[NMF-user] nmf error

[Fiona Christie]

Renaud,

I've been trying to run nmf on a data set with approx. 11000 rows and
10 columns. It will run fine when only one value for r is passed or
else a vector is passed to r and nrun is set to 1. However when I try
it with multiple runs an error occurs:


 *** caught segfault ***
address 0x110, cause 'memory not mapped'

 *** caught segfault ***
address 0x110, cause 'memory not mapped'

 *** caught segfault ***
address 0x110, cause 'memory not mapped'

Traceback:
 1: .local(object, ...)
 2: fitted(x)
 3: fitted(x)
 4: .rss(y, fitted(x))
 5: obj.fun(x, y, ...)
 6: .local(object, ...)
 7: deviance(strategy, nmfFit, v, ...)
 8: deviance(strategy, nmfFit, v, ...)
 9: trackError(nmfData, deviance(strategy, nmfFit, v, ...), niter = i,
    force = TRUE)
10: .local(object, y, x, ...)
11: run(strategyX, y, x, maxIter = maxIter, ...)
12: run(strategyX, y, x, maxIter = maxIter, ...)
13: .local(object, y, x, ...)
14: run(object = <S4 object of class "NMFStrategyIterative">, y =
c(0L, 38L, 44L, 0L, 378L, 0L, 16L, 189L, 89L, 43L, 34L, 14L, 10L, 85L,
209L, 0L, 17L, 356L, 0L, 0L, 0L, 19L, ......


I also tried to run one of the examples from NMF-vignette.pdf (from page 14):

res <- nmf(esGolub, 3, seed = 123456, nrun = 2)

and the same thing happens.

It's running on OS X Yosemite Version 10.10.3, 2.7 GHz Intel Core i5
processor, 12 GB 1333 MHz DDR3.

Any help would be much appreciated.

Regards,

Fiona
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/nmf-user/attachments/20150706/8e4472c6/attachment.html>