norm_phi_canonical_kmeans: Represent <x'(phi)_i, x'(phi)_j> as a quadratic function in...

In yiqunchen/KmeansInference: Selective inference for k-means clustering

Represent <x'(phi)_i, x'(phi)_j> as a quadratic function in phi —-

Description

Represent <x'(phi)_i, x'(phi)_j> as a quadratic function in phi —-

Usage

norm_phi_canonical_kmeans(
  X,
  last_centroids,
  XTv,
  XTv_norm,
  dir_XTv,
  v_norm,
  cl,
  k,
  v,
  i
)

Arguments

XTv,	vector p by 1
XTv_norm,	norm of XTv
dir_XTv,	vector p by 1 := XTv/XTv_norm
v_norm,	2-norm of vector v
cl,	factor vector n by 1 (most recent cluster assignment)
k,	cluster of interest
v,	contrast vector n by 1
i,	index of observation

Value

parameters: a, b, c the coefficients of the quadratic equation such that (ax^2 + bx + c <= 0)

yiqunchen/KmeansInference documentation built on June 4, 2022, 1:20 p.m.