Row-wise Gaussian affinities with sigma solved to hit a target perplexity

Binary search on beta = 1/(2 sigma^2) per row until row entropy equals log(perplexity).

Usage

morie_geron_conditional_p(D2, perplexity, tol = 1e-05, max_steps = 100)

Arguments

D2

Squared-distance matrix (n, n).

perplexity

Target perplexity.

tol

Entropy tolerance.

max_steps

Binary-search iterations.

Value

list(P, betas).

Examples

morie_geron_conditional_p(D2 = matrix(c(1, 2, 3, 4, 5, 6), nrow = 2), perplexity = 5L)
#> $P
#>      [,1] [,2]
#> [1,]    0    1
#> [2,]    1    0
#> 
#> $betas
#> [1] 7.888609e-31 7.888609e-31
#>