f_md

PURPOSE

Penalised density of optimal hyperplane with normal vector (V)

SYNOPSIS

function [fmin, bmin, df, dkde] = f_md(v,X,pars)

DESCRIPTION

Penalised density of optimal hyperplane with normal vector (V)
[FVAL, SPLIT, DF, DKDE] = F_MD(V,X,PARS)

 Inputs:
    (V): Projection vector
    (X): Data matrix
    (PARS): Structure array containing MDH parameters (alpha, eta, epsilon,h)

 Output: 
    (FVAL): penalised density integral for optimal hyperplane orthogonal to (v)
    (SPLIT): optimal split point on (v) after (v) is scaled to unit-length
    (DF): derivative of penalised density integral at (split)
    (DKDE): derivative of 1D kernel density estimator at (split)

CROSS-REFERENCE INFORMATION

This function calls:

dpkde Penalised density on hyperplane: Projection index for MDH
pkde Penalised density on of one-dimensional (projected) dataset at (x)

This function is called by:

df_md Derivative of penalised density on hyperplane criterion
f_df_md Function value and derivative for penalised density
mdhp MDHP implements Minimum Density Hyperplane (inherits from HP class)
mdpp Minimum Density Projection Pursuit (MDPP) algorithm