Kiefer Wolfowitz NPMLE for Student t non-centrality parameter mixtures
Model: *y_{ig} = mu_{g} + e_{ig}, e_{ig} ~ N(0,sigma_{g}^{2})*
x is the vector of t statistics for all groups, which follows t dist
if *mu_g = 0*, and noncentral t dist if *mu_g \neq 0*,
with *ncp_{g} = μ_g / σ_{g}*.
This leads to a mixture of t distribution with ncp as the mixing parameter.
df (degree of freedom) is determined by the group size in the simplest case.

1 2 |

`x` |
Data: Sample Observations |

`v` |
bin boundaries defaults to equal spacing of length v |

`u` |
bin boundaries for histogram binning: defaults to equal spacing |

`df` |
Number of degrees of freedom of Student base density |

`hist` |
If TRUE then aggregate x to histogram weights |

`weights` |
replicate weights for x obervations, should sum to 1 |

`...` |
optional parameters passed to KWDual to control optimization |

An object of class density with components:

`x` |
midpoints of evaluation on the domain of the mixing density |

`y` |
estimated function values at the points x of the mixing density |

`g` |
estimated function values at the observed points of mixture density |

`logLik` |
Log likelihood value at the proposed solution |

`dy` |
Bayes rule estimates of location at x |

`status` |
Mosek exit code |

Roger Koenker

Kiefer, J. and J. Wolfowitz Consistency of the Maximum
Likelihood Estimator in the Presence of Infinitely Many Incidental
Parameters *Ann. Math. Statist*. 27, (1956), 887-906.

GLmix for Gaussian version

Questions? Problems? Suggestions? Tweet to @rdrrHQ or email at ian@mutexlabs.com.

All documentation is copyright its authors; we didn't write any of that.