I'm trying to use a contributed package (rmetasim) to generate simulated
genetic datasets. My scripts work fine when I run them on a Sun
workstation running Solaris 7 and when I run them on a ~4 year old
laptop PC that I have. However, when I run them on my new laptop (Dell
Latitude D410 purchased in August 2005), the simulation will run for a
short (variable) period of time, then R crashes. When it crashes, I
either get an Application Popup that says
'The instruction at "0x5ad71531" referenced memory at
"0x00000014". The
memory could not be "read".',
or I simply get the popup saying that RGUI has experienced a problem,
would I like to send a report to Microsoft. When the latter happens, an
event log is generated that says
'Fauling application rgui.exe, version 2.30.37590.0, fauling module
ntdll.dll, version 5.1.2600.2180, fault address 0x0001888f.'
In both types of errors, the memory addresses listed are different every
time a crash occurs.
I experience this problem primarily when attempting to simulate large
datasets. My script saves the 'landscape' (the object being simulated)
every 100 time steps. When R crashes, I can start it back up, reload
the last landscape saved before the crash, and continue simulating from
there. Thus, there appears to be nothing wrong with the landscapes I'm
simulating.
I've been in contact with the author of the contributed package I'm
using, and he has been unable to find any bugs in the package or
reproduce the error. The bug seems to be platform specific since, as I
mentioned, I'm able to run my scripts crash-free on other computers.
However, I've tried running my scripts on another computer that is
nearly identical to mine (a Dell D610 purchased in August 2005), and
experience the same problems on that computer. I've tried running my
scripts from a command window using 'R --vanilla', and experience the
same problems. I've tried using the development version of R
(R-2.3.0alpha) and experience the same problems.
In reading other bug reports I found that some people have had crashes
caused by buggy video drivers. I've updated my video drivers, but it
didn't help. However, I was only able to upgrade to the newest version
of the video driver; I haven't been able to find an alternative driver
to work with my display adapter. The display adapter used by my
computer is the mobile Intel 915GM/GMS,910GML Express Chipset.
I'm running R 2.2.1 (though I've tried R 2.0.1 and 2.3.0alpha and had
the same problem). The machine on which I've had problems is a Dell
Latitude D410 laptop running Windows XP Pro SP2, 2GHz Pentium processor
with 1GB of RAM. The output of the R command 'version' is;
platform i386-pc-mingw32
arch i386
os mingw32
system i386, mingw32
status
major 2
minor 2.1
year 2005
month 12
day 20
svn rev 36812
language R
The exact sequence of commands I type, after launching the Rgui, is as
follows:
>library(rmetasim)
>source("ArchI_burnin_NOdensdep-mod.R")
The script I'm running (which obviously requires that the package
'rmetasim' be installed) follows. Beware that the script takes about 15
minutes to run on my old computer (the one on which it doesn't crash.
For what it's worth, that computer is a Dell Latitude C400 running
Windows XP Pro SP2, 866 MHz Pentium III processor, 512 MB of RAM). Any
suggestions would be greatly appreciated.
############################################################################
#ArchI_burnin_NOdensdep-mod.R#
habitats <- 1
carrycap <- 7500
stages <- 5
rland <- NULL
numreps <- 2
numsteps <- 50
stepsize <- 100
runlength <- numsteps*stepsize
rland <- new.landscape.empty()
rland <- new.intparam.land(rland, h = habitats, s = stages, totgen =
runlength)
rland <- new.switchparam.land(rland, mp = 1)
rland <- new.floatparam.land(rland)
#life history matrices at zero population density
SZ <- matrix (c(0.730, 0, 0, 0, 0,
0.210, 0, 0, 0, 0,
0, 0.470, 0, 0.946, 0,
0, 0, 0.946, 0, 0,
0, 0.470, 0, 0, 0.954), nrow=5, byrow = T)
RZ <- matrix (c(0, 0, 0, .94, 0,
0, 0, 0, 0, 0,
0, 0, 0, 0, 0,
0, 0, 0, 0, 0,
0, 0, 0, 0, 0), nrow = 5, byrow = T)
M <- matrix (c(0, 0, 0, 0, 0,
0, 0, 0, 0, 0,
0, 0, 0, 0, 0,
0, 0, 0, 0, 1,
0, 0, 0, 0, 0), nrow = 5, byrow = T)
rland <- new.local.demo(rland,SZ,RZ,M)
rland <-
new.epoch.island(rland,0,c(0,0,0,0,0),c(0,0,0,0,0),0,c(0,0,0,0,0),c(0,0,0,0,0),
0,c(0,0,0,0,0),c(0,0,0,0,0), carry = rep(carrycap,habitats))
rland <- new.locus(rland, type = 1, ploidy = 2, transmission = 0,
numalleles = 1, mutationrate = 0.001)
rland <- new.locus(rland, type = 1, ploidy = 2, transmission = 0,
numalleles = 1, mutationrate = 0.001)
rland <- new.locus(rland, type = 1, ploidy = 2, transmission = 0,
numalleles = 1, mutationrate = 0.001)
rland <- new.locus(rland, type = 1, ploidy = 2, transmission = 0,
numalleles = 1, mutationrate = 0.002)
rland <- new.locus(rland, type = 1, ploidy = 2, transmission = 0,
numalleles = 1, mutationrate = 0.002)
rland <- new.locus(rland, type = 1, ploidy = 2, transmission = 0,
numalleles = 1, mutationrate = 0.002)
rland <- new.locus(rland, type = 1, ploidy = 2, transmission = 0,
numalleles = 1, mutationrate = 0.003)
rland <- new.locus(rland, type = 1, ploidy = 2, transmission = 0,
numalleles = 1, mutationrate = 0.003)
rland <- new.locus(rland, type = 1, ploidy = 2, transmission = 0,
numalleles = 1, mutationrate = 0.003)
rland <- new.locus(rland, type = 2, ploidy = 1, transmission = 1,
numalleles = 1, allelesize = 500, mutationrate = 0.005)
rland <- new.individuals(rland, as.integer(c(2775,525,1050,975,2175)))
rland.start <- rland
for (j in 1:numreps) {
rland <- rland.start
for (i in 1:numsteps) {
rland <- simulate.landscape(rland, stepsize)
save(rland,
file=paste("ArchI_test_",j,"_",i,".rda",sep=""))
}
}
--
***********************************
Dr. Karen Martien
Southwest Fisheries Science Center
8604 La Jolla Shores Drive
La Jolla, CA 92037
USA
Ph: 858-546-7058
Fax: 858-546-7003
