Segmentation Fault with Perf

[Brannon Imamura]

When running Perf, the program errors and closes out. The following is the code.

library(mixOmics)

print("reading")
x <- read.csv("GeneX.csv")
y <- read.csv("GeneY.csv")
print("Loaded")

# everything past this point will not work unless it is performed on a computer with more RAM

# PLS run with 6 components
tox6.pls <- pls(x,
y,
ncomp = 6,
mode = c("regression"),
scale = TRUE,
)

print("Running XValidation")
# Shows how much explanation each component adds
tune.pls <- perf(tox6.pls, validation = "Mfold", folds = 3, progressBar = FALSE, criterion = "all")

I will attach a Valgrind Dump

[Florian Rohart]

Dear Brannon, Thanks for your message.

What is your environment? Windows, Linux, Mac? Are you using Rstudio, Rgui? Are you running that on your machine, on a cluster? What is the size of your data?

PS: it seems your segfault.txt was run on tox10.pls whereas your R code is tox6.pls, is there just a different number of components?

[Brannon Imamura]

Sorry for the inconsistencies in my Valgrind vs the code I put in.

I changed the name to tox6.pls from tox10.pls to better reflect that I was using 6 components instead of 10, but the rest of the code is identical.

This was run on a Linux server, not a cluster. Data has 63 rows and 49,927 columns for a total size of 26,589KB.

R platform:

R version 3.4.1 (2017-06-30) -- "Single Candle" Copyright (C) 2017 The R Foundation for Statistical Computing Platform: x86_64-pc-linux-gnu (64-bit)

[Brannon Imamura]

This appears to work fine on a windows machine.

We are going to run a memory check on the Linux machine to see if a bad stick of ram is the issue.