StochGradient: Functional quantization with stochastic gradient method

In elamrireda/FunctQuant: Data-Driven Functional Quantization

Description

Data-driven functional quantization with stochastic gradient method

Usage

StochGradient(data,mKL,size)

Arguments

data	matrix that we want to quantize.
mKL	truncation argument for the dimension reduction.
size	size of the quantization grids.

Value

data	the input matrix.
quantizer	the quantizer grid.
weights	the associated weight of each curve (calculated using the input matrix).

See Also

CVT and GFQ

Examples

##### function to generate realizations of BM
BM <- function(N=1000,M=1,x0=0,t0=0,T=1,Dt=NULL)
{
  Dt <- (T - t0)/N
  t <- seq(t0, T, by=Dt)
  res <- data.frame(sapply(1:M,function(i) c(0,cumsum(rnorm(N,mean  =0,sd=sqrt(Dt))))))
  names(res) <- paste("X",1:M,sep="")
  X <- ts(res, start = t0, deltat = Dt)
  return(X)
}
data <- t(BM(N = 200 - 1, M = 200))
size <- 10
mKL <- 2
quant <- StochGradient(data,mKL,size)

elamrireda/FunctQuant documentation built on Jan. 1, 2021, 2:50 p.m.