logsumexpc: Logsumexp technique to prevent underflow

In DavidKLim/FSCseq: Feature Selection and Clustering of RNA-seq Count Data

Description

This function takes in a vector, and uses the logsumexp technique to prevent underflow issues. Used on log_pi + loglike to prevent it being rounded to 0

Usage

logsumexpc(v)

Arguments

v	a vector or matrix of numbers

Value

log(sum(exp(v))), without underflow issues

Examples

log_pi=log(c(0.1,0.02,0.07))
l=matrix(c(-2000,-2100,-2050,-1000,-1050,-1100),nrow=3,ncol=2)
logsumexpc(log_pi+l)
log(sum(exp(log_pi+l)))

DavidKLim/FSCseq documentation built on Dec. 12, 2021, 3:46 a.m.