Downsample code based on the neighbors

downsample_by_neighbors.R

mat <- matrix(rnorm(100000), ncol=10)

library(BiocNeighbors)
res <- findKNN(mat, k=20)
d <- res$distance[,ncol(res$distance)]

processed <- logical(nrow(mat))
candidates <- logical(nrow(mat))

while (!all(processed)) {
    keep <- which(!processed)
    coverage <- vapply(keep, function(x) sum(processed[res$index[x,]]), 0)
    o <- order(coverage, d[keep])
    limit <- min(coverage)
    print(limit)

    for (i in o) {
        pt <- keep[i]
        if (processed[pt]) {
            next
        } else if (coverage[i] > limit) {
            break
        }
        
        neighbors <- res$index[pt,]
        updated.coverage <- sum(processed[neighbors])
        if (updated.coverage == limit) {
            candidates[pt] <- TRUE
            processed[pt] <- TRUE
            processed[neighbors] <- TRUE
        }
    }
}
mean(candidates)