[section:nc_f_dist Noncentral F Distribution]

``#include <boost/math/distributions/non_central_f.hpp>``

   namespace boost{ namespace math{ 

   template <class RealType = double, 
             class ``__Policy``   = ``__policy_class`` >
   class non_central_f_distribution;

   typedef non_central_f_distribution<> non_central_f;

   template <class RealType, class ``__Policy``>
   class non_central_f_distribution
   {
   public:
      typedef RealType  value_type;
      typedef Policy    policy_type;

      // Constructor:
      non_central_f_distribution(RealType v1, RealType v2, RealType lambda);

      // Accessor to degrees_of_freedom parameters v1 & v2:
      RealType degrees_of_freedom1()const;
      RealType degrees_of_freedom2()const;

      // Accessor to non-centrality parameter lambda:
      RealType non_centrality()const;
   };
   
   }} // namespaces
   
The noncentral F distribution is a generalization of the __F_distrib.
It is defined as the ratio 

[expression F = (X/v1) / (Y/v2)]
   
where X is a noncentral [chi][super 2]
random variable with /v1/ degrees of freedom and non-centrality parameter [lambda], 
and Y is a central [chi][super 2] random variable with /v2/ degrees of freedom.

This gives the following PDF:

[equation nc_f_ref1]

where ['L[sub a][super b](c)] is a generalised Laguerre polynomial and ['B(a,b)] is the 
__beta function, or

[equation nc_f_ref2]

The following graph illustrates how the distribution changes
for different values of [lambda]:

[graph nc_f_pdf]

[h4 Member Functions]

      non_central_f_distribution(RealType v1, RealType v2, RealType lambda);
      
Constructs a non-central beta distribution with parameters /v1/ and /v2/
and non-centrality parameter /lambda/.

Requires /v1/ > 0, /v2/ > 0 and lambda >= 0, otherwise calls __domain_error.

      RealType degrees_of_freedom1()const;
      
Returns the parameter /v1/ from which this object was constructed.

      RealType degrees_of_freedom2()const;
      
Returns the parameter /v2/ from which this object was constructed.

      RealType non_centrality()const;
      
Returns the non-centrality parameter /lambda/ from which this object was constructed.

[h4 Non-member Accessors]

All the [link math_toolkit.dist_ref.nmp usual non-member accessor functions]
that are generic to all distributions are supported: __usual_accessors.

The domain of the random variable is \[0, +[infin]\].

[h4 Accuracy]

This distribution is implemented in terms of the
__non_central_beta_distrib: refer to that distribution for accuracy data.

[h4 Tests]

Since this distribution is implemented by adapting another distribution, 
the tests consist of basic sanity checks computed by the
[@http://www.r-project.org/ R-2.5.1 Math library statistical
package] and its pbeta and dbeta functions.

[h4 Implementation]

In the following table /v1/ and /v2/ are the first and second
degrees of freedom parameters of the distribution, [lambda]
is the non-centrality parameter,
/x/ is the random variate, /p/ is the probability, and /q = 1-p/.

[table
[[Function][Implementation Notes]]
[[pdf][Implemented in terms of the non-central beta PDF using the relation:

[role serif_italic f(x;v1,v2;[lambda]) = (v1\/v2) / ((1+y)*(1+y)) * g(y\/(1+y);v1\/2,v2\/2;[lambda])]

where [role serif_italic g(x; a, b; [lambda])] is the non central beta PDF, and:
 
[role serif_italic y = x * v1 \/ v2]
]]
[[cdf][Using the relation:

[role serif_italic p = B[sub y](v1\/2, v2\/2; [lambda])]

where [role serif_italic B[sub x](a, b; [lambda])] is the noncentral beta distribution CDF and

[role serif_italic y = x * v1 \/ v2]

]]

[[cdf complement][Using the relation:

[role serif_italic q = 1 - B[sub y](v1\/2, v2\/2; [lambda])]

where [role serif_italic 1 - B[sub x](a, b; [lambda])] is the complement of the 
noncentral beta distribution CDF and

[role serif_italic y = x * v1 \/ v2]

]]
[[quantile][Using the relation:

[role serif_italic x = (bx \/ (1-bx)) * (v1 \/ v2)]

where

[role serif_italic bx = Q[sub p][super -1](v1\/2, v2\/2; [lambda])]

and 

[role serif_italic Q[sub p][super -1](v1\/2, v2\/2; [lambda])]

is the noncentral beta quantile.

]]
[[quantile

from the complement][
Using the relation:

[role serif_italic x = (bx \/ (1-bx)) * (v1 \/ v2)]

where

[role serif_italic bx = QC[sub q][super -1](v1\/2, v2\/2; [lambda])]

and 

[role serif_italic QC[sub q][super -1](v1\/2, v2\/2; [lambda])]

is the noncentral beta quantile from the complement.]]
[[mean][[role serif_italic v2 * (v1 + l) \/ (v1 * (v2 - 2))]]]
[[mode][By numeric maximalisation of the PDF.]]
[[variance][Refer to, [@http://mathworld.wolfram.com/NoncentralF-Distribution.html
    Weisstein, Eric W. "Noncentral F-Distribution." From MathWorld--A Wolfram Web Resource.]  ]]
[[skewness][Refer to, [@http://mathworld.wolfram.com/NoncentralF-Distribution.html
    Weisstein, Eric W. "Noncentral F-Distribution." From MathWorld--A Wolfram Web Resource.],
    and to the [@http://reference.wolfram.com/mathematica/ref/NoncentralFRatioDistribution.html
    Mathematica documentation]  ]]
[[kurtosis and kurtosis excess]
    [Refer to, [@http://mathworld.wolfram.com/NoncentralF-Distribution.html
    Weisstein, Eric W. "Noncentral F-Distribution." From MathWorld--A Wolfram Web Resource.],
    and to the [@http://reference.wolfram.com/mathematica/ref/NoncentralFRatioDistribution.html
    Mathematica documentation]  ]]
]

Some analytic properties of noncentral distributions
(particularly unimodality, and monotonicity of their modes)
are surveyed and summarized by:

Andrea van Aubel & Wolfgang Gawronski, Applied Mathematics and Computation, 141 (2003) 3-12.

[endsect] [/section:nc_f_dist]

[/ nc_f.qbk
  Copyright 2008 John Maddock and Paul A. Bristow.
  Distributed under the Boost Software License, Version 1.0.
  (See accompanying file LICENSE_1_0.txt or copy at
  http://www.boost.org/LICENSE_1_0.txt).
]