>>>>> "MM" == Martin Maechler <[EMAIL PROTECTED]>
>>>>> on Thu, 11 Dec 2003 18:29:05 +0100 (CET) writes:
>>>>> "Torsten" == Torsten Hothorn <[EMAIL PROTECTED]>
>>>>> on Thu, 11 Dec 2003 18:03:07 +0100 (CET) writes:
Torsten> On Thu, 11 Dec 2003, Jeffrey Chang wrote:
>>> Hello everybody,
>>>
>>> I'm running R 1.8.1 on both Linux and OS X compiled with
>>> gcc 3.2.2 and 3.3, respectively. The following call
>>> seems to freeze the interpreter on both systems: >
>>> chisq.test(matrix(c(233, 580104, 3776, 5786104), 2, 2),
>>> simulate.p.value=TRUE)
>>>
>>> By freeze, I mean, the function call never returns
>>> (running > 10 hours so far), the process is unresponsive
>>> to SIGINT (but I call kill it with TERM), and the
>>> process still consumes cycles on the CPU.
>>>
Torsten> This is due to calling `exp' with a very small
Torsten> value leading to a zero return value in rcont2
Torsten> (src/appl/rcont.c) line 70:
MM> yes, that's correct.
However, as I have found out now, the real reason *is* integer
overflow (in multiplication), as I had guessed originally (but
didn't believe anymore for a while after your exp() underflow diagnosis).
Note that this also affects r2dtable(), and e.g., the following
gives a segmentation fault:
M <- 50000 ; r2dtable(1, c(2*M,3), M+1:2)
Further, even if I protect against the exp() underflow, I can
still see examples which "freeze", i.e. end in an infinite loop.
-------
The fix is to ensure that integer multiplications i * j do not happen
but rather i * (double) j
I will commit a fix to R-patched and R-devel later this evening.
--
Martin Maechler <[EMAIL PROTECTED]> http://stat.ethz.ch/~maechler/
Seminar fuer Statistik, ETH-Zentrum LEO C16 Leonhardstr. 27
ETH (Federal Inst. Technology) 8092 Zurich SWITZERLAND
phone: x-41-1-632-3408 fax: ...-1228 <><
______________________________________________
[EMAIL PROTECTED] mailing list
https://www.stat.math.ethz.ch/mailman/listinfo/r-devel
