Integer overflow about somoclu

[peterwittek]

I cannot reproduce the error on 64-bit Linux with 32GB of RAM. R is insanely inefficient with memory use: during the conversion, it peaks around 19GB. Once the conversion is over, I run garbage collection and the final memory use is around 1.18GB. So I suspect your problem has to do with your configuration, and not with Somoclu/kohonen. This is what I do:

library('Rsomoclu')
library('kohonen')
input_data<-matrix(runif(1), nrow=1400000, ncol=50)
nSomX <- 20
nSomY <- 20
nEpoch <- 10
radius0 <- 0
radiusN <- 0
radiusCooling <- "linear"
scale0 <- 0
scaleN <- 0.01
scaleCooling <- "linear"
kernelType <- 0
mapType <- "planar"
gridType <- "rectangular"
compactSupport <- FALSE
codebook <- NULL
neighborhood <- "gaussian"
res <- Rsomoclu.train(input_data, nEpoch, nSomX, nSomY,
                      radius0, radiusN,
                      radiusCooling, scale0, scaleN,
                      scaleCooling,
kernelType, mapType, gridType, compactSupport, neighborhood, codebook)
gc()
sommap = Rsomoclu.kohonen(input_data, res)
gc()

[apastore]

Thanks I will try to test more in detail. I am using a centOS bases qsub cluster I request 50GB Ram. So I was not expecting to be the problem.

Thanks!

Sent from my iPhone

On Oct 2, 2016, at 5:53 AM, Peter Wittek [email protected] wrote:

I cannot reproduce the error on 64-bit Linux with 32GB of RAM. R is insanely inefficient with memory use: during the conversion, it peaks around 19GB. Once the conversion is over, I run garbage collection and the final memory use is around 1.18GB. So I suspect your problem has to do with your configuration, and not with Somoclu/kohonen. This is what I do:

library('Rsomoclu')
library('kohonen')
input_data<-matrix(runif(1), nrow=1400000, ncol=50)
nSomX <- 20
nSomY <- 20
nEpoch <- 10
radius0 <- 0
radiusN <- 0
radiusCooling <- "linear"
scale0 <- 0
scaleN <- 0.01
scaleCooling <- "linear"
kernelType <- 0
mapType <- "planar"
gridType <- "rectangular"
compactSupport <- FALSE
codebook <- NULL
neighborhood <- "gaussian"
res <- Rsomoclu.train(input_data, nEpoch, nSomX, nSomY,
radius0, radiusN,
radiusCooling, scale0, scaleN,
scaleCooling,
kernelType, mapType, gridType, compactSupport, neighborhood, codebook)
gc()
sommap = Rsomoclu.kohonen(input_data, res)
gc()
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub, or mute the thread.