On Dec 3, 2009, at 3:02 PM, Stephen Grubb wrote:
Hello,
We are occasionally getting matrix results that appear to be
corrupted... here are the last several rows of an example, copy-
pasted out of the R command window. These are supposed to be
floating point numbers.
[25015,] 1.820848e-01-3.2090e-06i
[25016,] 2.178046e-01-4.8140e-06i
[25017,] 1.820848e-01-3.2090e-06i
[25018,] 1.820848e-01-3.2090e-06i
[25019,] 1.144594e-01-1.6657e-06i
[25020,] 1.820848e-01-3.2090e-06i
[25021,] -1.293271e-01+4.3889e-06i
[25022,] 1.144594e-01-1.6657e-06i
[25023,] 1.820848e-01-3.2090e-06i
[25024,] 1.820848e-01-3.2090e-06i
[25025,] 1.173487e-01-4.4415e-07i
[25026,] 1.820848e-01-3.2090e-06i
[25027,] 1.375304e-01-3.6167e-06i
[25028,] 1.820848e-01-3.2090e-06i
[25029,] -1.293271e-01+4.3889e-06i
[25030,] 1.820848e-01-3.2090e-06i
[25031,] 1.820848e-01-3.2090e-06i
[25032,] 1.820848e-01-3.2090e-06i
[25033,] 1.820848e-01-3.2090e-06i
Any general idea what may be going on here?
Those are all complex numbers.
It is a sporadic problem... it occurs maybe 2% or 3% of the time
when running this particular script on various data.
You are most likely "solving" or inverting <something> that does not
have all real roots or eigenvalues ... or something.
I apologize for not including a pared-down example that reproduces
the problem.... we are using an R script written elsewhere on large
data sets. If someone wants more specifics please follow up.
Steve Grubb
David Winsemius, MD
Heritage Laboratories
West Hartford, CT
______________________________________________
R-help@r-project.org mailing list
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.