244|isbn=0-534-20934-3}}</ref>

==Characterization==
A random variable ''X'' that is gamma-distributed with scale ''θ'' and shape ''k'' is denoted

:<math>X \sim \Gamma(k, \theta)\text{ or }X \sim \textrm{Gamma}(k,\theta ). \, </math>

===Probability density function===
The [[probability density function]] of the gamma distribution can be expressed in terms of the [[gamma function]] parameterized in terms of a shape parameter ''k'' and scale parameter ''θ''.  Both ''k'' and ''θ'' will be positive values.

The equation defining the probability density function of a gamma-distributed random variable ''x'' is

:<math> f(x;k,\theta) = x^{k-1} \frac{e^{-x/\theta}}{\theta^k \, \Gamma(k)}\text{ for } x > 0\text{ and }k, \theta > 0.\,</math>

(This parameterization is used in the infobox and the plots.)

Alternatively, the gamma distribution can be parameterized in terms of a shape parameter ''α''&nbsp;=&nbsp;''k'' and an inverse scale parameter ''β''&nbsp;=&nbsp;1/''θ'', called a rate parameter:

:<math>g(x;\alpha,\beta) = \frac{\beta^{\alpha}}{\Gamma(\alpha)} x^{\alpha-1} e^{-\beta {x}} \text{ for }x > 0. \,</math>

If ''α'' is a positive integer, then

: <math> \Gamma(\alpha) = (\alpha - 1)!.\, </math>

Both parametrizations are common because either can be more convenient depending on the situation.
[[Image:Gamma-PDF-3D.png|thumb|right|320px|Illustration of the Gamma PDF for parameter values over ''k'' and ''x'' with ''&theta;'' set to 1,&nbsp;2,&nbsp;3,&nbsp;4,&nbsp;5&nbsp;and&nbsp;6. One can see each ''&theta;'' layer by itself here [http://commons.wikimedia.org/wiki/File:Gamma-PDF-3D-by-k.png] as well as by&nbsp;''k'' [http://commons.wikimedia.org/wiki/File:Gamma-PDF-3D-by-Theta.png] and&nbsp;''x''. [http://commons.wikimedia.org/wiki/File:Gamma-PDF-3D-by-x.png].]]

===Cumulative distribution function===

The [[cumulative distribution function]] is the regularized gamma function: 

:<math> F(x;k,\theta) = \int_0^x f(u;k,\theta)\,du  
  =\frac{\gamma(k, x/\theta)}{\Gamma(k)} \,</math>

where <math>\gamma(k, x/\theta)</math> is the lower [[incomplete gamma function]].

It can also be expressed as follows, if ''k'' is a positive [[integer]] (i.e., the distribution is an [[Erlang distribution]])<ref>Papoulis, Pillai, ''Probability, Random Variables, and Stochastic Processes'', Fourth Edition</ref>:

:<math>F(x;k,\theta) = 1-\sum_{i=0}^{k-1} \frac{(x/\theta)^i}{i!} e^{-x/\theta}.</math>

<!-- The parameter ''&theta;'' used here is equivalent to the ''&beta;'' defined above! -->

==Properties==
===Summation===

If ''X''<sub>''i''</sub> has a Γ(''k''<sub>''i''</sub>, θ) distribution for ''i''&nbsp;=&nbsp;1,&nbsp;2,&nbsp;...,&nbsp;''N'', then

:<math>
\sum_{i=1}^N X_i
\sim
\mathrm{Gamma}  \left( \sum_{i=1}^N k_i, \theta \right) \,\!
</math>

provided all ''X''<sub>''i'</sub> are [[statistical independence|independent]].

The gamma distribution exhibits [[Infinite divisibility (probability)|infinite divisibility]].

===Scaling===

If

<math>X\,\sim\,\mathrm{Gamma}(k,\theta)</math>

then for any ''&alpha;''&nbsp;>&nbsp;0,

<math>\alpha X\,\sim\,\mathrm{Gamma}(k,\alpha\theta).</math>

===Exponential family===

The Gamma distribution is a two-parameter [[exponential family]] with [[natural parameters]] ''k''&nbsp;&minus;&nbsp;1 and &minus;1/''θ'', and [[natural statistics]] ''X'' and ln&nbsp;(''X'').

===Information entropy===
The [[information entropy]] is given by

: <math>
\begin{align}
& {}\qquad\frac{-1}{\theta^k \Gamma(k)} \int_0^\infty \frac{x^{k-1}}{e^{x/\theta}} \left[ (k-1)\ln x - x/\theta - k \ln\theta - \ln\Gamma(k) \right] \,dx \\[8pt]
& = -\left[ (k-1) (\ln\theta + \psi(k)) - k - k \ln\theta - \ln\Gamma(k) \right] \\[8pt]
& = k + \ln\theta + \ln\Gamma(k) + (1-k)\psi(k)
\end{align}
</math>

where ''ψ''(''k'') is the [[digamma function]].

One can also show that (if we use the shape parameter ''k'' and the inverse scale parameter&nbsp;''&beta;''),

: <math>\mathbb{E}[\ln(x)] = \psi(k) - \ln(\beta). \, </math>

Or alternately, using the scale parameter&nbsp;''&theta;'',

: <math>\mathbb{E}[\ln(x)] = \psi(k) + \ln(\theta). \, </math>

[[Image:Gamma-KL-3D.png|thumb|right|320px|Illustration of the Kullback–Leibler (KL) divergence for two Gamma PDF's. Here ''&beta;''&nbsp;=&nbsp;''&beta;''<sub>0</sub>&nbsp;+&nbsp;1 which are set to 1,&nbsp;2,&nbsp;3,&nbsp;4,&nbsp;5&nbsp;and&nbsp;6.  The typical asymmetry for the KL divergence is clearly visible.]]

===Kullback–Leibler divergence===
The directed [[Kullback–Leibler divergence]] between Γ(θ<sub>0</sub>, β<sub>0</sub>) ('true' distribution) and Γ(θ, β) ('approximating' distribution), for shape parameter θ and inverse scale parameter β is given by

:<math>
D_{\mathrm{KL}}(\theta_0,\beta_0 || \theta, \beta) = \log\left(\frac{\Gamma({\theta})\beta_0^{\theta_0}}{\Gamma(\theta_0)\beta^{\theta}}\right)+(\theta_0-\theta)(\psi(\theta_0)-\log \beta_0)+\theta_0\frac{\beta-\beta_0}{\beta_0}
</math>

=== Laplace transform ===
The [[Laplace transform]] of the gamma PDF is

:<math>
F(s)=\left(1+\theta s\right)^{-k}=\frac{\beta^\alpha}{(s+\beta)^\alpha}.</math>

== Parameter estimation ==
=== Maximum likelihood estimation ===

The likelihood function for ''N'' [[independent and identically-distributed random variables|iid]] observations (''x''<sub>1</sub>,&nbsp;...,&nbsp;''x''<sub>''N''</sub>) is

:<math>L(k,\theta)=\prod_{i=1}^N f(x_i;k,\theta)\,\!</math>

from which we calculate the log-likelihood function

:<math>\ell(k,\theta) = (k-1) \sum_{i=1}^N \ln{(x_i)} - \sum_{i=1}^N x_i/\theta - Nk\ln{(\theta)} - N\ln{\Gamma(k)}.</math>

Finding the maximum with respect to ''θ'' by taking the derivative and setting it equal to zero yields the [[maximum likelihood]] estimator of the θ parameter:

:<math>\hat{\theta} = \frac{1}{kN}\sum_{i=1}^N x_i. \,\!</math>

Substituting this into the log-likelihood function gives

:<math>\ell=(k-1)\sum_{i=1}^N\ln{(x_i)}-Nk-Nk\ln{\left(\frac{\sum x_i}{kN}\right)}-N\ln{(\Gamma(k))}. \,\!</math>

Finding the maximum with respect to ''k'' by taking the derivative and setting it equal to zero yields

:<math>\ln{(k)}-\psi(k)=\ln{\left(\frac{1}{N}\sum_{i=1}^N x_i\right)}-\frac{1}{N}\sum_{i=1}^N\ln{(x_i)} \,\!</math>

where

: <math>\psi(k) = \frac{\Gamma'(k)}{\Gamma(k)} \!</math>

is the [[digamma function]].

There is no closed-form solution for ''k''.  The function is numerically very well behaved, so if a numerical solution is desired, it can be found using, for example, [[Newton's method]].  An initial value of ''k'' can be found either using the [[method of moments (statistics)|method of moments]], or using the approximation

:<math>\ln(k)-\psi(k) \approx \frac{1}{k}\left(\frac{1}{2} + \frac{1}{12k+2}\right). \,\!</math>

If we let 

:<math>s = \ln{\left(\frac{1}{N}\sum_{i=1}^N x_i\right)} - \frac{1}{N}\sum_{i=1}^N\ln{(x_i)},\,\!</math> 

then ''k'' is approximately

:<math>k \approx \frac{3-s+\sqrt{(s-3)^2 + 24s}}{12s}</math>

which is within 1.5% of the correct value.{{Fact|date=February 2007}} An explicit form for the Newton-Raphson update of this initial guess is given by Choi and Wette (1969) as the following expression:

:<math>k \leftarrow k - \frac{ \ln k - \psi\left(k\right) - s }{ 1/k - \psi\;'\left(k\right) }</math>

where <math>\psi\;'\left(\cdot\right)</math> denotes the trigamma function (the derivative of the digamma function).

The digamma and trigamma functions can be difficult to calculate with high precision. However, approximations known to be good to several significant figures can be computed using the following '''approximation formulae''':

:<math>
\psi\left(k\right) = \begin{cases}
\ln(k) - ( 1 + ( 1 - ( 1/10 - 1 / ( 21 k^2 ) ) / k^2 ) / ( 6 k ) ) / ( 2 k ), \quad k \geq 8 \\
\psi\left( k + 1 \right) - 1/k, \quad k < 8
\end{cases}
</math>

and

:<math>
\psi\;'\left(k\right) = \begin{cases}
( 1 + ( 1 + ( 1 - ( 1/5 - 1 / ( 7 k^2 ) ) / k^2 ) / ( 3 k ) ) / ( 2 k ) ) / k, \quad k \geq 8, \\
\psi\;'\left( k + 1 \right) + 1/k^2, \quad k < 8.
\end{cases}
</math>

For details, see Choi and Wette (1969).

=== Bayesian minimum mean-squared error ===

With known ''k'' and unknown <math> \theta </math>, the posterior PDF for theta (using the standard scale-invariant prior for <math>\theta</math>) is

:<math>
P(\theta | k, x_1, ..., x_N) \propto 1/\theta \prod_{i=1}^N f(x_i;k,\theta).\,\!
</math>

Denoting

:<math> y \equiv \sum_{i=1}^N x_i , \qquad  P(\theta | k, x_1, \dots , x_N) = C(x_i)  \theta^{-N k-1} e^{-y / \theta}. \! </math>

Integration over θ can be carried out using a change of variables, revealing that 1/θ is gamma-distributed with parameters <math>\scriptstyle \alpha = N k,\ \  \beta = y</math>.

:<math>
\int_0^{\infty} \theta^{-N k-1+m} e^{-y / \theta}\, d\theta = \int_0^{\infty} x^{N k -1 -m} e^{-x y} \, dx = y^{-(N k -m)} \Gamma(N k -m). \!
</math>

The moments can be computed by taking the ratio (''m'' by ''m'' = 0)

:<math>
E(x^m) = \frac {\Gamma (N k -m)} {\Gamma(N k)} y^m, \!
</math>

which shows that the mean ± standard deviation estimate of the posterior distribution for theta is

:<math> \frac {y} {N k -1} \pm \frac {y^2} {(N k-1)^2 (N k-2)}. </math>

== Generating gamma-distributed random variables ==

Given the scaling property above, it is enough to generate gamma variables with <math>\theta = 1</math> as we can later convert to any value of <math>\beta</math> with simple division.

Using the fact that a <math>\Gamma(1,1)</math> distribution is the same as an <math>Exp(1)</math> distribution, and noting the method of [[exponential distribution#Generating exponential variates|generating exponential variables]],
we conclude that if <math>U</math> is [[uniform distribution (continuous)|uniformly distributed]] on <math>(0,1]</math>,
then &minus;<math>\ln(U)</math> is distributed <math>\Gamma(1,1)</math>.
Now, using the "&alpha;-addition" property of gamma distribution, we expand this result:

:<math>\sum_{k=1}^n {-\ln U_k} \sim \Gamma(n, 1),</math>

where <math>U_k</math> are all uniformly distributed on <math>(0,1]</math> and [[statistical independence|independent]].

All that is left now is to generate a variable distributed as <math>\Gamma(\delta,1)</math> for <math>0 < \delta < 1</math> and apply the "&alpha;-addition" property once more. This is the most difficult part.

We provide an algorithm without proof.  It is an instance of the [[rejection sampling|acceptance-rejection method]]:

# Let <math>m</math> be 1.
# Generate <math>V_{3m - 2}</math>, <math>V_{3m - 1}</math> and <math>V_{3m}</math> as independent uniformly distributed on <math>(0,1]</math> variables.
# If <math>V_{3m - 2} \le v_0</math>, where <math>v_0 = \frac e {e + \delta}</math>, then go to step 4, else go to step 5.
# Let <math>\xi_m = V_{3m - 1}^{1 / \delta}, \ \eta_m = V_{3m} \xi _m^ {\delta - 1}</math>.  Go to step 6.
# Let <math>\xi_m = 1 - \ln {V_{3m - 1}}, \ \eta_m = V_{3m} e^{-\xi_m}</math>.
# If <math>\eta_m > \xi_m^{\delta - 1} e^{-\xi_m}</math>, then increment <math>m</math> and go to step 2.
# Assume <math>\xi = \xi_m</math> to be the realization of <math>\Gamma  (\delta, 1)</math>
Now, to summarize,

:<math> \theta \left( \xi - \sum _{i=1} ^{\lfloor{k}\rfloor} {\ln U_i} \right) \sim \Gamma (k, \theta),</math>
where
<math>\lfloor{k}\rfloor</math> is the integral part of <math>k</math>, and <math>\xi</math> has been generated using the algorithm above with
<math>\delta = \{k\}</math> (the fractional part of <math>k</math>),
<math>U_k</math> and <math>V_l</math> are distributed as explained above and are all independent.

==Related distributions==
=== Specializations ===
* If <math>X \sim {\Gamma}(k=1, \theta=1/\lambda)\,</math>, then ''X'' has an [[exponential distribution]] with rate parameter λ. 
* If <math>X \sim {\Gamma}(k=\nu/2, \theta=2)\,</math>, then ''X'' is identical to χ<sup>2</sup>(''ν''), the [[chi-square distribution]] with ''ν'' degrees of freedom. Conversely, if <math>Q \sim {\chi}^2(\nu)\,</math> and ''c'' is a positive constant, then <math>c\cdot Q \sim {\Gamma}(k=\nu/2, \theta=2c)\,</math>.
* If <math>k</math> is an [[integer]], the gamma distribution is an [[Erlang distribution]] and is the probability distribution of the waiting time until the <math>k</math>-th "arrival" in a one-dimensional [[Poisson process]] with intensity 1/θ.
* If <math>X^2 \sim {\Gamma}(3/2, 2a^2)\,</math>, then ''X'' has a [[Maxwell-Boltzmann distribution]] with parameter ''a''.
<!--
* <math>Y \sim N(\mu = \alpha \beta, \sigma^2 = \alpha \beta^2)</math> is a [[normal distribution]] as <math>Y = \lim_{\alpha \to \infty} X</math> where <math>X \sim \Gamma (\alpha, \beta)</math>. --> 
*<math>X \sim \mathrm{SkewLogistic}(\theta)\,</math>, then <math>\mathrm{log}(1 + e^{-X}) \sim \Gamma (1,\theta)\,</math>, i.e. an [[exponential distribution]]: see [[skew-logistic distribution]].

=== Conjugate prior ===
In [[Bayesian inference]], the '''gamma distribution''' is the [[conjugate prior]] to many likelihood distributions: the [[Poisson distribution|Poisson]], [[Exponential distribution|exponential]], [[Normal distribution|normal]] (with known mean), [[Pareto distribution|Pareto]], gamma with known shape σ, and [[Inverse-gamma distribution|inverse gamma]] with known shape parameter. <!-- reference: see article [[conjugate prior]] //-->

The Gamma distribution's [[conjugate prior]] is <ref name="fink">Fink, D. 1995 [http://www.stat.columbia.edu/~cook/movabletype/mlm/CONJINTRnew%2BTEX.pdf A Compendium of Conjugate Priors]. In progress report: Extension and enhancement of methods for setting data quality objectives. (DOE contract 95‑831).</ref>:

:<math>p(k,\theta | p,q,r,s) = \frac{1}{Z} \frac{p^{k-1} e^{-\theta^{-1} q}}{\Gamma(k)^r \theta^{k s}}</math>

Where Z is the normalizing constant, which has no closed form solution. The posterior distribution can be found by updating the parameters as follows.

<math>p' = p\prod_i x_i</math>
<math>q' = q + \sum_i x_i</math>
<math>r' = r + n</math>,<math>s' = s + n</math>

Where <math>n</math> is the number of observations, and <math>x_i</math> is the <math>i^{th}</math> observation.

=== Others ===
* If ''X'' has a Γ(''k'', θ) distribution, then 1/''X'' has an [[inverse-gamma distribution]] with parameters ''k'' and θ<sup>-1</sup>. 
* If ''X'' and ''Y'' are independently distributed Γ(α, θ) and  Γ(β, θ) respectively, then ''X''&nbsp;/&nbsp;(''X''&nbsp;+&nbsp;''Y'') has a [[beta distribution]] with parameters α and β.
* If ''X<sub>i</sub>'' are independently distributed Γ(α<sub>''i''</sub>,θ) respectively, then the vector (''X''<sub>1</sub>&nbsp;/&nbsp;''S'',&nbsp;...,&nbsp;''X<sub>n</sub>''&nbsp;/&nbsp;''S''), where ''S''&nbsp;=&nbsp;''X''<sub>1</sub>&nbsp;+&nbsp;...&nbsp;+&nbsp;''X<sub>n</sub>'', follows a [[Dirichlet distribution]] with parameters α<sub>1</sub>,&nbsp;...,&nbsp;α<sub>''n''</sub>.
* For large ''k'' the gamma distribution converges to Gaussian distribution with mean <math>\mu = k\theta</math> and variance <math>\sigma^2 = k\theta^2</math>.
* The Gamma distribution is the [[conjugate prior]] for the precision of the [[normal distribution]] with known [[mean]].
* The [[Wishart distribution]] is a multivariate generalization of the gamma distribution (samples are positive-definite matrices rather than positive real numbers).
* The Gamma distribution is a special case of the [[generalized gamma distribution]].
* Among the discrete distributions, the [[negative binomial distribution]] is sometimes considered the discrete analogue of the Gamma distribution

==Applications==
{{Expand-section|date=March 2009}}
The gamma distribution has been used to model the size of insurance claims and rainfalls. This means aggregate insurance claims and the amount of rainfall accumulated in a reservoir are modelled by a [[gamma process]].

== See also ==
{{wikibooks|Statistics|Distributions/Gamma|Gamma distribution}}
* [[Gamma process]]
* [[Lukacs's proportion-sum independence theorem]]

==Notes==
<references/>

== References ==
*  R. V. Hogg and A. T. Craig. ''Introduction to Mathematical Statistics'', 4th edition. New York: Macmillan, 1978. ''(See Section 3.3.)''
*  {{MathWorld|urlname=GammaDistribution|title=Gamma distribution}}
* [http://www.itl.nist.gov/div898/handbook/eda/section3/eda366b.htm Engineering Statistics Handbook]
* S. C. Choi and R. Wette. (1969) ''Maximum Likelihood Estimation of the Parameters of the Gamma Distribution and Their Bias'', '''Technometrics''', '''11'''(4) 683&ndash;690

{{ProbDistributions|continuous-semi-infinite}}
{{Common univariate probability distributions}}

[[Category:Continuous distributions]]
[[Category:Factorial and binomial topics]]

[[az:Qamma paylanması]]
[[ca:Distribució gamma]]
[[de:Gammaverteilung]]
[[es:Distribución gamma]]
[[fa:توزیع گاما]]
[[fr:Distribution Gamma]]
[[ko:감마 분포]]
[[it:Distribuzione Gamma]]
[[he:התפלגות גמא]]
[[hu:Gamma-eloszlás]]
[[nl:Gamma-verdeling]]
[[ja:ガンマ分布]]
[[pl:Rozkład gamma]]
[[pt:Distribuição gama]]
[[ru:Гамма-распределение]]
[[sl:Porazdelitev gama]]
[[fi:Gamma-jakauma]]
[[sv:Gammafördelning]]
[[tr:Gamma dağılımı]]
[[uk:Гамма-розподіл]]
[[zh:伽玛分布]]