montecarlo - Cannot figure out Reduce and Monte Carlo Simulations in R, calculating VaR -

apologies title. not think of how title this...

(also, know shite question, hoping out there can help.)

i have following mean vector , covariance matrix:

> mu0        msft        aapl  0.001250251 0.001060690  > sig0              msft         aapl msft 1.275625e-04 3.334225e-05 aapl 3.334225e-05 1.484212e-04 

i have following chunk of code (i know 500 simulations. brevity's sake. ramp when figure out code.):

wta = 0.562546911; wtm = 0.437453089; ct = -1000000*c(wtm,wta); lambda = .97; k = 20;  m = 500; l1 = 0.0; l1 = rep(0,m);  mu1 = mu0; sig1 = sig0; l1=0.0; rando = rmvnorm(m, mu1, sig1) for(i in 1:m){    xtd = t(rando[i,])   mu1 = lambda*mu1+(1-lambda)*xtd   sig1 = lambda*sig0+(1-lambda)*t(xtd)%*%xtd; l1 = l1+t(ct)%*%t(xtd)   l1=l1+t(ct)%*%t(xtd)   for(j in 2:k){      xtd=rmvnorm(1,mu1, sig1)     mu1=lambda*mu1+(1-lambda)*xtd     sig1=lambda*sig0+(1-lambda)*t(xtd)%*%xtd     l1=l1+t(ct)%*%t(xtd)   }    l1[i]=l1 } 

i hate code. slow, , perhaps more importantly, ugly. feel ripe sort of optimization using reduce or do.call or something. cannot figure out how write functional reduce function. honest, not understand reduce. have been trying read through this, struggling apply here.

the problem seems have several value must 'accumulate' through pulls of random numbers, 'x' reduce list instead of vector, doesn't seem make sense. tried writing this:

runstuff <- function(theprev, thenext, thelam, thect, thesig){   xtd=rmvnorm(1, theprev[1], theprev[2])   thenext[3]=theprev[3]+t(thect)%*%t(xtd)   thenext[1]=thelam*theprev[1]+(1-thelam)*xtd   thenext[2]=thelam*thesig+(1-lambda)*t(xtd)%*%xtd   thenext } 

and using (which expected not work), , not work:

reduce(x=list(mu1, sig1, l1), f=runstuff,         thelam = lambda, thect = ct, thesig = sig0, accumulate=true) 

any how use functional in scenario?