[R] Memory problem

Thu Apr 7 08:08:01 CEST 2016

Dear Sir,

Yes I am using the plyr and in the end I am writing the output to the data.frame. Earlier I had the problem of process time and hence I made some changes in the code and now I am fetching all the required inputs needed for valuation purpose using ddply, store the results in a data.frame and once that is over, I am carrying out the calculations.

Here is part of my R code-

library(plyr)
library(reshape)

tx <- read.csv('transaction_fxdeal.csv')
tx$id  <-  as.character(tx$id)

n        <- max(unique(simulated_exchange$id))

result <- NULL
current  <- 1
rcount   <- 0
current1 <- 1
rcount1  <- 0
current2 <- 1
rcount2  <- 0
for (env in 0:n) {

  if (rcount == 0) rcount <- nrow(subset(simulated_interest, id==env))
  temp <- current+rcount-1
  env_rates  <- simulated_interest[current:temp,]
  env_rates  <- env_rates[order(env_rates$curve, env_rates$day_count), ]
  if (rcount1 == 0)rcount1 <- nrow(subset(simulated_exchange, id==env))
  temp <- current1+rcount1-1
  exch_rates <- simulated_exchange[current1:temp,]
  if (rcount2 == 0)rcount2 <- nrow(subset(simulated_instruments, id==env))
  temp <- current2+rcount2-1
  instr_rates<- simulated_instruments[current2:temp,]
  current <- current+rcount
  current1 <- current1+rcount1
  current2 <- current2+rcount2

  curve       <- daply(env_rates, 'curve', function(x) {
    return(approxfun(x$day_count, x$rate, rule = 2))
  })

result <- rbind(result, ddply(tx, 'id', function(x) {

intrate_from <- curve[[x$currency_from]](x$maturity_from)
intrate_to   <- curve[[x$currency_to]](x$maturity_to)
cross_rate   <- subset(exch_rates, key==paste(x$currency_from_exch, x$currency_to_exch, sep='_'))$rate
base_rate    <- subset(exch_rates, key==paste(x$currency_to_exch, x$currency_base, sep='_'))$rate

return(data.frame(env=env, intrate_from=intrate_from, intrate_to=intrate_to, cross_rate=cross_rate, base_rate=base_rate))

  }))
}

sorted <- result[order(result$id, result$env),]

sorted$currency_from_exch <- rep(tx$currency_from_exch, each = length(unique(sorted$env)))
sorted$currency_to_exch <- rep(tx$currency_to_exch, each = length(unique(sorted$env)))
sorted$currency_base <- rep(tx$currency_base, each = length(unique(sorted$env)))
sorted$transaction_type <- rep(tx$transaction_type, each = length(unique(sorted$env)))
sorted$amount_fromccy <- rep(tx$amount_fromccy, each = length(unique(sorted$env)))
sorted$amount_toccy <- rep(tx$amount_toccy, each = length(unique(sorted$env)))
sorted$intbasis_fromccy <- rep(tx$intbasis_fromccy, each = length(unique(sorted$env)))
sorted$intbasis_toccy <- rep(tx$intbasis_toccy, each = length(unique(sorted$env)))
sorted$maturity_from <- rep(tx$maturity_from, each = length(unique(sorted$env)))
sorted$maturity_to <- rep(tx$maturity_to, each = length(unique(sorted$env)))
sorted$currency_from <- rep(tx$currency_from, each = length(unique(sorted$env))) 
sorted$currency_to <- rep(tx$currency_to, each = length(unique(sorted$env))) 

sorted$from_mtm <- sorted$cross_rate * (sorted$amount_fromccy / ((1 + (sorted$intrate_from/100))^(sorted$maturity_from / sorted$intbasis_fromccy)))

sorted$to_mtm       <- (sorted$amount_toccy   / ((1 + (sorted$intrate_to/100))^(sorted$maturity_to / sorted$intbasis_toccy)))

mtm_base <- function(from_mtm, to_mtm, base_rate)
{
mtm <- (from_mtm + to_mtm)
mtm_bc = mtm*base_rate[1]

return(data.frame(mtm_bc = mtm_bc))
}

sorted1 <- ddply(.data=sorted, .variables = "id", .fun=function(x) mtm_base(from_mtm = x$from_mtm, to_mtm = x$to_mtm, base_rate = x$base_rate))

sorted$mtm <- sorted1$mtm
sorted$mtm_bc <- sorted1$mtm_bc

sorted2 <- ddply(sorted, .(currency_from_exch, id), mutate, change_in_mtm_bc = mtm_bc - mtm_bc[1])

sorted$change_in_mtm_bc <- sorted2$change_in_mtm_bc

sorted <- sorted[order(sorted$id, sorted$env),]

write.csv(data.frame(sorted), file='MC_result_fxdeal.csv', row.names=FALSE)

# ________________________________________________________

# END of Code

With regards

Amelia

On Wednesday, 6 April 2016 7:43 PM, Jeff Newmiller <jdnewmil at dcn.davis.ca.us> wrote:

As Jim has indicated, memory usage problems can require very specific diagnostics and code changes,  so generic help is tough to give. 

However, in most cases I have found the dplyr package to be more memory efficient than plyr, so you could consider that. Also, you can be explicit about only saving the minimum results you want to keep rather than making a list of complete results and extracting results later. 
-- 
Sent from my phone. Please excuse my brevity.

On April 6, 2016 4:39:59 AM PDT, Amelia Marsh via R-help <r-help at r-project.org> wrote:
Dear R Forum,
>
>I have about 2000+ FX forward transactions and I am trying to run 1000 simulations. If I use less no of simulations, I am able to get the desired results. However, when I try to use more than 1000 simulations, I get following error.
>
>
>sorted2 <- ddply(sorted, .(currency_from_exch, id), mutate, change_in_mtm_bc = mtm_bc - mtm_bc[1]) 
>>
>Error: cannot allocate vector of size 15.6 Mb 
>
>
>In addition: Warning messages: 
>1: Reached total allocation of 3583Mb: see help(memory.size) 
>2: Reached total allocation of 3583Mb: see help(memory.size) 
>3: In output[[var]][rng] <- df[[var]] : 
>Reached total allocation of 3583Mb: see help(memory.size) 
>4: In output[[var]][rng] <- df[[var]] : 
>Reached total allocation of 3583Mb: see help(memory.size) 
>5: In output[[var]][rng] <- df[[var]] : 
>Reached total allocation of 3583Mb: see help(memory.size) 
>6: In output[[var]][rng] <- df[[var]] : 
>Reached total allocation of 3583Mb: see help(memory.size) 
>7: In output[[var]][rng] <- df[[var]] : 
>Reached total allocation of 3583Mb: see help(memory.size) 
>8: In output[[var]][rng] <- df[[var]] : 
>Reached total allocation of 3583Mb: see help(memory.size)
>
>
>When I checked -
>
>
>memory.size() 
>>[1] 846.83 
>
>memory.limit() 
>>[1] 3583
>
>
>The code is bit lengthy and unfortunately can't be shared.
>
>Kindly guide how this memory probelm can be tackled? I am using R x64 3.2.0
>
>Regards
>
>Amelia
>
>>________________________________
>
>R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
>https://stat.ethz.ch/mailman/listinfo/r-help
>PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
>and provide commented, minimal, self-contained, reproducible code.
>